ICE-B 2008

Proceedings of the
International Conference on
e-Business

Porto, Portugal

July 26 – 29, 2008

Organized by
INSTICC – Institute for Systems and Technologies of Information, Control
and Communication

Co-Sponsored by
WfMC – Workflow Management Coalition – Process Thought Leadership

Technically Co-Sponsored by
IEEE SMC – Systems, Man, and Cybernetics Society

I
 Copyright © 2008 INSTICC – Institute for Systems and Technologies of
Information, Control and Communication
All rights reserved

Edited by Joaquim Filipe, David A. Marca, Boris Shishkov and Marten van Sinderen

Printed in Portugal
ISBN: 978-989-8111-58-6
Depósito Legal: 279019/08

http://www.ice-b.org
[email protected]

II
 BRIEF CONTENTS

INVITED SPEAKERS........................................................................................................................IV

TUTORIAL .....................................................................................................................................IV

ORGANIZING AND STEERING COMMITTEES .................................................................................... V

PROGRAM COMMITTEE .................................................................................................................VI

AUXILIARY REVIEWERS ............................................................................................................. VIII

SELECTED PAPERS BOOK ........................................................................................................... VIII

FOREWORD....................................................................................................................................IX

CONTENTS .....................................................................................................................................XI

III
 INVITED SPEAKERS

David A. Marca
University of Phoenix
U.S.A.

Yaakov Kogan
AT&T Labs
U.S.A.

Hsiao-Hwa Chen
National Sun Yat-Sen University
Taiwan

Nuno Borges Carvalho

Instituto de Telecomunicações / Universidade de Aveiro
Portugal

Ueli Maurer
ISwiss Federal Institute of Technology (ETH)
Switzerland

Bart Preneel
University of Leuven
Belgium

Ingemar Cox
University College London
U.K.

TUTORIAL
PATENTABILITY OF E-BUSINESS AND COMPUTER-IMPLEMENTED INVENTIONS AT THE EPO
Falk Giemsa, European Patent Office, Germany

IV
ORGANIZING AND STEERING COMMITTEES
CONFERENCE CO-CHAIRS
Joaquim Filipe, Polytechnic Institute of Setúbal / INSTICC, Portugal
Mohammad S. Obaidat, Monmouth University, U.S.A.

PROGRAM CO-CHAIRS
Marten van Sinderen, University of Twente, The Netherlands
Boris Shishkov, University of Twente, The Netherlands
David A. Marca, University of Phoenix, U.S.A.

PROCEEDINGS PRODUCTION
Helder Coelhas, INSTICC, Portugal
Vera Coelho, INSTICC, Portugal
Andreia Costa, INSTICC, Portugal

Bruno Encarnação, INSTICC, Portugal

Bárbara Lima, INSTICC, Portugal
Vitor Pedrosa, INSTICC, Portugal

Vera Rosário, INSTICC, Portugal

CD-ROM PRODUCTION
Paulo Brito, INSTICC, Portugal

GRAPHICS PRODUCTION
Helder Coelhas, INSTICC, Portugal

SECRETARIAT, WEBDESIGNER AND WEBMASTER

Mónica Saramago, INSTICC, Portugal

V
 PROGRAM COMMITTEE
Geetha Abeysinghe, Middlesex University, U.K. Jen-Yao Chung, IBM T. J. Watson Research Center,
U. S.A.
Ajith Abraham, Norwegian University of Science
and Technology, Norway Oscar Corcho, Universidad Politécnica de Madrid,
Spain
Fahim Akhter, Zayed University, U.A.E.
Alfredo Cuzzocrea, University of Calabria, Italy
Damminda Alahakoon, Monash University, Australia
George Dafoulas, Middlesex University, U.K.
Antonia Albani, Delft University of Technology,
The Netherlands Hepu Deng, RMIT University, Australia
Sarmad Alshawi, Brunel University, U.K. Claudia Diaz, K. U. Leuven, Belgium
Panagiotes Anastasiades, University of Crete, Greece Asuman Dogac, Middle East Technical University,
Turkey
José Luis Arjona, University of Huelva, Spain
Schahram Dustdar, T. U. Wien, Austria
Anteneh Ayanso, Brock University, Canada
Joerg Evermann, Memorial University of Canada,
Gilbert Babin, HEC Montréal, Canada
Canada
Ladjel Belllatreche, Poitiers University, France Jinan Fiaidhi, Lakehead University, Canada
Morad Benyoucef, University of Ottawa, Canada Xiang Fu, Georgia Southwestern State University,
Hans Bjornsson, Chalmers University of Technology, U.S.A.
Sweden George Giaglis, Athens University of Economics and
Peter Bodorik, Dalhousie University, Canada Business, Greece
Indranil Bose, University of Hong Kong, Hong Kong Paul Grefen, Eindhoven University of Technology,
The Netherlands
Vesna Bosilj-Vuksic, University of Zagreb, Faculty of
Economics and Business, Croatia Volker Gruhn, University of Leipzig, Germany
Christos Bouras, University of Patras, Greece Haresh Gurnani, University of Miami, U.S.A.
Stephane Bressan, National University of Singapore, Mohand-Said Hacid, University Lyon 1, France
Singapore Milena Head, McMaster University, Canada
Rongzeng Cao, IBM China Research Lab, China Vlatka Hlupic, University of Westminster, U.K.
Barbara Carminati, Univeristy of Insubria, Italy Birgit Hofreiter, University of Technology Sydney,
Teuta Cata, Northern Kentucky University, U.S.A. Australia
Michelangelo Ceci, University of Bari, Italy Andreas Holzinger, IMI, Research Unit HCI,
Medical University Graz, Austria
Wojciech Cellary, Poznan University of Economics,
Poland Christian Huemer, Vienna University of Technology,
Austria
Patrick Y.K. Chau, The University of Hong Kong,
Hong Kong Patrick C. K. Hung, University of Ontario Institute of
Technology, Canada
Michael Chau, The University of Hong Kong,
Hong Kong Takayuki Ito, Nagoya Institute of Technology, Japan
Harry Chen, Image Matters LLC/University of Arun Iyengar, IBM T.J. Watson Research Center,
Maryland, Baltimore County, U.S.A. U.S.A.
Dickson Chiu, Dickson Computer Systems, Nallani Iyengar, Vellore Institute of Technology
Hong Kong University, India
Soon Ae Chun, City University of New York, U.S.A. James Joshi, University of Pittsburgh, U.S.A.
Matjaz B. Juric, University of Maribor, Slovenia

VI
 PROGRAM COMMITTEE (CONT.)
Sherif Kamel, The American University in Cairo, Hossein Sharifi, Liverpool University Management
Egypt School, U.K.
Kay Hooi Keoy, Bradford University School of Quan Z. Sheng, The University of Adelaide, Australia
Management, U.K.
Mario Spremic, University of Zagreb, Croatia
David Kreps, University of Salford, U.K.
Katarina Stanoevska-Slabeva, University of St.
Anton Lavrin, The Technical University of Kosice, Gallen, Switzerland
Slovakia
York Sure, SAP Research, CEC Karlsruhe, Germany
Lundy Lewis, Southern New Hampshire University,
Paula Swatman, University of South Australia,
U.S.A.
Australia
Dahui Li, University of Minnesota Duluth, U.S.A.
Ramayah T., Universiti Sains Malaysia, Malaysia
Yinsheng Li, Fudan University, China
Thompson Teo, National University of Singapore,
Chin Lin, South China University of Technology, Singapore
China
Thanassis Tiropanis, University of Southampton,
Tokuro Matsuo, Yamgataga University, Japan U.K.
Jan Mendling, QUT Brisbane, Australia David Trastour, HP Labs, U.K.
Brian Mennecke, Iowa State University, U.S.A. Roland Traunmüller, University Linz, Austria
Adrian Mocan, Digital Enterprise Research Institute, Jan Vanthienen, Katholieke Universiteit Leuven,
Austria Belgium
Sabah Mohammed, Lakehead University, Canada Tomas Vitvar, National University of Ireland, Ireland
Ali Reza Montazemi, McMaster University, Canada Adam Vrechopoulos, Athens University of
Economics and Business, Greece
Wee-Keong Ng, Nanyang Technological University,
Singapore Yan Wang, Macquarie University, Australia
Dan O'Leary, University of Southern California, Krzysztof Wecel, Poznan University of Economics,
U.S.A. Poland
Georgios Papamichail, National Centre of Public Michael Weiss, Carleton University, Canada
Administration and Local Government, Greece
Erik Wilde, UC Berkeley, U.S.A.
Cesare Pautasso, University of Lugano, Switzerland
Jongwook Woo, California State University Los
Krassie Petrova, Auckland University of Technology, Angeles, U.S.A.
New Zealand
Lai Xu, SAP Research, Switzerland
Pascal Poncelet, LGI2P/EMA, France
Benjamin Yen, The University of Hong Kong,
Pak-Lok Poon, The Hong Kong Polytechnic Hong Kong
University, Hong Kong
Soe-Tsyr Yuan, National Chengchi University,
Philippos Pouyioutas, University of Nicosia, Cyprus Taiwan
Dimitris Rigas, University of Bradford, U.K. Guangquan Zhang, University of Technology
Sydney, Australia
David Ruiz, University of Seville, Spain
Jarogniew Rykowski, Poznan University of
Economics, Poland
Demetrios Sampson, University of Piraeus, Greece

VII
 AUXILIARY REVIEWERS
Hameed Al-Qaheri, Kuwait University, Kuwait Boudewijn van Dongen, Eindhoven University of
Technology, The Netherlands
María del Carmen Suárez de Figueroa Baonza,
Universidad Politécnica de Madrid, Spain José María García, University of Seville, Spain
Raúl García Castro, Universidad Politécnica de Mehmet Olduz, Middle East Technical University,
Madrid, Spain Turkey
Remco Dijkman, Eindhoven University of Jochem Vonk, Eindhoven University of Technology,
Technology, The Netherlands The Netherlands

SELECTED PAPERS BOOK

A number of selected papers presented at ICETE 2008 will be published by Springer-Verlag in a CCIS Series book.
This selection will be done by the Conference Co-chairs and Program Co-chairs, among the papers actually
presented at the conference, based on a rigorous review by the ICETE 2008 program committee members.

VIII
 FOREWORD
We warmly welcome you to ICE-B 2008 - the International Conference on E-Business, which is held, this
year, in Porto, Portugal. This conference reflects a growing effort to increase the dissemination of
recent research results among professionals who work on the e-business field. ICE-B is integrated
as one of the modules of the ICETE conference, which occurs concurrently with ICE-B.
The major goal of ICETE is to bring together researchers, engineers and practitioners interested in
information and communication technologies, including e-business, wireless networks and
information systems, security and cryptography, signal processing and multimedia applications.
These are the main knowledge areas that define the four component conferences, namely: ICE-B,
SECRYPT, SIGMAP and WINSYS, which together form the ICETE joint conference.
In the program for ICETE, we have included keynote lectures, tutorials, papers, and posters to
present the widest possible view on these technical areas. With its four tracks, we expect to appeal
to a global audience of the engineers, scientists, business practitioners and policy experts, interested
in the research topics of ICETE. All tracks focus on research related to real world applications and
rely on contributions not only from Academia, but also from industry, with different solutions for
end-user applications and enabling technologies, in a diversity of communication environments.
The four volume set of proceedings demonstrate a number of new and innovative solutions for e-
business and telecommunication, and demonstrate the vitality of these research areas.
ICETE has received 440 papers in total, with contributions from more than 40 different countries,
from all continents, which demonstrates the success and global dimension of ICETE 2008. To
evaluate each submission, a double blind paper evaluation method was used: each paper was
reviewed by at least two experts from the International Program Committee, in a double-blind
review process, and most papers had 3 reviews or more. In the end, 174 papers were selected for
oral presentation and publication, corresponding to a 39% acceptance ratio. Of these only 77 were
accepted as full papers (17% of submissions) and 97 as short papers. Additionally, 87 papers were
accepted for poster presentation. These acceptance ratios demonstrate that ICETE 2008 strives to
achieve a high quality standard which we will keep and enhance in order to ensure the success of
next year conference, to be held in Milan/Italy. A short list of about thirty papers will be also
selected to appear in a book that will be published by Springer.
We would like to emphasize that ICETE 2008 includes several outstanding keynote lectures, which
are relevant to today’s lines of research and technical innovation. These talks are presented by
distinguished researchers who are internationally renowned experts in all ICETE areas, and their
contributions heighten the overall quality of our Conference.
A successful conference involves more than paper presentations; it is also a meeting place, where
ideas about new research projects and other ventures are discussed and debated. Therefore, a social
event including a conference diner/banquet has been planned for the evening of July 28 in order to
promote this kind of social networking.

IX
We would like to express our thanks, first of all, to all authors including those whose papers were
not included in the program. Next, we would like to thank all the members of the program
committee and reviewers, who helped us with their expertise, dedication and time. We would also
like to thank the invited speakers for their invaluable contribution, in sharing their vision and
knowledge. Lastly, but certainly not least, we give our deep appreciation to the secretariat and to all
the other members of the organizing committee, whose diligence in dealing with all organizational
issues were essential to a collaborative effort of a dedicated and highly capable team.
We hope that you will find these proceedings interesting and to be a helpful reference in the future
for all those who need to address the areas of e-business and telecommunications.
Enjoy the program and your stay in Porto.

Marten van Sinderen

University of Twente, The Netherlands

Boris Shishkov
University of Twente, The Netherlands

David A. Marca
University of Phoenix, U.S.A.

X
 CONTENTS
INVITED SPEAKERS

KEYNOTE LECTURES

E-BUSINESS INNOVATION - Surviving the Coming Decades IS-5

David A. Marca

IMPROVING RELIABILITY IN COMMERCIAL IP NETWORKS IS-17

Yaakov Kogan

CRYPTOGRAPHIC ALGORITHMS - Successes, Failures and Challenges IS-21

Bart Preneel

WATERMARKING, STEGANOGRAPHY AND CONTENT FORENSICS IS-29

Ingemar J. Cox

RETHINKING DIGITAL SIGNATURES IS-31

Ueli Maurer

THE IMPORTANCE OF METROLOGY IN WIRELESS COMMUNICATION SYSTEMS - From

AM/FM to SDR Systems IS-35
Nuno Borges Carvalho

NEXT GENERATION CDMA TECHNOLOGIES FOR FUTURISTIC WIRELESS

COMMUNICATIONS IS-37
Hsiao-Hwa Chen

TUTORIAL

PATENTABILITY OF E-BUSINESS AND COMPUTER-IMPLEMENTED INVENTIONS AT THE

EPO IS-41
Falk Giemsa

COMMUNICATION AND SOFTWARE TECHNOLOGIES AND ARCHITECTURES

FULL PAPERS

A PERFORMANCE EVALUATION OF AN ULTRA-THIN CLIENT SYSTEM

Colin Pattinson and Tahir Siddiqui 5

USING CO-OCCURRENCE TO CLASSIFY UNSTRUCTURED DATA IN TELECOMMUNICATION

SERVICES
Motoi Iwashita, Ken Nishimatsu and Shinsuke Shimogawa 12

SHORT PAPERS

FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

Toshio Kodama, Tosiyasu L. Kunii and Yoichi Seki 21

XI
ANALYZING DECENTRALIZED GOVERNABILITY OF BUSINESS PROCESSES BY EXTENDED
PETRI NETS AND MODAL LOGICS
Takashi Hattori, Hiroshi Kawakami, Osamu Katai and Takayuki Shiose 29

AN ARCHITECTURE FOR DYNAMIC INVARIANT GENERATION IN WS-BPEL WEB SERVICE

COMPOSITIONS
M. Palomo Duarte, A. García Domínguez and I. Medina Bulo 37

A NEW REINFORCEMENT SCHEME FOR STOCHASTIC LEARNING AUTOMATA - Application

to Automatic Control
Florin Stoica, Emil M. Popa and Iulian Pah 45

BUSINESS INTELLIGENCE THROUGH REAL-TIME TRACKING - Using a Location System

Towards Behaviour Pattern Extraction
Pedro Abreu, Vasco Vinhas and Pedro Mendes 51

A FRAMEWORK FOR DYNAMIC KNOWLEDGE ACQUISITION

Ana Aguilera and Alberto Subero 58

PRES – PERSONALIZED EVALUATION SYSTEM IN A WEB COMMUNITY - A Conceptual Model

Designed to Evaluate Reputation in Order to Achive a Personalised View on the System for Each User
Lenuta Alboaie 64

POSTERS

A STUDY OF THE EFFECTIVENESS OF “WAKE UP ON LAN” AS A MEANS OF POWER

MANAGEMENT
Colin Pattinson and Linton Robinson 73

ALGORITHM AND AN ELEVATOR CONTROL SYSTEM EXAMPLE FOR CTL MODEL UPDATE
Cacovean Laura Florentina, Pah Iulian, Popa Emil Marin and Brumar Cristina Ioana 77

COMMON TEXTILE VOCABULARIES AND DOCUMENTS - A Conceptual Foundation of a Globally

Interoperable Textile e-Marketplace
Jingzhi Guo and Zhuo Hu 81

PROPOSAL OF AN ARCHITECTURE FOR DIGITAL CITIES CREATION - Proposal of an

Architecture P2P for Digital Cities Creation
André M. Panhan, Denys G. Santos and Leonardo S. Mendes 89

OPEN SOURCE SOFTWARE AND LEVERAGING OF BUSINESS EFFECTIVENESS IN SMES - A

Case Study
Steven Butler, Dotun Adebanjo and Hossam Ismail 93

AN ONTOLOGY-BASED ARCHITECTURE FOR MULTI-AGENT SYSTEM ENVIRONMENT

Roberto Paiano, Anna Lisa Guido and Enrico Pulimeno 101

SECURE IT/TELCO ENVIRONMENT PLANNING MADE EASY - A Concept of a Tool for Planning
Secure IT/Telco Infrastructure and Applications
Wolfgang Haidegger 107

WEB AND MOBILE BUSINESS SYSTEMS AND SERVICES

FULL PAPERS

MOBILE TOURISM SERVICES - Experiences from Three Services on Trial

Niklas Eriksson and Peter Strandvik 115

XII
ADOPTION VERSUS USE DIFFUSION - Predicting User Acceptance of Mobile TV in Flanders
Tom Evens, Lieven De Marez and Dimitri Schuurman 124

ADOPTION OF MOBILE SERVICES IN FINLAND - Conceptual Model and Application-based Case

Study
Mari Ervasti and Heli Helaakoski 131

SHORT PAPERS

THE ROLE OF AUDIO-VISUAL METAPHORS IN AIDING THE COMMUNICATION

OF CUSTOMER KNOWLEDGE - User Satisfaction Prespective
Dimitrios I. Rigas and Mutlaq B. Alotaibi 143

A PERVASIVE NUTRITIONAL MONITORING AND ADVISE SYSTEM - NutriMe

Vitor Basto Fernandes, João Varajão and António Cunha 149

REPLICATION OF WEB SERVICES FOR QOS GUARANTEES IN WEB SERVICE COMPOSITION

Dirk Thissen and Thomas Brambring 155

AN E-VORTAL FOR THE PORTUGUESE BAKING INDUSTRY - Requirements Model

João Varajão, Jorge Gouveia and Paula Oliveira 161

ON EXPLORING CONSUMERS’ TECHNOLOGY FORESIGHT CAPABILITIES - An Analysis of

4,000 Mobile Service Ideas
Petteri Alahuhta, Pekka Abrahamsson and Antti Nummiaho 169

AN ANALYSIS OF CONTEXT-AWARENESS IN COMMERCIAL MOBILE SERVICES

Ana M. Bernardos, Daniel Marcos and José R. Casar 177

AN ACCESS-CONTROL MODEL FOR MOBILE COMPUTING WITH SPATIAL CONSTRAINTS -

Location-aware Role-based Access Control with a Method for Consistency Checks
Michael Decker 185

A PLATFORM FOR INVESTIGATING EFFECTIVENESS FOR STATIC, ADAPTABLE, ADAPTIVE,

AND MIXED-INITIATIVE ENVIRONMENTS IN E-COMMERCE
Khalid Al-Omar and Dimitris Rigas 191

AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT

PRIORITIZATION - Case Study on Fast-food Chain Restaurants
Slobodan Lukovic, Nikola Puzovic and Milos Stanisavljevic 197

MOBILE BUSINESS EXPERT ADVISOR

Danco Davcev, Marjan Arsic and Dalibor Ilievski 203

A MOBILE BUSINESS PROCESS DEPLOYMENT FRAMEWORK FOR DEVICE INDEPENDENCE

AND CONTEXT-AWARE ENVIRONMENTS
Torab Torabi, Saqib Ali and Hassan Ali 209

GIS-BASED MAP GENERATION USING NEW SURVEY TECHNIQUES

Balqies Sadoun and Omar Al-Bayari 217

POSTERS

eCT: THE B2C E-COMMERCE TOOLKIT FOR THE WEBCOMFORT PLATFORM

Frederico de Carvalho Baptista, João de Sousa Saraiva and Alberto Rodrigues da Silva 225

XIII
SURVEY OF CONSUMERS’ DECISION MAKING PROCESS FOR ONLINE MUSIC SERVICE
Ayako Hiramatsu, Takahiro Yamasaki and Kazuo Nose 229

TYPICAL PROBLEMS WITH DEVELOPING MOBILE APPLICATIONS FOR HEALTH CARE -

Some Lessons Learned from Developing User-centered Mobile Applications in a Hospital Environment
Andreas Holzinger, Martin Höller, Marcus Bloice and Berndt Urlesberger 235

A STUDY OF FACTOR AFFECTING CUSTOMER SWITCHING BEHAVIOR OF MOBILE

TELECOMMUNICATION 3.5G SERVICES
Jungwoo Lee, Eok Baek and Sujung Sung 241

ADOPTION OF NEAR FIELD COMMUNICATION TECHNOLOGY IN BUSINESS TO

CONSUMER SERVICES
Arto Wallin, Juha Häikiö and Jaana Määttä 247

TAXONOMY FOR MOBILE TERMINALS - A Selective Classification Scheme

Gunther Schiefer and Michael Decker 255

IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND EVOLUTIVE UI-COTS

COMPONENTS ARCHITECTURES
José Andrés Asensio, Luis Iribarne, Nicolás Padilla and Rosa Ayala 259

BUSINESS AND SOCIAL APPLICATIONS

FULL PAPERS

DEALING WITH BUSINESS PROCESS EVOLUTION USING VERSIONS

Mohamed Amine Chaâbane, Eric Andonoff, Lotfi Bouzguenda and Rafik Bouaziz 267

A THEORY-DRIVEN FRAMEWORK FOR CONSUMERS TO ADOPT M-COMMERCE DEVICES

Vincent Cho, Humphry Hung and Y. H. Wong 279

IT APPLICATIONS IN PRODUCTION PLANNING AND CONTROL - A Survey of Medium Sized

Business in German-speaking Europe
Jakob Lewandowski, Matthias Buhl and Burkhard Kittl 285

CONDITIONS FOR TECHNOLOGY ACCEPTANCE - Broadening the Scope of Determinants of ICT

Appropriation
Pieter Verdegem and Lieven De Marez 292

BARRIERS TO MOBILE BANKING ADOPTION - A Cross-national Study

Tommi Laukkanen and Pedro Cruz 300

BUSINESS AND TECHNICAL WORKFLOWS FOR E-BUSINESS IN A VIRTUAL CLUSTER OF

ISPS
Jane Hall and Klaus-Peter Eckert 307

FACTORS AFFECTING THE USAGE OF T-GOVERNMENT SERVICES - An Exploratory Study

Michele Cornacchia, Filomena Papa, Stefano Livi, Bartolomeo Sapio, Enrico Nicolò and Gaetano Bruno 315

BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL

DATABASE MODEL
Kelly Rosa Braghetto, João Eduardo Ferreira and Calton Pu 323

ON THE USE OF “QUALIFIED” DIGITAL SIGNATURES

Maurizio Adriano Strangio 334

XIV
RESEARCHING SEARCH - A Study into Search Engine Marketing Practices in Ireland
Chris Barry and Debbie Charleton 339

SHORT PAPERS

AN OPTIMIZATION METHOD FOR REDEMPTION AND DUE DATE MATCHING IN

ASSIGNMENT OF ELECTRONIC RECEIVABLES BY USING INTEGER LINEAR
PROGRAMMING
Toshiyuki Moritsu and Norihisa Komoda 349

E-LEARNING FOR NEW GRADUATE EMPLOYEES - Another Function of e-Learning for New
Graduate Employees of Japanese and Korean Companies
Jiro Usugami 357

OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF E-SUPPLY CHAINS AMONG

MALAYSIAN’S SMES
Kay Hooi Keoy, Mohamed Zairi and Khalid Hafeez 362

A COMPARISON OF WEB SITE ADOPTION IN SMALL AND LARGE PORTUGUESE FIRMS

Tiago Oliveira and Maria F. O. Martins 370

PROTOCOL OF AUTHENTICITY TO PROVIDE LEGAL SECURITY IN E-CONTRACTS - A

Prototype
João Fábio de Oliveira, Cinthia O. de A. Freitas and Altair Santin 378

USING TECHNOLOGY ACCEPTANCE MODEL TO EVALUATE USERS’ ATTITUDE AND

INTENTION OF USES
Dauw-Song Zhu and Chih-Te Lin 384

SEMANTIC INTEROPERABILITY FOR E-BUSINESS IN THE ISP SERVICE DOMAIN

Jane Hall and Stefanos Koukoulas 390

ON-DEMAND MOBILE CRM APPLICATIONS FOR SOCIAL MARKETING - Business and

Technology Perspective
Viktor Kaufman, Yuri Natchetoi and Vasily Ponomarev 397

E-COMPLEMENTARITY - The Link to e-Business Value

Pedro Soto-Acosta and Angel L. Meroño-Cerdan 405

A RESEARCH MODEL OF CUSTOMER RELATIONSHIP MANAGEMENT SYSTEMS FOR

MOBILE DEVICES - Description of a Research Model about Customer Relationship Management
Projects
Rebecca Bulander 413

POSTERS

A STUDY OF INNOVATION DIFFUSION OF ELECTRONIC PATIENT RECORDS FOR

SUPPORTING MEDICAL PRACTICE
Vincent Cho and Geoffrey Lieu 421

MANAGEMENT INFORMATION SYSTEMS IN ROMANIAN UNIVERSITIES

Ana-Ramona Bologa, Razvan Bologa, Gheorghe Sabau and Mihaela Muntean 425

A FEATURE EXTRACTING METHOD FOR TAMPER DETECTION IN PRINTED DOCUMENTS

Yoshiyasu Takahashi, Takaaki Yamada and Seiichi Susaki 429

XV
THE AFFECTION OF DEMOGRAPHIC CHARACTERISTICS ON MEDIA CHOICE - A Case about
Chinese Miniature Automobile Consumers
Dao-ping Chen and Wei Liu 433

WEB BASED COLLABORATIVE DOCUMENT CREATION AND REVIEW SYSTEM

Marius Ioan Podean, Raluca Arba and Loredana Muresan 437

ANTECEDENCES AND CONSEQUENCES OF E-SERVICE QUALITY ACROSS INDUSTRY

SECTORS
Dauw-Song Zhu, Chih-Te Lin and Yu-Ling Su 443

RFID PASSWORD MANAGEMENT METHODS FOR FALSIFICATION PREVENTION IN

BOOKSTORE MANAGEMENT USING SECURE RFID TAGS
Yuichi Kobayashi, Yoji Taniguchi, Toshiyuki Kuwana and Masanori Akiyoshi 447

WORK LISTS FOR THE TRANSPORT OF PATIENTS - A Case for Mobile Applications in Health Care
Andreas Holzinger, Jürgen Trauner and Stefan Biffl 454

TECHNOLOGY VS BUSINESS NEEDS IN BUSINESS INTELLIGENCE PROJECTS

Ana-Ramona Bologa, Razvan Bologa and Adela Bara 460

WEBSITE INTERACTIVITY - e-Commerce Usability Perspectives in Indonesia

Vincent Didiek Wiet Aryanto 464

THE HONG KONG GOVERNMENT AUTOMATED PASSENGER CLEARANCE SYSTEM

(E-CHANNEL) - A Study of Channel Management Strategies
Tak Ming Lam 470

THE PRIVATE AND PUBLIC PARTNERSHIP STRATEGY IN E-GOVERNMENT ESDLIFE IN

HONG KONG
Tak Ming Lam 473

A WEB-BASED ARCHITECTURE FOR E-GOV APPLICATION DEVELOPMENT

Marcelo Tilli, André M. Panhan, Osman Lima and Leonardo S. Mendes 476

PROSPECTS OF GRID IN THE CURRENT VOTER REGISTRATION SCENARIO

OF BANGLADESH
Sazia Mahfuz 480

WEBSITE CREDIBILITY - A Proposal on an Evaluation Method for e-Commerce

Katsuya Watanabe, Masaya Ando and Noboru Sonehara 484

AUTHOR INDEX 489

XVI
 INVITED
SPEAKERS
KEYNOTE
LECTURES
 E-BUSINESS INNOVATION
Surviving the Coming Decades

David A. Marca
University of Phoenix, One Research Drive, Westborough, Massachusetts 01581, U.S.A.
[email protected]

Keywords: e-Business, innovation, strategy, architecture, Internet, wireless, broadband, video.

Abstract: Innovation is invention or application of technologies or theories that radically alters business and the
economy. For the last 200 years, innovation and the economy have been locked in 80-year cycles, which
might imply that innovation is an economic driver, and vice versa. Based on this relationship, some forecast
that innovation and the economy will decrease sharply due to several forces: a) rapidly decreasing economic
growth, b) increasing demand for custom services, c) more entrepreneurial work environments, and d) urban
and environmental degradation. Should such forecasts hold true, business may need to alter its offerings,
operations and organization to survive. Such a scenario may also necessitate applied e-Business innovation:
the combining of existing internet, wireless, broadband, and video technologies. One possible result: highly
flexible front offices seamlessly integrated with highly efficient back offices. Such an e-Business could
comprise: a) a customer-based and transaction-based organization, b) functions for adaptive offerings that
anticipate consumer need, c) highly responsive, real-time, operations having no inventory, and d) value-
based front-end, and automated back-end, decision making.

1 DEMOGRAPHIC CYCLES cycle (i.e. a peak spending drop) and the end of the
current 80-year technology innovation cycle (i.e.
The economy is complex but its change over time is fewer new companies, products and jobs)1. Pundits
not1. A key to understanding its change is to focus see the cause for the downturn being primarily due
on underlying forces. For example, demographic to four forces: a) rapidly decreasing economic
cycles can tell you if a market crash is just an growth, b) increasing demand for personalized
extreme correction due to an overvaluation cycle or services, c) more entrepreneurial work settings, and
the beginning of a long-term economic decline1. The d) urban and environment degradation1.
economy appears to be cyclic: it over-expands when 4,700

growth sets in and then cuts back to continue future Births

growth2. For example, since 1985, the U.S. economy 4,200

has been growing due to rising earnings, spending,

and productivity of the baby-boom generation. This 3,700

demographic has driven economic growth (Figure 1)

3,200 Peak
and corporate growth (Figure 3). Spending
2,700
1.1 The Next Decade
2,200
Innovation is often associated with invention or
1900 1920 1940 1960 1980 2000 2020 2040 2060
application of technologies or theories that alter
business and the economy3. Innovation is a major Figure 1: U.S. Births (000s) and Peak Spending (50-year 1

economic driver, and the economy is a major birth lag approximates peak spending).
innovation driver. For example, in the last 200 years,
technology innovation and the economy were locked
in upward 80-year cycles. Some forecast future U.S.
economic downturn, due in part to the demographic

IS-5
1.2 The Birth Cycle technology changes business fundamentals (e.g.
railroads brought Sears goods to remote towns).
These forecasts are based on expected population Second, no one first knows how to profit from the
growth peaks in many developed countries by 2010 new technology (e.g. 30-second TV ads). Third,
and by 2070 worldwide. The rate of change is shared infrastructure requiring large investment is
noteworthy: prosperity and urbanization appear to be needed (e.g. internet, wireless, broadband, video –
causing rapid birth rate declines in industrialized see Figure 2). Fourth, the economy is healthy, and
countries (e.g. China, India, Europe, and Japan). The thus can make such investment (e.g. the U.S.
impact to e-Business innovation could be profound. economy from roughly 1980 to 2010). Fifth, low
Decreasing population growth over the next 30 years inflation favours investment in innovative firms. By
may cause less fundamental technology innovation4 the end of the cycle, a few companies survive (e.g.
(which occurs 25-35 years after birth) and less Dell, eBay, Amazon, AOL, Yahoo!, Cisco, Google).
fundamental business innovation5 - decision making,
organizational design, management theory (which 2.1 The 80-Year Cycle
occurs 45-60 years after birth) - in that same period.
Major technology cycles last roughly 80 years and
1.3 The Spending Cycle have four stages12: Startup: Fundamental innovation
causes new companies to emerge. Growth: Those
Spending is correlated to births. U.S. Department of firms grow into the main-stream. Shakeout: Slowing
Labor reports6 show two major components to U.S. growth and overexpansion cause a consolidation.
spending. The first, the weaker of the two, is family Maturity: The surviving companies compete for
formation, which occurs 25-35 years after birth (e.g. final market share. For example, the last major U.S.
it drove the U.S. economy from 1955 to 1985). The technology cycle, computing technology (Figure 2),
second component, the stronger of the two, is peak began around 1950 with the advent of the early
spending, which occurs 45-60 years after birth and mainframes. Cheap computers, fast networks, and
may drive the economy upward until 2010. At that massive storage emerged in 1995. Internet, wireless,
time, population growth will slow, causing spending broadband and video will reach maturity by 2030.
to slow likewise. This is causing some weakening in 100% 100 GB
U.S. wireless sales7, paid internet phone usage8, 10 Tb/s

search ad revenues9, and online sales10. 1 2 3 4 5 6

80% 10 GB
10 Gb/s
1.4 The Economic Cycle S
60% 1 GB
Spending is coupled to innovation due to innovation 10 Mb/s

creating new products and jobs. For example,

fundamental (e.g. electricity, steel, and motors) and 40% 0.1 GB
10 Kb/s
applied (e.g. assembly lines) innovation brought N
standard products to mass markets, along with 20% 0.01 GB
factory jobs and urban living. Today, fundamental 10 b/s

(e.g. wireless, internet, broadband, video) and

applied (e.g. produce-to-order systems) innovation 0% 0 GB
0 b/s
are bringing custom products to affluent markets, 1940 1960 1980 2000 2020 2040
along with outsourced jobs and exurban living11 (e.g.
small quality towns and distant urban rings). But, .Figure 2: The Last Major 80-Year Technology Cycle.
sustaining such remote work depends on today’s e-
Business technologies and future innovation. Approximated market adoption s-curves:
1=mainframe, 2=minicomputer, 3=personal
computers, 4=Internet, 5=wireless, 6=broadband.

The right y-axis is logarithmic with two unit sets:

2 TECHNOLOGY CYCLES S=storage capacity in gigabytes (GB), and
N=network capacity in bites-per-second (b/s).
Technology cycles and demographic cycles feed off
each other1: population grows and then innovation
occurs, which enables more population growth. Five
forces create a technology cycle: First, radical

IS-6
2.2 The Internet S-Curve fundamental business innovation (Sloan’s product
divisions and functional units) that enabled office
It took about 20 years for the Internet to have widely work and suburbs. The coming period of decreased
accepted standards13. Similarly, cars reached 10% of spending may again require business to reorganize –
U.S. homes in 1914, and then jumped to 90% market this time to become hyper-aware of customers, and
penetration by 1928, with a shakeout occurring in the socio-economic factors that affect their buying16.
1921. Likewise, the Internet reached 10% of U.S.
homes in 1996, grew fast, then hit 50% penetration 3.1 Arrival of Mass Affluence
in 2001. Near that point, a major industry shakeout
was expected and did occur – the “Dot Com Crash” Mass affluence is the current U.S. economy3, where
happened about 80 years after the 1921 automobile “affluence” is defined as a household having income
industry crash. Internet usage has now reached its over $100,000 and net worth over $500,000, apart
maturity stage, and firms such as eBay, Amazon, from the home. In 2001, there were 20 million such
AOL and Google made it through the shakeout. households, with 30 million expected by 2009. The
latter may account for 50% of total spending. This
2.3 The Wireless S-Curve affluent market is not wildly wealthy, but is beyond
the middle-class standard of living that emerged in
Technology adoption follows an S-curve pattern: a the last economic cycle. As it ages and becomes
new technology goes “main stream” and then grows austere, this group could dominate U.S. markets, and
fast, much faster than the economy, until it reaches could reshape business for several decades (e.g.
90% market penetration. Wireless technology (e.g. demanding premium products at value prices104.
mobile phones, PDAs) penetrated U.S. markets on
pace with Internet adoption, and now electronic 3.2 Premium Market Growth
commerce services are being offered to end-users14.
It hit 10% market penetration in 1994, 50% in 2001, From 1970 to 1990, discount firms (e.g. Wal-Mart)
and hit 90% this year. Most importantly, the made goods very affordable (Figure 3). This freed
combination of internet, wireless, broadband and up discretionary income for premium goods (e.g.
video technologies is being looked upon by many as Callaway). So, standard firms (e.g. Sears) are getting
one likely next wave of e-Business innovation103. squeezed out. But the expected austerity wave may
create a new value market segment, where aging
2.4 The Broadband S-Curve consumers obtain premium products at discount
prices by using electronic bargaining agents92. In
So, a major change in e-Business is possible, and it such a market, an e-Business could exploit XML to
may occur as broadband and video technologies define and create product fragments that can later be
reach most individuals. Broadband connections (e.g. combined into highly customizable solutions17.
DSL, cable modems) hit 10% market penetration in 100%

2001, over 30% in 2004, and the shakeout is Value

formally over. Digital cameras and wi-fi networks Segment
80%
are emerging at similar rates. By 2030, the combined
technologies of internet, wireless and broadband will
reach 90% market penetration. This maturity, 60%
Standard Premium
coupled with multi-modal e-Business transactions15, Segment Segment
sets the stage for applying technology combinations 40%

to support affluent and niche markets.

20% Discount
Segment

3 CONTEXT AWARE BUSINESS 0%

1940 1960 1980 2000 2020 2040 2060

Economic and social success drives the rise-and-fall
business cycle. Each rise calls business to reorganize Figure 3: A Characterization of Product Growth.
its structures for higher population, wealth, and
standard of living. This occurred before 1914, when
fundamental technology innovation (electricity, steel
motors) enabled the assembly line, followed by

IS-7
3.3 Old Culture Scrutinized platform for distributed office work, has given way
to the instantaneous, network-centric, World Wide
Firms are now recognizing mass affluence. Affluent Web. But e-Business 2.0 cannot be fully realized
households: a) value quality over quantity, b) value without simplifying the front-end, and that occurs
service over price, c) make their own decisions, and with decentralized decision making1. Thus, complex
d) make a difference versus just doing a job1. This front-end processes, and back-end bureaucracy, must
means new consumption, and a new business model be simultaneously simplified (e.g. Figure 4).
(e.g. individual pricing for mobile e-Commerce
services18). If the old culture (devalue customer 4.1 End-to-End Transaction
time, optimize worker time, suppress worker talent
and motivation, and maximize shareholder return When e-Business is designed top-down, it leaves a
while service levels drop) is scrutinized, that may set firm incapable of defining and executing the optimal
the stage for a corporate power shift. response to the customer event that initiated a
transaction. “Produce-to-order” lets the e-Business
3.4 Corporate Power Shifts focus on the entire, end-to-end transaction from the
customer’s perspective23. The result: the spanning of
While at General Motors, Alfred Sloan invented a all parties, activities, events, responses, messages
new corporate model that gave trade-up brands to and data across the entire supply chain. For example,
the middle class19. But this model is now meeting multiple firms participating in a federated execution
diminishing returns in the face of premium product of a single transaction require: a) coordinated and
growth. Rising self-esteem and self-actualization of secure message flows, b) information exchange
workers are causing them to exit to start their own agreements, and c) coordinated message tracking24.
business or to work for higher-growth firms. A new
business model – producing personalized products – 4.2 Customer-Based Business Design
is emerging that can give a competitive edge to those
firms that understand and implement that design Satisfying an affluent market means adapting the e-
early in the next major 80-year economic cycle. Business for each customer and small market. This
means personalizing each front-end sub-transaction
3.5 New Management Model according to the customer’s language, culture, etc.
This is context-sensitive design25. The context is:
Personalized products require many fragments and customer, events, and expected responses. The
combination options. The old business model has design is what some now call “customer-based inter-
too many top-down policies trying to coordinate too organizational systems26,” and some call it activity
many processes to allow for personalized products building authenticity into operations104. As a design
or service20. The solution is the produce-to-order aid, models can be developed to explain how
model21. It coordinates real-time production through customers could use an e-Business, and such models
the automation of logistical and scheduling tasks. It have been shown to help people and systems adjust
permits direct ordering and delivery of customized operations to accommodate customer expectations27.
goods with “little to no” inventories102 and less
Buy-Specific Front-End Supply-Specific Back-End
bureaucracy – Dell being one example. Companies
that embrace this new model can use e-Business Buyer Provider Supplier Commodity

concepts and constructs to reorganize into a network Fulfill

Order
of smaller businesses, subcontractors and vendors.
Source

4 CUSTOMER-BASED Service
Service
Service Deliver Satisfy

ORGANIZATION
Charge

The produce-to-order model, with its bottom-up

Pay Bill
management powered by internet, wireless and
broadband22, leverages software, data and networks
to maximize business response and productivity. Figure 4: A Customer-Based Organization.
Wintel technology that created the client-server

IS-8
4.3 Business Replication by Market instantly across a company, a market, or a supply
chain. It is an ideal platform upon which a solution
All customers are not alike; and now, using current can be structured, configured, delivered and serviced
internet technology, like-minded customers can band to meet affluent buyers who demand personalization.
together via electronic brokers, which bargain on This section and Figure 5 give an example of how an
their behalf to acquire custom products at value adaptive38 service solution can be architected.
prices28. The result is a plethora of dynamic small
market segments! So, the front-end sub-transactions 5.1 Value Chain
should be built on an adaptable e-Business platform
that enables rapid replication and alteration to fit the The increasing solution complexity that is now
language, culture and nuances of each customer or driving buyer-supplier relations in consumer-based
market segment. Otherwise, an e-Business may not markets39 is addressed in three stages. First, all
keep up with customers changing their events and companies that touch an end-to-end transaction are
expectations around the responses to those events29. organized into a value chain38. Second, that
transactional organization is made to operate on an
4.4 Full Service Response internet-based platform that can: a) be rapidly
branded, and b) selectively opened or closed to any
To achieve high adaptability, the response to an end- customer or market segment38. Third, ebXML (i.e.,
to-end transaction should come from a distinct today’s defacto standard for message exchange,
functional entity that is standardized to yield profit30. trading protocol, common terminology, and
A best-in-class design has: a) highly modular, plug- registered process) is used by each company in the
and-play, responses, b) a public interface of defined value chain to implement its standard response40.
events and responses, and c) dynamic, context-
sensitive, class loading of non-standard responses31. 5.2 Internet Branding
This “full service response” comprises modular,
coupled, and optimized sub-transactions32. It is Each value chain member has a brand41. Highly
object-oriented with functionally-oriented, standard personalized solutions require the e-Business to: a)
components33, each of which is designed to achieve preserve brand equity for each member42, b) enable
the same level of: reputation, user trust, information brand change, emphasis and transparency, and c)
quality, functional availability and readiness, speed enable member differentiation, identification and
of response, and domain-specific characteristics33. interactivity43. For example, individuals are part of
the value chain for a contract labour solution. Today,
4.5 Browsers and Butlers people brand themselves on social networks44. The
staffing industry calls this a “video resume.” This
Personalization increases product complexity and language falls short: a resume documents experience
information distribution34. Thus, human assistants, and skills, whereas social networks can demonstrate
narrowly focused “browsers1,” are needed to help competency, demeanor, presence, and articulation.
customers choose, personalize, and use products35.
Buy-Specific Front-End Supply-Specific Back-End
These e-Professionals35 oversee: trust, security,
privacy, version and access control, configuration Buyer Provider Supplier Commodity

management and delivery dispersion34,35. Similarly, Fulfillment

Order Workflow
back-end assistants are needed, since similar buyers Management

can have conflicting priorities36. These “butlers1” Sourcing

Workflow
have a wide focus: they know the whole end-to-end
transaction in depth. This is crucial; service failures Service
Service
Service Delivery
Satisfy
Customer
are directly related to a lack of in-depth knowledge, Management Workflow Need

resulting in product/service personalization errors37. Charge

Workflow

Payment Billing

5 ADAPTIVE SOLUTIONS Management Workflow

Figure 5: A Service Solution with Real-Time Branding.

The Internet is an adaptive medium. Its mechanisms
can change a product, a service, or a brand faster
than other media. It can distribute changes almost

IS-9
5.3 Brand Transparency trust by: a) engaging with the customer, b) listening
and framing needs, c) envisioning new solutions60.
During design, each internet brand goes through a
chartering process: create, structure, communicate,
direct, manage and maintain45. “Transparency” is the 6 RESPONSIVE OPERATIONS
decision to: a) hide who is responding to achieve
transactional continuity, or b) show who is Moving to a transactional organization that produces
responding to build trust. The design is implemented adaptive solutions is a migratory activity. Success
using internet frames46 to “nest” brands within starts with mapping current operations: organization,
brands. For example, the checkout sub-transaction of codes, process, rules, and so on. Using that picture,
an online purchase may make visible the checkout operational elements are identified and modularized
vendor’s brand or logo to build trust with the buyer. into autonomous business functions that correctly
contribute to each and every response to each and
5.4 Configurable Workflows every business event, and any sub-transaction that
initial event spawns. This section and Figure 6 give
When an e-Business is a service, it comprises many an example of the result of this design activity: a full
sub-transactions, each having an operational life of service response having specific operational goals:
its own. One can describe each sub-transaction by a order and service management, sourcing strategy,
workflow47. For adaptability, all workflows should transactional alignment, and real-time reporting38.
be highly modular and granular, with well-defined
configuration rules. This enables reconfiguration of
6.1 Order Management
workflows in real-time to meet customer preferences
and business standards. For example, today, artificial Orders actually start when customer and provider
intelligence technology is used to deduce customer share organizational knowledge61 and XML forms
need, buying intent, and tendencies. This knowledge for orders and fulfillment62. The former comprises
is then used to reconfigure the workflows for each codes (e.g. departments), structure (e.g. hierarchy),
specific customer or small market segment48. and rules (e.g. designees). Once this knowledge is
shared, optimal ordering can be achieved using
5.5 Data Integrity and Privacy electronic flow down of orders and fulfilment rules.
This kind of digital binding of companies using a
Data integrity49 across workflows permits flawless single management scheme transforms operations
end-to-end transactions. Failure or delay occurs if and reduces variations and handoffs64. Note here that
sub-transactions use different data definitions or if “optimal” does not mean “minimal.” In a complex
they shirk their responsibility50. Such cases can be value chain, some suppliers become inefficient when
mitigated using secure message-oriented middleware they are electronically integrated63.
having a shared data object pool51, and passing extra
messages to verify data and function alignment. In Buy-Specific Front-End Supply-Specific Back-End
like fashion, data privacy52 should be built-in: By Buyer Provider Supplier Commodity
isolating the back-end from the front-end, response
data is protected by limiting export to the functions OLTP OLTP OLTP Internet

needed for a response53. Data privacy must include a Order Order Order
Job
Boards
policy, automated audits, and formal consent54. Structure Service Sourcing Search
Engines
Payment Billing Charges
5.6 Anticipation of Customer Need Blogs + Wikis

Social
Networks
Reengineering is not anticipatory55. Instead, back- OLAP OLAP OLAP
Digital
end butlers study analytical reports of customer Objectives Benchmarks Metrics
Communities

activity to understand collective needs56, and use that +

Run Charts
+
Trends
+
Dashboards
Niche
Web Sites
knowledge to migrate standard workflows to better
serve all customers57. At the same time, front-end
browsers are building trust58 and reputation59 with Figure 6: Adaptive Architecture for Operational Response.
customers and the value chain. Once in a trusted
position, they participate in strategic-level dialog for
new uses of the value chain. They create rational

IS-10
6.2 Sourcing Strategy 6.5 Real-Time Reporting
Sourcing is complex, especially when opportunity A best-in-class e-Business has three types of reports:
timing is a deciding factor65. Some sourcing can be Progress Reports tell how the value chain is doing
automated by: a) locating opportunities via strategy- against the customer's outsourcing objectives74.
based profiles, b) adjusting selection criteria using Best-in-class versions have two dimensions: product
prior query experience, c) maintaining opportunity or service quality, and delivery process75. Predictive
classifications, and d) using analytics to refine that Reports visualize order patterns and deduce their
classification65. Best-in-class profiles have three causes: need, business cycle, economic context,
dimensions: differentiation, cost, personalization66. etc76. Best-in-class versions anticipate customer need
Many tools now exist to "mine" internet sources67, and predict if that need can be met76. Performance
and their effectiveness hinges on building trust68. Reports trend defect, compliance, productivity, etc77.
But, companies still rely on manual sourcing when: Best-in-class versions use a variety of performance
a) supplier trust is unknown, b) duties, taxes or metrics, capture highly granular data, and display
quotas are required, c) logistics or transportation is summaries and trends via desktop dashboards78.
involved, d) transaction risks exist69. All these
factors go into an internet sourcing strategy.
7 OPTIMAL DECISION MAKING
6.3 Service Management
A transactional organization becomes virtual when
Uncontrolled product personalization (during sales) multiple companies participate in the e-Business79.
leads to inaccurate demand forecasts, high inventory Optimal decision making happens if the e-Business
investment, and poor customer service. So, just like permits autonomy, cooperation and control among
with product replication by market, the variations of business functions80. Today’s top companies use this
service delivery must be carefully controlled71. approach to adjust their operations to generate new
Service management70 after product delivery is thus electronic revenue streams and enable new service
crucial. For example, a contract labour management mix strategies81. But since functions need rules to
service is often implemented along two dimensions: operate and coordinate, a transactional organization
the actual work of the hired person, and the human becomes durable only with a flexible framework of
resources management of that person. When service rules38; specifically an XML-based framework for
management succeeds, the tactical objective of the optimal decision making. This framework has two
value chain solution is met: well-managed orders; dimensions: standards-oriented (industry, company,
fulfilment within customer-desired service levels; customer)83, process-oriented (language, practices,
satisfactory quality of the delivered commodity; service)82. This creates a design space (e.g. Figure
good follow-on service. Tactical fulfilment paves the 7a, b, and c) for XML-based code sets that together,
way for the provider to become a trusted advisor. can support value-based front-end and automated
back-end decision-making for an e-Business1.
6.4 Transactional Alignment
Controls
Nothing can ruin response faster than transactional
misalignment. Response requires sub-transactions to
work correctly, and correctly together72.
Misalignment causes product, process, or data errors
– the results of which require costly adjustments Inputs e-Business Outputs
after transaction closure. For example, incorrect Function
state tax on labour, an error found in contract labour
solutions, creates charges which require subsequent
adjustment. These can be eliminated if all sub-
transactions share the same tax rules. So, the design
or deployment of an e-Business should be reviewed Mechanisms
end-to-end to ensure function, data, and rule
alignment among all of its sub-transactions73. Figure 7a: An e-Business Function and its Four Interfaces.

IS-11
7.1 Language-based Decision Rules 7.3 Service-based Decision Rules
The semantic automation of an e-Business starts With language and practices in place, service levels
with industry-specific terminology that is captured are then defined. The first set comprises an industry
in an ontology84. This framework comprises five benchmark for each sub-transaction96. When added
distinct sets of terms: a) process, b) information, c) together, a true measure of "response" is obtained
application, d) data, and e) infrastructure85. Optimal for each end-to-end transaction96. These metrics feed
operations can occur when terms are standardized86. the performance reporting system. Inside industry
But language must be a strategic consideration; a benchmarks are metrics for SOP. Besides speed and
shared company language is insufficient unless it cost97, they can measure usability, trust, loyalty,
accounts for nationalistic interpretations of words innovation, flexibility and financial impact98, and
and their implied meanings87. Within the context of they can occur when the buyer, or the supplier, or
intensified competition in a market that demands both, are mobile98. These metrics feed the predictive
high personalization, the ability to market, sell, and reporting system. If the customer requires service
support using the customer's language becomes levels that exceed industry benchmarks and SOP99,
imperative88. So, an e-Business architecture should the customer-facing business functions receive new
contain: a) a message-passing mechanism for buyer- metrics. Besides speed and cost100, service levels can
supplier interactions, and b) a declarative language define product quality and service quality101. These
for expressing customer requests89. metrics feed the progress reporting system.
Interface Scope Contents How Implemented Interface Focus Industry Standard Company Standard Customer Required

Input What Data Definitions XML, XML Schema Input Language Ontology Meta Data Meta Data

Output What Data Definitions XML, XML Schema Output Language Ontology Meta Data Meta Data

Control When Events, Triggers, Coordination XML, XML Schema, ebXML Control Service Benchmarks Company SLAs Customer SLAs

Standard Operating
Mechanism How Algorithms, Standards, Regulations ebXML Mechanism Practice Best Practices Customer Functions
Procedure

Overlay Sequence
Figure 7b: Process-Oriented Interface Definitions. 1. Achieve 2. Create 3. Comply with
Parity with Competitive Contract Terms
Industry Advantage and Conditions
7.2 Practice-based Decision Rules
Figure 7c: Standards-Oriented Rule Set Overlays.
Once language is settled upon, standard practices
can emerge. At the highest level, these are industry
best practices90. When encoded in XML, including 8 SUMMARY AND
measurable objectives and dependency relationships
among practices91, they can easily be embedded into
CONCLUSIONS
manuals or systems. The second level of practices,
often called standard operating procedures (SOP), e-Business is here to stay, but it may likely change
simultaneously enables business replication by over the next 15 years as population growth declines
market and full service response92. When encoded in impact innovation and how internet, wireless,
XML, this company knowledge can be immediately broadband and video technologies are combined and
deployed exactly where and when needed93. The used to conduct e-Business. One possible future
third level of practices, used during replication by scenario is the increase in applied innovation, built
market, adjusts the SOP to meet customer needs on new XML-based architectures, for providing
(e.g. market niche, natural business cycle, unique highly flexible front-end and highly standard back-
value proposition, special contracts) 94. The best way end e-Business platforms. To summarize:
to adjust is to overlay customer operations atop the
SOP. This way, company functions are not changed, 8.1 More Applied Innovation
but instead are replaced, by customer functions95.
Innovation is invention or application of theories or
technologies that radically alters business and the
economy. Since innovation is a major economic
driver, and vice versa, any economic downturn could
impact e-Business innovation. Slowing population

IS-12
growth and exiting aged workforces can deplete both their customer best. They can apply standard rules
the number of inventors and those capable of using appropriately, and can overlay customer-specific
investments for fundamental business innovation. In rules atop the corporate standard. At the same time,
parallel, the simultaneous maturation of the Internet, back-end “butlers” use their “all customer”
wireless, broadband and video may provide new perspective to improve the standard rule set. In this
opportunities for combining these technologies into way, best practices are centrally defined and
solutions for the mass affluence economy. More immediately distributed to the front lines. Context-
applied, and less fundamental, innovation is likely. sensitive e-Business (i.e. high-flex front-ends with
standard back-ends) may become the norm.
8.2 Architecture Innovation
Those consumers capable of demanding premium, ACKNOWLEDGEMENTS
personalized products dominate the mass affluence
economy. These consumers are also tech-savvy, and
Thanks go to Mr. Harry S. Dent (www.hsdent.com)
can exploit the Internet by creating intelligent agents
that build dynamic small markets of like-minded for his research on the interrelationship among
consumers to negotiate with suppliers for premium, population growth, the economy, and technology
personalized products at value prices! Since this innovation, and to Dr. Clement McGowan for his
market could grow to over 50% of the overall insightful review of earlier drafts of this paper.
economy, e-Business must become hyper-aware of
their needs and buying patterns, and must be
“architected” into a highly flexible front-end (to flex REFERENCES
to each individual or small market) and a highly
standardized back-end (to enable cost-effective 1. Dent, H. (2004). The Next Great Bubble Boom. The
operations). Architecture innovation could be a key Free Press.
to e-Business survival in the coming decades. 2. Gunderson, L., Holling, C. (2002). Panarchy:
Understanding Transformations in Human and Natural
Systems. Island Press.
8.3 XML Platform of Contracts 3. Nicholas, A. (2002). “Innovation: The Pathway to
Threefold Sustainability,” in Technology and Law.
Internet technology now exists to “architect” Massachusetts Institute of Technology.
flexibility with standardization. A crucial technology 4. Pesola, J. (2005). “Banking Fragility and Distress,”
is XML. It is an internet-based software language for Bank of Finland Discussion Papers, Bank of Finland.
defining business language, practices and service 5. Hoge, J. (2004). “A Global Power Shift in the
levels (thus enabling standardization), and allowing Making,” Foreign Affairs, 83(4).
them to be easily changed and distributed across all 6. U.S. Department of Labor, (2007). “Consumer
Expenditure Survey,” Bureau of Labor Statistics,
business functions (thus enabling flexibility). Couple http://www.bls.gov/cex/.
this mechanism with an architecture of “customer 7. Richtel, M., Tedeschi, B. (2007). “Online Sales Lose
within company within industry,” and you have a Steam,” New York Times, Jun 17.
platform for standardizing business contracts and 8. Sullivan, D. (2006). “The Need for Search Ad
then overlaying the more esoteric rules required for Revenue to Stand Alone,”
meeting the varied and changing demands of the http://blog.searchenginewatch.com/.
individuals and small markets in the mass affluence 9. Yen, Y. (2007). “Can't Anyone Make Internet Phones
economy. Future market forces could likely make an Pay?” Fortune, October 2.
XML-based contracts platform commonplace. 10. Harrison, C. (2001). “Nokia Predicts Slow Recovery
for Wireless Sales,” Dallas Morning News, Nov 28.
11. Silverstein, M., Fiske, N. (2004). “Trading Up,”
8.4 Front-Line Decision-Making Journal of Macromarketing, Dec.
12. Poiré, N. (1994). “Social Dynamics and the
Such architecture requires an investment – shed top- Investment Cycle,” CYCLES magazine, Mar/Apr.
down control, reduce bureaucracy and decentralize 14. Silva, F., Rosa, P. (2006). “The Quest for the Web
decision-making. The XML platform of contracts Services Stack,” in proc. ICE-B’06.
lets a company define its operational rules while also 15. Sanchez, E., et. al. (2006). “Service-Oriented
enabling giving up control for how to apply those Architecture for Mobile Phones,” in proc. ICE-B’06.
16. Rigas, D., Gazepidis, N. (2006). “Facial Expressions
rules for each customer or small market segment.
and Body Gestures as Metaphors for B2C Interfaces,”
That is the job of front-line “browsers,” who know in proc. ICE-B’06.

IS-13
17. Sanchez, E., Chavez, F. (2006). “Smart Diffusion of of Physical Distribution & Logistics Management,
Government Contents,” in proc. ICE-B’06. 30(3).
18. Isiklar, G., et. al. (2006). ”A CRM-Based Pricing 40. Bechini, A., et. al. (2007). “Document Management
Model for m-Commerce Services” in proc. ICE-B’06. for Collaborative e-Business,” in proc. ICE-B’07, Jul.
19. Drucker, P. (2003) Peter Drucker on the Profession of 41. http://en.wikipedia.org/wiki/brand.
Management. Harvard Business School. 42. Keller , K. (1993). “Conceptualizing, Measuring, and
20. Microsoft (2006). “Unified Communications and Managing Customer-Based Brand Equity,” Journal of
Collaboration: Pervasive Capabilities for Where and Marketing, 57(1).
How People Work,” Microsoft White Paper, July. 43. Ries, A., Ries, L. (2000). The 11 Immutable Laws of
21. Dejonckheere, J., et. al. (2003). “Measuring and Internet Branding. Harper Information.
Avoiding the Bullwhip Effect,” in European Journal of 44. http://en.wikipedia.org/wiki/social_networks
Operational Research 147(3). 45. Macrae, C., Uncles, M. (1997). “Rethinking Brand
22. Kalakota, R., Robinson, M. (1999). E-Business 2.0: Management: the Role of Brand Chartering,” Journal
Roadmap for Success. Addison-Wesley. of Product & Brand Management, 6(1).
23. Bhimani, A. (1996). “Securing the Commercial 46. Tolly, K. (1997). “The Great Networking Correction:
Internet,” CACM, 39(6). Frames Reaffirmed,” Internet Computing, 1(5).
24. Sahai, A., et. al. (2002). “Message Tracking in SOAP- 47. http://en.wikipedia.org/wiki/workflow
Based Web Services,” in proc. Network Operations 48. Lopes, C., Melo, J. (2006). “Business Improvement
and Management Symposium. through Automatic Workflow Modeling,” in proc.
25. Otto, S., Kirn, S. (2006). “Adaption in Distributed ICE-B’06, Aug.
Systems,” in proc. 8th Annual conf. on Genetic and 49. http://en.wikipedia.org/wiki/data_integrity.
Evolutionary Computation. 50. Bhimani, A. (1996). “Securing the Commercial
26. Grover, V., (2007). “An Empirically Derived Model Internet,” CACM, 39(6).
for the Adoption of Customer-based Inter- 51. Yamada, H., Kawaguchi, A. (2006). “Message
organizational Systems,” Decision Sciences, Jul. Queuing Middleware on VPN and Its Traffic Control
27. Vilpola, I., et. al. (2006). “Applying Contextual design Scheme,” in proc. ICE-B’06, Aug.
to ERP System Implementation,” in proc. conf. on 52. http://en.wikipedia.org/wiki/privacy.
Human Factors in Computing Systems, CHI'06. 53. Guha, S., Francis, P. (2005). “Architecting a Secure
28. Elofson, G., et. al. (1998). “Creating a Custom Mass- Internet,” in proc. SOSP’05, Oct.
Production Channel on the Internet,” CACM, 41(3). 54. Friedman, R. (2008). “Protecting Customer Privacy,”
29. Otto, S., Kirn, S. (2006). “Adaption in Distributed InformationToday, Jan.
Systems,” in proc. GECCO/’06. 55. Bessant, J., Caffyn, S. (1997). “High-involvement
30. Palmer, J., et. al. McKee, P., Fisher, M. (2007). Innovation through Continuous Improvement,” Intl.
“Optimizing Revenue,” in proc. ICE-B’07. Journal of Technology Management, 14(1), Jan.
31. Sponziello, A. (2007). “Design Strategies for Web 56. http://en.wikipedia.org/wiki/continuous_improvement.
Based ITS Applications,” in proc. ICE-B’07. 57. Blumenthal D., Kilo C. (1998). “A Report Card on
32. Kirchmer, M. (1999). Business Process Oriented Continuous Quality Improvement,” The Milbank
Implementation of Standard Software: How to Quarterly, 74(4), Apr.
Achieve Competitive Advantage, Springer-Verlag. 58. Amoretti, M., et. al. (2006). “Reputation Management
33. Palmer, J., Lindemann, M. (2003). “Business Models Service for Peer-to-Peer Enterprise Architectures,” in
and Market Mechanisms,” The DATABASE for proc. ICE-B’06, Aug.
Advances in Information Systems, 34(2). 59. Louta, M., et. al. (2006). “Reputation Based Intelligent
34. Sheu, R., et. al. (2000). “A Collaborative File Agent Negotiation Frameworks in the e-Marketplace,”
Management System on the WWW,” in proc. 8th intl. in proc. ICE-B’06, Aug.
conf. on High Performance Computing, Springer- 60. Maister, D., et. al. (2000). The Trusted Advisor. The
Verlag. Free Press.
35. Martinez, A., et. al. (2007). “Towards a Semantically 61. Holsapple, C. (2004). Handbook on Knowledge
Augmented Collaborative Working Environment,” in Management. Springer-Verlag.
proc. ICE-B’07. 62. Warkentin, M., et. al. (2001). “e-Knowledge Networks
36. Curvelo, T., et. al. (2007). “THOTH – A framework for Inter-Organizational Collaborative e-Business,”
for Building Secure TV Broadcasters Networks over Logistics Information Management 14(1/2).
the Internet,” in proc. ICE-B’07. 63. Stewart, G. (1995). “Supply Chain Performance
37. Oppenheimer, D., et. al. (2003). “Why Do Internet Benchmarking Study Reveals Keys to Supply Chain
Services Fail, and What Can Be Done About It,” in Excellence,” Logistics Info. Management, 8(2).
proc. 4th USITS. 64. Kalakota, R., Robison, M. (1999). e-Business:
38. Marca, D. (2005). Open Process Frameworks, IEEE Roadmap for Success. Addison-Wesley Information
Press, John Wiley Publishing Company. Technology Series.
39. Graham, G., Hardaker, G. (2000). “Supply-Chain 65. Bai, J., et. al. (2006). “Facilitating e-Business by
Management across the Internet,” International Journal Retrieving Relevant Business Opportunities on the
Internet,” in proc. ICE-B’06.

IS-14
66. Pearcy, D., Giunipero, L. (2007). “An Empirical 86. Marschan, R., et. al. (1999). “Adopting a Common
Investigation of the Impact of Corporate Strategy and Corporate language: IHRM Implications,”
Existing Buyer-Supplier Relationships on the Use of International Journal of Human Resource Management
Electronic Procurement Applications in US-Based 10(3).
Firms,” in proc. ICE-B’06. 87. Marschan, R. (1997). “Language: The Forgotten
67. Tucker, D., Jones, L. (2000). “Leveraging the Power Factor in Multinational Management,” European
of the Internet for Optimal Supplier Sourcing,” Intl. Management Journal, 15(5).
Journal of Physical Distribution & Logistics 88. Lankshear, C. (1998). “Language and the New
Management, 30(3/4). Capitalism,” Intl. Journal of Inclusive Education.
68. Amoretti, M., et. al. (2006). “Reputation Management 89. David J., et. al. (1999). “SMART: Towards Spatial
Service for Peer-to-Peer Enterprise Architectures,” in Internet Marketplaces,” Geoinformatica, 3(2), Jun.
proc. ICE-B’06. 90. Bhutta, K., Huq, F. (1999). “Benchmarking – Best
69. Sowinski, L. (2000). “Sourcing Over the Internet,” Practices: an Integrated Approach,” Benchmarking:
World Trade, 13(8). An International Journal 6(3).
70. Fitzsimmons, J., Fitzsimmons, M. (2001). Service 91. Davies, A., Kochhar, A. (2000). “A Framework for the
Management: Operations, Strategy, and Information, Selection of Best Practices,” Intl. Journal of
Technology. McGraw-Hill. Operations & Production Management, 20(10).
71. Lee, H., (1996). “Effective Inventory and Service 92. Resinas, M., et. al. (2006). “Towards Automated
Management through Product and Process Redesign,” Service Trading,” in proc. ICE-B’06.
Operations Research, 44(1). 92. Sveda, M., Vrba, R. (2006). “Internet-Based
72. Blumenthal, M., Clark, D. (2001). “Rethinking the Embedded System Architectures,” in proc. ICE-B’06.
Design of the Internet,” ACM trans. on Internet 94. Yahia, I. (2006). “B2C Virtual Communities:
Technology, 1(1). Typology & Associated Benefits,” in proc. ICE-B’06.
73. Anderson, B., et. al. (2006). “The Application of 95. Raivio, Y., Luukkainen, S. (2006). “Digital Rights
Model Checking for Securing e-Commerce Management in the Mobile Environment,” in proc.
Transactions,” CACM, 49(6). ICE-B’06.
74. Storck, J., Hill, P. (2000). “Knowledge Diffusion 96. Hellerstein, J., et. al. (1999). “ETE: a Customizable
through Strategic Communities,” in Knowledge and Approach to Measuring End-to-End Response Times
Communities, Butterworth-Heineman. and their Components in Distributed Systems,” in
75. Hawkins, A. (1968). “A Predictive Reporting System proc. 19th intl. conf. on Distributed Computing Sys.
for Management,” Journal of Mgt Studies, 5(3). 97. Decker, M., Bulander, R. (2006). “Notification
76. Shan, J., et. al. (2005). “Dynamic Modeling and Services for Mobile and Wireless Terminals,” in proc.
Forecasting on Enterprise Revenue with Derived ICE-B’06.
Granularities,” in proc. of Granular Computing. 98. Nojoumian, M., Lethbridge, T. (2006). “A New
77. Banker, R., et. al. (1993). “Manufacturing Approach for the Trust Calculation in Social
Performance Reporting for Continuous Quality Networks,” in proc. ICE-B’06.
Improvement,” Management International Review. 99. Osterwalder, A. (2002). “An e-Business Model
78. Walker, B., (1996). “Corporate Performance Ontology for Modeling e-Business,” in proc. 15th
Reporting Revisited,” Journal of Industrial Bled Electronic Commerce Conference.
Management & Data Systems 96(3). 100. Parasuraman, A., et. al. (2006). “Perceived Service
79. Travica, B. (2005). “Virtual Organization and Quality as a Customer-based Performance Measure,”
Electronic Commerce,” The DATABASE for Human Resource Management, 30(3).
Advances in Information Systems, 36(3). 101. Cheung, C., Lee, M. (2002). “Consumer Satisfaction
80. Keidel, R. (1995). Seeing Organizational Patterns: A with Internet Shopping,” in proc. 7th Intl. conf. on
New Theory and Language of Organizational Design. Electronic Commerce.
Berrett-Koehler Publishers. 102. Anderson, C. (2006). The Long Tail. Hyperion
81. Gerwig, K. (2001). “Disruptive Behavior,” Business: Publishers.
the 8th Layer, Sep. 103. BBC. (2008). “Nokia Pushes 4G Wireless Plans,”
82. Heasley, J. (2005). “Securing XML Data,” in proc. BBC News Wire, April 15.
InfoSecCD’04, Oct. 104. Kiviat, B., Cloud, J., Fox, J. (2008). “10 Ideas that
83. Chifu, V., et. al. (2007). “Ontology Model for are Changing the World,” Time Magazine, March 24.
Traceability in the Food Industry,” in ICE-B’07.
84. Alexopoulos, P., et. al. (2007). “Towards a Generic
Fraud Ontology,” in proc. ICE-B’06.
85. Petersen, M. (2000). “Demand Activated BRIEF BIOGRAPHY
Manufacturing Architecture for Supply Chain
Collaboration,” Resource Relation Intl. conf. on David A. Marca is on the Adjunct Faculty at the
Modeling and Analysis of Semiconductor University of Phoenix. His six books and 24 papers
Manufacturing, Mar. cover e-Business, e-Commerce, business process
reengineering, and software engineering. He holds a

IS-15
patent in workflow technology. His last book,
entitled “Open Process Frameworks: Patterns for the
Adaptive e-Enterprise,” was published by the IEEE
in 2006. David is also President of OpenProcess,
Inc. – an e-Business consulting firm since 1997 –
that helps firms implement workforce management
and e-Business solutions. He has consulted in Italy,
Norway, Mexico and the United States. David is a
member of the Institute of Electrical and Electronic
Engineers (IEEE), the Association for Computing
Machinery (ACM), and the Project Management
Institute (PMI).

IS-16
 IMPROVING RELIABILITY IN COMMERCIAL IP NETWORKS

Yaakov Kogan
AT&T Labs, 200 Laurel Avenue, Middletown, NJ, U.S.A.
[email protected]

Keywords: IP network, IP router, availability, reliability, redundancy.

Abstract: As the Internet becomes an increasingly critical communication infrastructure for business, education and
civil society in general, the need to understand and systematically analyze its reliability becomes
progressively more important. An Internet Service Provider (ISP) faces a challenge of providing service that
meets customer expectations in terms of price and reliability while continuously reducing its cost. In fact,
changes in Internet technology (particularly software) are significantly more frequent and less rigorously
tested than used to be in circuit-switching telephone networks. An ISP can wait until the technology will
mature but then it faces the risk of losing customers and revenues in a situation where many customers care
more about low prices than claims of better quality and reliability. A large ISP has to meet high reliability
requirements for critical applications like financial transactions, Voice over IP and IPTV. This results in
variety of redundancy solutions at the edge and resilient core which is shared by traffic from all
applications. In this paper we review redundancy solutions for eliminating the customer impact in typical
failure modes in IP networks and present a constructive metric for evaluating the reliability of commercial
IP networks.

1 INTRODUCTION more economically and technically efficient

multiplexing of network resources. Second,
The ARECI Study (Bell Labs, 2007) conducted for protocols like MPLS TE with FRR are more
the European Commission states “European economically efficient in restoration from failures of
security, economic stability, and the public safety transmission facilities than traditional 1+1
and welfare of its citizens increasingly depend on redundancy. Third, commercial IP networks can
the availability and robustness of its electronic provide different level of redundancy to multiple
communications infrastructures”. The main focus of services and price the service dependent on its
the Study is the crucial subject of the end-to-end reliability. That allows Internet service providers to
availability of communication networks. It satisfy many customers who care more about low
emphasizes that “European citizens are used to the price than reliability (claffy, Meinrath and Bradner,
high reliability of legacy telephone service and come 2007) and recover high cost of redundancy with
to expect new services (e.g., VoIP, Internet, IPTV) higher price for mission critical applications.
to have similar level of reliability. Legacy telephone This paper describes redundancy solutions for
service uses circuit switching, allocating an eliminating the customer impact in typical failure
exclusive path with fixed capacity across the modes in IP networks and provides the framework
network for the duration of the call regardless of the for quantification of reliability for commercial IP
actual usage of the circuit capacity. The reliability networks.
objective of ‘no more than 2 hours downtime in 40
years’ has been applied to voice communication
since 1964 (Malec, 1998). It has been achieved 2 REDUNDANCY SOLUTIONS
using expensive redundancy solutions for both
switches and transmission facilities. Commercial IP Redundant elements in IP routers and networks are
networks have three main advantages in comparison necessary to provide service restoration without
with legacy telephone networks. First, packet human intervention. However special mechanisms
switching as the transport mechanism provides far and protocols are required to reduce restoration time

IS-17
to sub-second level (required for video Tele- ƒ Bidirectional Forwarding Detection (BFD)
Presence) and further below 100 milliseconds provides fast failure detection time (on the order
(required for broadcast video and some financial of tens of milliseconds) between customer and
applications). The total restoration time in IP access routers. BFD can be used to trigger
network can be broken up into the following three reroute to the backup access router if the
phases:
primary access router fails.
ƒ Detection: time required to detect and react to a
failure;

ƒ Notification: time required to communicate the 3 THE FRAMEWORK FOR

failure in the previous phase to all relevant RELIABILITY ANALYSIS
devices in the network;
For reliability analysis, IP network consists of
ƒ Restoration: time required to move traffic to an routers interconnected by transmission facilities.
alternate network device or path. Routers are responsible for packet delivery to the
destination while facilities provide the necessary
In this section we list key redundancy solutions
bandwidth to avoid packet loss. An ISP typically
and protocols which eliminate service interruptions
sells to customers access ports. These ports are
that can be caused by router and facility outages.
housed on Line Cards that are on the “drop side” of
ƒ Non Stop Routing (NSR) provides hitless an Access Router, where facilities from a Customer
failover to backup Route Processor when Router terminate.
primary RP fails. It eliminates simultaneous There are five element types in a typical IP
failure of all line cards in case of RP failure. Access Network topology whose failure can cause
downtime for customer facing Line Cards:
NSR is essential for achieving the required
MTBO level for the Backbone network. NSR is ƒ Line Card on the customer facing side. Any
the first building block for In Service Software failure in the electronic or optical components
Upgrade (ISSU) that must eliminate or of the Line Card will result in Line Card
significantly reduce planned outages due to downtime.
software upgrade.
ƒ Access Routers that form an edge on an ISP
ƒ Fast Reroute (FRR) is a feature in an MPLS Backbone network. A total router failure will
network that enables the service provider to result in downtime for all Line Cards
provide a redundant data path should a link
ƒ Facilities and supporting elements such as
(facility) or router line card fail. FRR is
cross-connects, which link Access Routers to
intended to offer practically hitless restoration
Backbone Routers. To increase the reliability of
with failover times on the order of tens of
the Access Network, an ISP usually provides
milliseconds. To achieve such a fast restoration
redundancy by connecting each Access Router
time, a backup Label Switched Path (LSP) is
to two Backbone Routers at the same access
established at each node. The traffic is switched
node using two independent sets of uplinks.
to the backup LSP once a failure has been
This permits customer traffic to enter the
detected in the primary LSP. Typical budget for
Backbone in the following failure scenarios:
achieving 50 milliseconds for the total
restoration time is 10 milliseconds for Detection - A failed uplink
and Notification and 40 milliseconds for
Restoration (traffic redirection). Today only - A failed card supporting an uplink
facility failures can be detected and notified
- A failed Backbone Router at the access node
within 10 milliseconds. Reduction of detection
and notification time for other failures is work If all facilities linking an Access Router to a
in progress. Backbone Router fail, then all Line Cards at
the Access Router will experience downtime.

IS-18
 ƒ Backbone Routers linked to Access Routers. If two diverse access routers. To obtain the desired
both Backbone Routers at an access node fail (a End-to-End MTBO of 40 years we need to allocate
rare event) then all Line Cards on the Access 75 years to the Backbone MTBO. Assume that mean
Routers at this node lose connection to the time between failures of Backbone facilities or
Backbone. router cards is 2000 hours, and that, on the average,
one such failure without hitless restoration affects
ƒ Facilities linking Backbone Routers at an access traffic on 2% of customer facing Line Cards. Then
node, to Backbone Routers at other Backbone the Backbone coverage is 85% which can be
realistically achieved with NSR and FRR described
nodes. Such facility failures decrease the
in the previous section.
available bandwidth from Access Routers to the
Backbone. Note that if all Backbone Router
uplinks at an access node fail (a rare event), REFERENCES
then all Line Cards on the Access Routers at
this node lose connection to the Backbone. Bell Labs and Professional Services–Alcatel-Lucent,
2007. Availability and robustness of electronic
Thus, the access Line Card acts as a common
communications infrastructures. “The ARECI study”.
denominator for all of the above failure types. Any Final Report, March 2007. ECSC–EC–EAEC,
one of these failures results in downtime for the Brussels–Luxembourg.
impacted Line Cards. In addition, failures in the http://ec.europa.eu/information_society/newsroom/cf/i
Backbone without hitless restoration may interrupt temdetail.cfm?item_id=3334.
traffic between many access Line Cards. Malec, H., 1998. Communications reliability: A historical
Let N be the total number of access (customer perspective. IEEE Transactions on Reliability, 47, 333
facing) Line Cards and traffic were impacted on K – 345.
of these card during a fixed time interval T (e.g., one claffy, kc, Meinrath, S. and Bradner, S. 2007. The
(un)Economic Internet? IEEE Internet Computing, 11,
month measured in hours). Then Mean Time
53– 58.
between Outages (MTBO) is defined as

MTBO = NT/K (1)

BRIEF BIOGRAPHY
Table 1 demonstrates application of MTBO for
calculation of the probability of successful failover Yaakov Kogan received the Ph.D. and Doctor of
(Coverage) in the Backbone required to achieve the Sciences degrees from the USSR Academy of
End-to-End MTBO of 40 years. Assume that MTBO Sciences in 1969 and 1987, respectively. He joined
for an Access Router without redundancy is 7 years. the Network Design and Performance Analysis
This case is listed in Table 1 as ‘Single Access’. Department at AT&T in 1993 where he is currently
a Lead Member of Technical Staff. During 1989 –
Table 1: Steps toward End-to-End MTBO of 40 years. 1993 he was a professor at the Faculty of Industrial
Engineering and Management of the Technion
Connection Coverage MTBO, years
University. After receiving the Ph.D. he worked on
Single Access 7
performance analysis and measurements of
Dual Access 96% 172
Both Ends 86 computer and communication systems and
Backbone 85% 75 developed nonparametric and asymptotic methods
End-to-End 40 for solving stochastic models of large dimension. He
published 5 books and more than 100 papers. Dr.
The probability of successful failover to the backup Kogan is a member of IFIP Working Group 7.3 on
device (e.g., router) when primary device fails is Computer Systems Modeling since 1979 and IEEE
referred to as Coverage. Row ‘Dual Access’ in Table Fellow since 2001. His recent activities include
1 shows that MTBO can be dramatically increased performance and reliability analysis of large IP and
by providing dual homing, where customer router is Frame Relay networks.
connected to two access routers located in different
Backbone nodes if we assume that the respective
Coverage is 96%. The next row in Table 1 takes into
account ‘another end’ of the network with similar

IS-19
 CRYPTOGRAPHIC ALGORITHMS
Successes, Failures and Challenges

Bart Preneel
Katholieke Universiteit Leuven, Dept. Electrical Engineering-ESAT/COSIC
Kasteelpark Arenberg 10, bus 2446, B-3001 Leuven, Belgium
IBBT, Van Crommenlaan, B-9000 Gent, Belgium
[email protected]

Keywords: Cryptology, information security.

Abstract: The exponential progress of hardware during the past decades and the explosion of wired and wireless net-
works has resulted in a large scale deployment of cryptography in financial applications and electronic com-
merce, in mobile phones and in electronic identity cards. In parallel with his deployment the insight has grown
that building secure information systems is a very hard problem: cryptography is an essential building block
that may have helped us to win some battles, but one still has the impression that we are losing the information
security “war.” This article clarifies the role of cryptology in information security. Subsequently it discusses
the challenges that are faced by this discipline. There is a strong need for algorithms that offer better trade-
offs between performance, cost and security. In addition we face challenges related to secure implementations
in software and hardware and in the area of algorithm agility. We will conclude with a brief discussion of
opportunities offered by the progress made in the area of cryptographic protocols.

1 INFORMATION SECURITY such as secure key storage and remote attestation in

AND CRYPTOLOGY tens of millions of computers. Most governments
are introducing electronic identity cards or electronic
Cryptology is the science that studies mathematical health cards to offer secure on-line services to their
techniques to protect digital information (Menezes citizens. One can anticipate that in a future world with
et al., 1997). It allows to identify entities in a re- trillions of electronic devices (known as wireless sen-
liable way, to establish secure communications over sor networks, the Internet of things, ubiquitous or per-
open networks, and to protect stored information. vasive computing,. . . ) we will see a further increase
At a higher abstraction level, cryptology enables the of cryptographic deployments.
collaboration of mutually distrusting parties towards During the last thirty years, cryptology has
achieving a common goal. Cryptology is essential for evolved from very limited use in restricted military
securing communications; one can think of mobile and government environments towards a massive de-
networks such as GSM and 3GSM, Virtual Private ployment. This development has been driven by the
Networks, Wireless LANs and Personal Area Net- needs of the information society and has been enabled
works (e.g., Zigbee and Bluetooth). Cryptology also by the same development in microelectronics that has
plays a key role in securing financial transactions; the created this society in the first place. In parallel the
financial world uses cryptology in its back-office, but science of cryptology has been progressed by build-
also for identification in terminals (e.g. the EMV stan- ing on complexity theory and information theory, in
dard (EMV, 2004)), and for Internet banking. The which security proofs (or more correctly security re-
TLS protocol in browsers is widely used for secur- ductions) play an ever growing role. In this context
ing e-commerce transactions. There is also a grow- the concepts are refined and we have substantially
ing use of cryptology for hard disk encryption; the improved our understanding of the properties needed
Trusted Computing Group has introduced the TPM from the building blocks. Nevertheless, the develop-
chip (Trusted Platform Module) that offers services ment of many of these building blocks is still an engi-

IS-21
neering discipline where we learn and develop design stantial progress is understanding how humans inter-
principles by trial and error methods. act with ICT systems and in particular how humans
It is important to realize that in spite of the strong interact with security.
scientific track record, the cryptography problem is
not solved. Cryptographic algorithms and protocols
get frequently cryptanalyzed, because of progress in
cryptanalysis and advances in computational power.
2 RESEARCH CHALLENGES
Once the cryptography is broken, the security will FOR CRYPTOGRAPHIC
probably collapse (e.g., decrypting sensitive informa- ALGORITHMS
tion or copying controlled information becomes pos-
sible). This is very similar to a scenario in which This section identifies two challenges for crypto-
a burglar would get access to a master key. In the graphic algorithms: offering improved trade-offs be-
best case one can use fraud detection and respon- tween cost, performance and security and secure im-
sive technologies to control the losses, which would plementations.
correspond in our building analogy to adding video
surveillance and security guards. However, most se-
curity breaches are completely unrelated to crypto-
2.1 Improved Trade-offs for
graphic weaknesses (see for example Anderson (An- Cryptographic Algorithms
dreson, 1994)). The most important reasons for these
breaches are: Cryptographic algorithms offer a trade-off between
• incorrect or incomplete requirements for the sys- cost (footprint in hardware or software, power and/or
tem (this includes the case where a system is de- energy consumption), security and performance.
signed for one application is later on used for Achieving any two of these three is straightforward
other applications with different security require- provided that one can neglect the third one. It is triv-
ments); ial to make an insecure system that is fast and cheap.
If cost is not an issue, it is rather easy to make a very
• implementation errors (e.g., software bugs, cryp- fast and very secure solution. Similarly, if a system
tographic hardware or software with side channel can be very slow, it can offer an acceptable security
weaknesses); at a low cost. The real challenge is to improve these
• configuration errors (this includes the case where three parameters at the same time, in particular for
algorithms and implementations known to be specific extremes beyond the current state of the art,
weak are not upgraded); namely:
• human issues: the weakest link in any security Extremely Low Cost Solutions are essential to get
system is humans; this has been illustrated by cryptography everywhere (ambient intelligence,
many war stories of hackers who use social en- sensor networks and RFIDs). A specific target is
gineering techniques to gain access. encryption with less than 1500 gates or an entity
The main reason why it is so difficult to make sub- authentication protocol that consumes less than
stantial progress is that our ICT infrastructure gets 50 mJoules.
ever more complex: ICs have hundreds of millions Extremely Fast Solutions for applications such as
of transistors, processors will have dozens of cores, bus encryption, and authenticated encryption for
operating systems consist of tens of millions of lines Petabyte storage devices and Terabit networks.
of code, applications are built from millions of lines
Need for High Security Solutions for applications
of code and we have hundreds of millions of devices
such as e-voting, e-health and national security we
connected to the Internet. All these numbers will
need cryptographic algorithms that provide guar-
increase by another order of magnitude in the next
anteed protection for 50 years or more. While this
decade. Securing complex systems is inherently diffi-
is conceivable for symmetric cryptography, this
cult, and our approach to build simple and trustworthy
goal is currently a major challenge for public key
subsystems with core cryptographic functionality is
cryptography, in view of progress in research to
only partially effective. One of the future challenges
attack hard mathematical problems and the antic-
is to learn how to isolate this functionality and how
ipated development of quantum computers.
to integrate it with the rest of the system. We will
also have to learn how to build resilient and adaptive This challenge will be illustrated by using as ex-
systems with multiple lines of defense that degrade amples block ciphers, stream ciphers, cryptographic
gracefully. A third area where we need to make sub- hash functions and public-key encryption schemes. A

IS-22
more detailed overview of recent progress in crypt- bination generators based on Linear Feedback Shift
analysis can be found in (Preneel, 2007). Registers (LFSRs), since it turns out that these gen-
In the area of block ciphers, a major changed oc- erators are vulnerable to fast correlation attacks (see
curred when DES (a 1977 FIPS 46 standard designed e.g. (Meier and Staffelbach, 1989)) and algebraic at-
by IBM (FIPS 46, 1977)) was replaced by AES (the tacks (see e.g. (Courtois and Meier)). Stream ci-
2001 FIPS 197 standard (FIPS 197, 2001)). The main phers with clock control such as the shrinking gen-
shortcomings of DES are that its key size (56 bits) and erator (Coppersmith) and the self-shrinking genera-
block length (64 bits) are too small. M. Wiener has tor (Meier and Staffelbach) seem to be more resistant
designed in 1993 a US$ 1 million hardware DES key to cryptanalysis; recently two new designs based on
search machine that can recover a 56-bit DES key in a Non-Linear Feedback Shift Register (NLFSR) have
about three hours (Wiener, 1996). If such a machine been introduced that have a very low footprint yet of-
would be built in 2008, it would be about 600 times fer a good performance in hardware: Grain (Hell et
faster, hence it would need about 10 seconds to find al., 2005) and Trivum (De Cannière and Prennel). In
a key. The 64-bit block length is no longer adequate software, the evolution seems to be towards designs
for most applications because there exist matching ci- that contain large tables that are updated at every it-
phertext attacks on the modes of operation of an n- eration; RC4 is the pioneer of this area and recently
bit block cipher which require about 2n/2 ciphertext HC (Wu, 2005) and PY (Biham and Seberry, 2007)
blocks (Knudsen, 1994). In 1978, triple-DES was have been added to this class. An alternative approach
proposed as an alternative to DES; the two-key variant takes its inspiration from the block cipher designs
(with a 112-bit key) is believed to offer a security level (e.g., SCREAM (Halevi et al.), MUGI (Watanabe et
of 70-80 bits, while the three-key variant (with a 168- al., 2004), and Salsa (Bernstein et al.)). Between 2004
bit key) offers a security level of 90-100 bits. Triple- and 2008 the ECRYPT Network of Excellence has
DES still has a 64-bit block size. Moreover, DES is run the eSTREAM initiative with as goal to advance
not very fast in software (45 cycles/byte on a Pentium- the state of the art in stream ciphers and to identify
type processor), and triple-DES requires more than promising candidates for stream ciphers (ECRYPT).
100 cycles/byte. AES is the result of an open compe- After an open call for proposals in 2004, 34 designs
tition with as winner the Rijndael algorithm designed were submitted; during the evaluation phase more
by Daemen and Rijmen (Daemen and Rijmen, 2001). than half of these were broken. In April 2008 eight
It has a block length of 128 bits and key lengths of stream ciphers have been identified as promising can-
128, 192 and 256 bits, which means that matching ci- didates for future research and standardization (for
phertext attacks and exhaustive search are completely more details see (Robshaw and Billet, 2008)).
out of reach. Moreover, it is a very flexible cipher that In 2004, the hash function crisis started: Wang
offers a good software performance (15 cycles/byte), et al. showed that finding collisions for the widely
but also allows for a very broad range of size/speed used hash function MD5 (Rivest, 1992) requires only
trade-offs in hardware. Today AES is a trusted design a few minutes on a PC rather than 264 hash function
that is being integrated in a growing number of ap- calls; one year later they also showed that collisions
plications; Intel has even announced that they plan to for SHA-1 (FIPS 180-2, 2002) can be found in less
integrate a hardware implementation of one round of than 280 function evaluations needed for a brute force
AES in their CPUs. For some low-end applications, a collision search (currently the best result announced
compact 64-bit block cipher with an 80-bit key would by a team from T.U.Graz requires between 260 and
be useful; PRESENT (Bogdanov et al., 2007) is a re- 261 function evaluations). In parallel, several struc-
cent design with these characteristics; it can be imple- tural vulnerabilities were identified in our hash func-
mented in 1600 gates, compared to 3400 for AES. tion designs, with perhaps as most surprising observa-
The area of synchronous stream ciphers has a tion that the concatenation of the result of two iterated
very long tradition that goes back to the 1960s (e.g., hash functions is only as secure as the strongest of the
Golomb (Golomb, 1967)). Stream ciphers became two (Joux, 2004). It should be pointed out that ex-
less important after the introduction of the open perts in the area of hash functions were perhaps less
block cipher standard DES, because DES was widely surprised by the results on MD5, since serious struc-
trusted and offers a flexible building block. How- tural weaknesses had been defined in MD5 in 1993
ever, today it is believed that stream ciphers still have by den Boer and Bosselaers (den Boer and Bosse-
the potential to outperform AES with a factor of two laers) and in 1996 by Dobbertin (Dobbertin, 1996).
or more for very low footprint applications of for A second observation is that the strength of MD5 and
very high speed in software. Modern stream ciphers SHA-1 against brute force collision search is only 264
tend to move away from filter generators and com- and 280 ; this means that brute force collision attacks

IS-23
against MD5 with dedicated hardware were feasible twenty years to understand what the security prop-
in the 1990s; for SHA-1 a brute force attack will be- erties are for a public-key encryption algorithm and
come feasible in the next years. One can conclude how one should use RSA in an application in such a
from this observation that MD5 should have been re- way that breaking the security of the encryption can
placed more than 10 years ago. Many hash function be reduced to breaking the RSA assumption. The
designs have been broken; it seems that the designers most efficient construction is RSA-OAEP (Fujisaki et
are too optimistic: if MD5 would have had 8 rounds al.); note that it requires the random oracle assump-
rather than 4 or if SHA-1 would have had 160 steps tion, which is somewhat controversial (for more de-
rather than 80, nobody would have spoken of a hash tails see Koblitz and Menezes (Koblitz and Menezes,
function crisis today. Due to their high speed in soft- 2007)).
ware and the freedom in usage (no license fees and
less export issues), hash functions have become in the 2.2 Secure Implementations
1990s the Swiss army knifes of cryptography: they
are used for a broad range of applications, includ-
The classified community realized already in the
ing digital signatures, commitments, pseudo-random
1950s that in addition to secure algorithms, secure im-
function, key derivation, identification protocols, con-
plementations are needed (see for example Wright’s
struction of MAC algorithms, etc. The current cri-
book (Wright, 1987)). The open community discov-
sis has generated an increased interest in the research
ered only in the 1990s the power of side channel at-
community to understand the properties required for
tacks on software and hardware implementations of
all these applications and to develop generic construc-
cryptographic algorithms. This includes passive at-
tions. NIST has published in 2002 FIPS 180-2 (FIPS
tacks such as timing attacks (Kocher; Tsunoo et al.,
180-2, 2002); this standard includes a.o. SHA-256
2003), power analysis (Kocher et al.) and electro-
and SHA-512 and is commonly known as the SHA-2
magnetic attacks (Gandolfi et al., 2001; Quisquater
family. While currently no viable attacks are known
and Samide, 2001) and active attacks such as fault at-
on this family, the similarity with the SHA-1 de-
tacks (Boneh et al., 1997). By now it is clear that
signs has created some concern. From 2008 to 2012,
defending against these attacks is very hard, in par-
NIST organizes an open competition for the AHS
ticular since new and improved attacks are discovered
(Advanced Hash Standard) or the SHA-3 family; the
on a regular basis. A broad range of countermeasures
resulting standard will offer an alternative to the SHA-
has been developed at all layers (circuit, logic gate,
2 family.
algorithms, protocols); very often they bring a sub-
In the area of public key encryption, all popu- stantial overhead and it happens regularly that coun-
lar algorithms depend on a small set of problems termeasures are shown to be insufficient. Researchers
in algebraic number theory (factoring (Rivest et al., are also attempting to develop a theory to model side
1978), discrete logarithm in Z p (Diffie and Hellman) channel attacks in order to develop a solid foundation
and discrete logarithm on an elliptic curve over a fi- of this area. Overall, the development of implementa-
nite field (Avanzi et al., 2005)). If a breakthrough tions that are both secure and efficient is a very diffi-
would be made in solving any of these problems, or cult challenge that will require a substantial research
if quantum computers could be built (), we would effort.
have to abandon all these schemes. There exists a
small number of alternatives based on coding theory
and lattices. These alternative public-key schemes are
known today under the name ‘post quantum cryptog- 3 RESEARCH CHALLENGES
raphy’ since they could resist the existence of a large FOR CRYPTOGRAPHIC
quantum computer. Typically they require more com- PROTOCOLS
putation and/or memory than the schemes used today;
in addition, more security analysis is needed before While this article focuses on cryptographic algo-
they can be widely adopted. In the last years several rithms, there are also important challenges in cryp-
schemes based on multivariate polynomial equations tographic protocols. A first challenge has already
have been cryptanalyzed. This shows that the design been mentioned: the development of low cost entity
of novel public key encryption schemes is a challeng- authentication protocols for small devices. Substan-
ing problem. tial progress has been made in proving the security
Today RSA is still the most widely used public of simple building blocks such as authenticated en-
key encryption algorithm. It is perhaps surprising cryption or two-party authenticated key agreement,
to learn that it has taken cryptographers more than but the development and analysis of more complex

IS-24
cryptographic protocols for distributing trust is a ma- ing complex applications and the security of crypto-
jor challenge. These protocols allow to reduce the re- graphic algorithms is to a large extent based on engi-
quirement of trusting a centralized system and/or spe- neering and heuristics. For future systems, we should
cific machines or hardware components by distribut- strive to create modular designs in which building
ing this trust over a larger number of entities. This is blocks can be replaced as needed. There are also
very important for privacy sensitive applications such important challenges in the area of secure implemen-
as voting, auctions and data mining (e.g., of medical tations; progress could be made here by developing
data). The creation of complex cryptographic proto- models and methods to verify the security and cor-
cols has been based on an approach for provable se- rectness.
curity based on assumptions for cryptographic algo- Finally, cryptography forms a building block of a
rithms; due to the complexity of the proofs and meth- larger security system. It is therefore important that
ods, there is a need for advanced tools to develop and we try to improve our understanding of the overall
verify such proofs. A second goal is to take into ac- system requirements in order to ensure that the cryp-
count more realistic deployment models such as con- tographic solutions are as useful and usable as possi-
current composition, asynchronous interactions, com- ble.
plex privacy models and protocols in which the play-
ers are not “malicious” or “honest” but rather “ratio-
nal”; the latter approach results in a novel game theo-
retic approach. ACKNOWLEDGEMENTS
A more practical problem is to design crypto-
graphic protocols in such a way that algorithms or This article is based in part on the results of the
building blocks can be upgraded easily. Section 2 ECRYPT project (ECRYPT, 2006). This author’s
shows that the lifetimes of cryptographic algorithms work was supported in part by the IAP Programme
are frequently shorted than anticipated. During the P6/26 BCRYPT of the Belgian State (Belgian Science
design of the SSL/TLS protocol, this was well under- Policy) and by the European Commission through
stood, yet it turns out that MD5 and SHA-1 are part of the IST Programme under Contract IST-2002-507932
the SSL negotiation process, and cannot be upgraded ECRYPT.
easily.

REFERENCES
4 CONCLUSIONS
R.J. Anderson “Why cryptosystems fail,” Communications
In this article we have not discussed the fundamental ACM, Vol. 37, No. 11, November 1994, pp. 32–40.
research questions in cryptology, that are also funda- Avanzi, R.M., Cohen, H., Doche, C., Frey, G., Lange, T.,
mental questions in computer science or mathematics, Nguyen, K., Vercauteren, F. Handbook of Elliptic and
such as: can we prove that certain problems are hard, Hyperelliptic Curve Cryptography, H. Cohen, G. Frey,
is it hard to factor the product of two large primes, or Eds., Chapman & Hall/CRC, 2005.
how difficult is it to solve large non-linear systems of Bernstein, D.J., Salsa20 Specification, Available via:
equations, even if these have a regular structure (e.g., http://www.ecrypt.eu.org/stream/salsa20p3.html.
because they have been deduced from a block cipher). Biham, E., Seberry, J. Tweaking the IV Setup
All of these problems are known to be very difficult, of the Py Family of Stream Ciphers—The Ci-
and researchers understand that so far we have only phers TPy, TPypy, and TPy6, Paper 2007/038 at
scratched the surface in this area, which forms the http://www.ecrypt.eu.org/stream/papers.html.
foundation of cryptography. Bogdanov, A., Knudsen, L.R., Leander, G., Paar, C.,
The goal of cryptography is to develop clear se- Poschmann, A., Robshaw, M.J.B., Seurin,Y., Vikkel-
curity definitions and develop an understanding of the soe, C. “PRESENT: An ultra-lightweight block ci-
relation between the security properties of the appli- pher,” Cryptographic Hardware and Embedded Sys-
tems – CHES 2007, Lecture Notes in Computer Sci-
cations and those of the building blocks: algorithms ence 4727, P. Paillier, I. Verbauwhede, Eds., Springer-
(e.g., AES), primitives (e.g., the OCB mode based on Verlag, 2007, pp. 450–466.
AES, RSA-OAEP), protocols (e.g., an e-voting pro-
Boneh, D., DeMillo, R., Lipton, R. “On the importance
tocol) and applications. So far we seem to be mak-
of checking cryptographic protocols for faults,” Ad-
ing progress to prove relationships between proto- vances in Cryptology, Proceedings Eurocrypt’97, Lec-
cols and primitives and between primitives and algo- ture Notes in Computer Science 1233, W. Fumy, Ed.,
rithms. However, we are still very far from analyz- Springer-Verlag, 1997, pp. 37–51.

IS-25
Courtois, N., Meier, W. “Algebraic attacks on stream ci- Halevi, S., Coppersmith, D., Jutla, C. S. “Scream: A
phers with linear feedback,” Advances in Cry ptol- software-efficient stream cipher,” Fast Software En-
ogy, Proc. Eurocrypt’03, Lecture Notes in Computer cryption, Lecture Notes in Computer Science 2365,
Science 2656, E. Biham, Ed., Springer-Verlag, 2003, J. Daemen, V. Rijmen, Eds., Springer-Verlag, 2002,
pp. 345–359. pp. 195–209.
Coppersmith, D., Krawczyk, H., Mansour, Y. “The Hell, M., Johansson, T., Meier, W. Grain – A Stream Ci-
shrinking generator,” Advances in Cryptology, Proc. pher for Constrained Environments, Paper 2005/020
Crypto’03, Lecture Notes in Computer Science 2729, at http://www.ecrypt.eu.org/stream/grainp3.html.
D. Boneh, Ed., Springer-Verlag, 2003, pp. 22–39.
Joux,A. “Multicollisions in iterated hash functions. Appli-
Daemen, J., Rijmen, V. The Design of Rijndael. AES – cation to cascaded constructions,” Advances in Cryp-
The Advanced Encryption Standard, Springer-Verlag, tology, Proceedings Crypto’04, Lecture Notes in Com-
2001. puter Science 3152, M.K. Franklin, Ed., Springer-
De Cannière, Preneel, C. B. Triv- Verlag, 2004, pp. 306–316.
ium Specifications, Available via Knudsen, L. R. Block Ciphers – Analysis, Design and Ap-
http://www.ecrypt.eu.org/stream/triviump3.html. plications, PhD thesis, Aarhus University, Denmark,
den Boer,B., Bosselaers, A. “Collisions for the compression 1994.
function of MD5,” Advances in Cryptology, Proceed- Koblitz, N., Menezes, A. “Another look at “provable secu-
ings Eurocrypt’93, Lecture Notes in Computer Sci- rity”,” Journal of Cryptology, Vol. 20, 2007, pp. 3–37.
ence 765, T. Helleseth, Ed., Springer-Verlag, 1994,
pp. 293–304. Kocher, P. “Timing attacks on implementations of Diffie-
Hellman, RSA, DSS, and other systems,” Advances
Diffie, W., Hellman, M.E. “New directions in cryptog-
in Cryptology, Proceedings Crypto’96, Lecture Notes
raphy,” IEEE Transactions on Information Theory,
in Computer Science 1109, N. Koblitz, Ed., Springer-
Vol. IT–22, No. 6, 1976, pp. 644–654.
Verlag, 1996, pp. 104–113.
Dobbertin, H. “The status of MD5 after a recent attack,”
CryptoBytes, Vol. 2, No. 2, Summer 1996, pp. 1–6. Kocher, P., Jaffe, J., Jun, B. “Differential power analysis,”
Advances in Cryptology, Proceedings Crypto’99, Lec-
ECRYPT Deliverable D.SPA.22, Challenges for Cryptol- ture Notes in Computer Science 1666, M. Wiener, Ed.,
ogy Research in Europe for 2007-2013, revision 1.0, Springer-Verlag, 1999, pp. 388–397.
26 May 2006, http://www.ecrypt.eu.org.
Meier, W., Staffelbach, O. “Fast correlation attacks on
ECRYPT eSTREAM initia- stream ciphers,” Journal of Cryptology, Vol. 1, 1989,
tive,http://www.ecrypt.eu.org/estream pp. 159–176.
EMV, Integrated Circuit Card Specifications for Payment Meier, W., Staffelbach, O. “The self-shrinking generator,”
Systems: Book 2 Security and Key Management, Ver- Advances in Cryptology, Proc. Eurocrypt’94, Lecture
sion 4.1, May 2004. Notes in Computer Science 950, A. De Santis, Ed.,
FIPS 46, Data Encryption Standard, Federal Information Springer-Verlag, 1995, pp. 205–214.
Processing Standard, NBS, U.S. Dept. of Commerce,
Menezes, A. J., van Oorschot, P. C., Vanstone, S.A. Hand-
January 1977 (revised as FIPS 46-1, 1988; FIPS 46-2,
book of Applied Cryptography, CRC Press, 1997.
1993, FIPS 46-3, 1999).
FIPS 180-2, Secure Hash Standard, Federal Information Preneel, B. “A survey of recent developments in crypto-
Processing Standard (FIPS), Publication 180-2, NIST graphic algorithms for smart cards,” Computer Net-
U.S. Dept. of Commerce, August 26, 2002 (Change works, Vol. 51, No. 9, 2007, pp. 2223-2233.
notice 1 published on December 1, 2003). Quisquater, J.-J., Samide, D. “ElectroMagnetic Analy-
FIPS 197, Advanced Encryption Standard, Federal In- sis (EMA): Measures and countermeasures for smart
formation Processing Standard, NIST, U.S. Dept. of cards,” Smart Card Programming and Security, Inter-
Commerce, November 26, 2001. national Conference on Research in Smart Cards, E-
smart 2001, Lecture Notes in Computer Science 2140,
Fujisaki, E., Okamoto, T., Pointcheval, D., Stern, J. RSA- I. Attali, T. Jensen, Eds., Springer-Verlag, 2001,
OAEP is secure under the RSA assumption, Advances pp. 200–210.
in Cryptology, Proceedings Crypto’01, Lecture Notes
in Computer Science 2139, J. Kilian, Ed., Springer- Rivest, R. L., Shamir, A., Adleman, L. “A method for
Verlag, 2001, pp. 260–274. obtaining digital signatures and public-key cryptosys-
tems,” Communications ACM, Vol. 21, No. 2, 1978,
Gandolfi, K., Mourtel, C., Olivier, F. “Electromag- pp. 120–126.
netic analysis: Concrete results,” Proceedings Cryp-
tographic Hardware and Embedded Systems – CHES Rivest, R. L. “The MD5 message-digest algorithm,” Re-
2001, Lecture Notes in Computer Science 2162, quest for Comments (RFC) 1321, Internet Activities
C.K. Koç, D. Naccache, C. Paar, Eds., Springer- Board, Internet Privacy Task Force, April 1992.
Verlag, 2001, pp. 251–261.
Robshaw, M. J. B., Billet, O., Eds. New Stream Ci-
Golomb, S. W. Shift Register Sequences, Holden-Day, pher Designs: The eSTREAM Finalists, LNCS 4986,
1967. Springer-Verlag, 2008.

IS-26
Shor, P. W. “Algorithms for quantum computation: dis- in the area of e-security. He is cofounder and conduc-
crete logarithms and factoring,” Proceedings 35nd An- tor of the jazz ensemble of the K.U.Leuven.
nual Symposium on Foundations of Computer Sci-
ence, S. Goldwasser, Ed., IEEE Computer Society
Press, 1994, pp. 124–134.
Tsunoo, Y., Saito, T., Suzaki, T., Shigery, M., Miyauchi,
H. “Cryptanalysis of DES implemented on computers
with cache,” Cryptographic Hardware and Embedded
Systems – CHES 2003, Lecture Notes in Computer
Science 2779, C.D. Walter, Ç.K. Koç, C. Paar, Eds.,
Springer-Verlag, 2003, pp. 62–76.
Wang, X., Yu, H. “How to break MD5 and other hash
functions,” Advances in Cryptology, Proceedings Eu-
rocrypt’05, Lecture Notes in Computer Science 3494,
R. Cramer, Ed., Springer-Verlag, 2005, pp. 19–35.
Wang, X., Lin, Y. L., Yu, H. “Finding collisions in the
ful SHA-1,” Advances in Cryptology, Proceedings
Crypto’05, Lecture Notes in Computer Science 3621,
V. Shoup, Ed., Springer-Verlag, 2005, pp. 17–36.
Watanabe, D., Furuya, S., Yoshida, H., Preneel, B. “A
new keystream generator MUGI,” IEICE Trans. Fun-
damentals, Vol. E87-A, No. 1, 2004, pp. 37-45.
Wiener, M. J. “Efficient DES key search,” Presented at the
Rump Session of Crypto’93. Reprinted in Practical
Cryptography for Data Internetworks, W. Stallings
(ed.), pp. 31–79, IEEE Computer Society, 1996.
Wright, P. (with P. Greengrass) “Spycatcher. The Can-
did Autobiography of a Senior Intelligence Officer,”
Heinemann (Australia), 1987.
Wu, H. The Stream Cipher HC-128, Paper 2005/020 at
http://www.ecrypt.eu.org/stream/hcp3.html.

BRIEF BIOGRAPHY
Bart Preneel received the Doctorate in Applied Sci-
ences from the Katholieke Universiteit Leuven (Bel-
gium) where he is currently a full professor. He was
visiting professor at several universities in Europe.
His main research interests are cryptography and in-
formation security. He has authored and co-authored
more than 200 scientific publications. He is president
of the IACR (International Association for Crypto-
logic Research) and a member of the Editorial Board
of the Journal of Cryptology and of the IEEE Trans-
actions on Forensics and Information Security. He
has participated to 25 research projects sponsored by
the European Commission, for five of these as project
manager. He has been program chair of ten interna-
tional conferences and he has been invited speaker at
more than 30 conferences. In 2003, he has received
the European Information Security Award in the area
of academic research. He has been a member of the
TCPA Advisory Board. He is president of L-SEC
vzw. (Leuven Security Excellence Consortium), an
association of 60 companies and research institutions

IS-27
InvisibleWhiteText.
 WATERMARKING, STEGANOGRAPHY
AND CONTENT FORENSICS

Ingemar J. Cox
University College London, U.K.

Abstract: Electronic watermarking is about 60 years old. However it was not until the beginning of the early 1990's
that watermarking received widespread interest, due to concerns about piracy of digital content. In the
subsequent decade, very significant progress has been made both in our theoretical understanding of digital
watermarking and in its applications. This progress is described here.
Steganography has a much longer history, dating back to at least the time of the ancient Greeks. While
Shannon dismissed steganography as "{primarily a psychological problem}", the last decade has seen the
application of information theory to steganography. Terrorist events at the beginning of the 21st century
motivated further attention and very interesting results have been described. At first sight, digital
watermarking and steganography would appear to share the same goals. However, while both seek to hide
information within other information or content, there are very significant differences in the constraints that
must be satisfied. The similarities and differences between digital watermarking and steganography are
highlighted here.
Steganography spawns steganalysis, the art and science of detecting the presence of a steganographic
message hidden in innocuous content. Recent research views steganalysis as a binary classification problem;
is a hidden message present or absent? Classification is based on testing for statistical anomalies in features
derived from the content. Content forensics shares similarities with steganalysis. At the simplest level,
content forensics is often asked whether, for example, an image is authentic or has been tampered with.
This problem can also be viewed as a binary classification problem and similar techniques can be applied.
We will review recent work in content forensics and steganalysis, and discuss the limitations of both.

BRIEF BIOGRAPHY Signal Processing Area) for a paper he co-authored

on watermarking. From 1997-1999, he served as
Ingemar J. Cox is currently Professor and BT Chair Chief Technical Officer of Signafy, Inc, a subsidiary
of Communications in the Departments of Computer of NEC responsible for the commercialization of
Science, and Electronic and Electrical Engineering watermarking. Between 1996 and 1999, he led the
at University College London and Director of UCL's design of NEC's watermarking proposal for DVD
Adastral Park Postgraduate Campus. He is currently video disks and later colloborated with IBM in
a holder of a Royal Society Wolfson Fellowship. He developing the technology behind the joint "Galaxy"
received his B.Sc. from University College London proposal supported by Hitachi, IBM, NEC, Pioneer
and Ph.D. from Oxford University. He was a and Sony. In 1999, he returned to NEC Research
member of the Technical Staff at AT\&T Bell Labs Institute as a Research Fellow. He is a Fellow of the
at Murray Hill from 1984 until 1989 where his IEEE, the IET (formerly IEE), and the British
research interests were focused on mobile robots. In Computer Society. He is a member of the UK
1989 he joined NEC Research Institute in Princeton, Computing Research Committee. He was founding
NJ as a senior research scientist in the computer co-editor in chief of the IEE Proc. on Information
science division. At NEC, his research shifted to Security and is an associate editor of the IEEE
problems in computer vision and he was responsible Trans. on Information Forensics and Security. He is
for creating the computer vision group at NECI. He co-author of a book entitled "Digital Watermarking"
has worked on problems to do with stereo and and its second edition "Digital Watermarking and
motion correspondence and multimedia issues of Steganography", and the co-editor of two books,
image database retrieval and watermarking. In 1999, `Autonomous Robots Vehicles' and `Partitioning
he was awarded the IEEE Signal Processing Society Data Sets: With Applications to Psychology,
Best Paper Award (Image and Multidimensional Computer Vision and Target Tracking'.

IS-29
 RETHINKING DIGITAL SIGNATURES

Ueli Maurer
Department of Computer Science, ETH Zurich, CH-8092 Zurich, Switzerland
[email protected]

Keywords: Digital signatures, digital evidence, non-repudiation, public-key certificates, public-key infrastructure (PKI),
revocation, time-stamping.

Abstract: Digital signatures are a core enabling technology for evidence management in the context of the automation
and digitization of business and government processes. The main advantages over conventional signatures are
that they are easy to transmit, archive, search, and verify, and, moreover, promise to achieve a high level of
security. Nevertheless the initial expectations for the usefulness of digital signatures were too optimistic. This
calls for a systematic treatment of digital signatures and more generally digital evidence. The goal of this talk
is to provide a foundation for reasoning about digital evidence systems and legislation, thereby identifying the
roles and limitations of digital evidence.

1 INTRODUCTION ally irrelevant when a signature was generated, that

a commitment to be liable for digital evidence cannot
This short paper accompanies the author’s talk at meaningfully be revoked, and that there is no need for
ICETE 2008, which is based on and extends the ar- mutually trusted authorities like certification authori-
ticle (Maurer, 2004). This short note only summa- ties.
rizes a few points at a non-technical level. Interested We also propose a new type of digital evidence
readers should consult (Maurer, 2004) and possibly called digital declarations, based on a digital record-
also (Maurer, 2003). ing of a willful act indicating agreement to a docu-
ment or contract.
This work draws its motivation from the appar-
ent discrepancy between the potential usefulness of
digital signatures as evidence and the fact that their
use in practical applications is still very limited. We 2 EVIDENCE FOR
believe that a primary reason for this discrepancy is NON-REPUDIATION
the lack of understanding of the semantics of digi-
tal evidence like digital signatures, certificates, time- A basic act in business and other contexts is to enter
stamps, etc. Our goal is therefore to provide a foun- a formal agreement, often called a contract, between
dation for reasoning about digital evidence systems two (or more) entities. Such an agreement requires
and legislation and to identify the roles and limitations the clear mutual understanding of all relevant param-
of digital evidence, in the apparently simple scenario eters, in particular the terms and conditions. A con-
where it should prove that an entity A agreed to a dig- tract is valid only if both parties formally entered it.
ital contract d. It is generally understood that a contract has been en-
The approach proposed in (Maurer, 2004) is in tered by a user or entity only if he or she (or an autho-
sharp contrast to the current general views docu- rized representative) performed a well-defined con-
mented in the technical literature and in digital sig- scious and willful act, for instance by shaking hands
nature legislation. We propose an entirely new view and/or by signing a paper document, or by activating
of the concepts of certification, time-stamping, revo- the generation of a digital signature on his smart-card.
cation, and other trusted services, potentially leading In order to prepare for a possible future dispute,
to new and more sound business models for trusted each party to a contract wants to keep sufficient evi-
services. Some of the perhaps provocative implica- dence for the claim that the other party agreed to the
tions of our view are that certificates are generally contract. This is a symmetric goal, although the evi-
irrelevant as evidence in a dispute, that it is gener- dence collected by each party may be different. How

IS-31
can an entity B obtain sufficient evidence that party A resulting in slow user acceptance.
entered the contract? What sufficient means must be An even more fundamental problem is that the
defined by the legal system. meaning of digital evidence is not well understood.
One can distinguish at least three types of evi- The intrinsic problem that a digital signature is not
dence, a combination of which may be used in a con- linked to any event in the real world has no solution.
crete setting: (1) physical evidence (e.g. a signed pa- It is inherently impossible to determine when, where,
per document), (2) statements by witnesses, and (3) how, and by whom a digital signature was generated,
digital evidence (digital signatures, time stamps, etc.). even when secure hardware, biometric identification,
Physical evidence and witnesses seem to contradict and time-stamping is used.
the new paradigm of digitized business processes. A
wide-spread expectation for digital signatures is that
they allow to avoid the need for physical evidence and 4 CONVENTIONAL VS. DIGITAL
witnesses.
SIGNATURES
As a motivation for digital declarations, it is instruc-
3 DIGITAL SIGNATURES: tive to discuss the role of conventional hand-written
PROMISES AND OBSTACLES signatures and why they are so useful in practice, de-
spite the fact that their technical security is generally
A digital signature scheme works as follows. A user quite low. Conventional signatures are a pragmatic
A generates a secret key, which she keeps secret, and and flexible mechanism.
the corresponding public key, which is made publicly An idealized (but of course naı̈ve) view of the use
available. The digital signature for a digital contract of conventional signatures can be described as fol-
can be generated only when given the secret key, but lows. A user’s signature is well-defined, for instance
it can be verified by anybody using the public key. by a master copy she has deposited. In case of a dis-
Hence it can be interpreted as a proof that A agreed to pute, a signature allegedly issued by the user can be
the contract. compared to her master signature. If one assumes that
In view of the continuing automation and digiti- forged signatures can be recognized, then a signature
zation of many business processes, the transmission, is convincing evidence for the user’s consent to the
storage, and verification of physical evidence, like signed document.
signed contracts, presents a major problem. In con- In practice, however, things are quite different
trast to physical evidence, digital signatures are easy from this idealized view. First, most people’s hand-
to transmit, archive, search, and verify. Moreover, written signatures are not very difficult to forge for
digital signatures are generally unambiguous because a dedicated forger. Second, in most settings (except
their verification corresponds simply to the evaluation for example in a bilateral business relationship with a
of a well-defined mathematical function, the signature bank), a person’s master signature is neither deposited
verification predicate relative to a given public key. nor defined. Third, a person could use a signature
For these reasons, digital signatures promise to pro- different from the master copy in order to be able to
vide an elegant solution to the non-repudiation prob- later repudiate it. To avoid this problem, the receiving
lem in the digitally operating economy. party would have to have on-line access to a master
Furthermore, due to the conjectured security of signature registry.
the underlying cryptographic mechanisms, digital sig- The value and purpose of a hand-written signa-
natures also promise substantially higher security ture is not primarily that it is difficult to forge, but
compared to conventional signatures, and hence fewer rather that it creates a situation in which a person A
disputes and simpler dispute resolution. knows whether or not she signed, thus guaranteeing
Despite the promises, in the context of non- her awareness of performing a conscious and will-
repudiation services digital signatures are currently ful act. Similarly, forging a signature also requires a
used only in isolated applications. We are still far conscious act. Due to this guaranteed awareness, the
from an internationally operational framework and in- denial of having signed a document is a precise and
frastructure. Some of the obstacles are the lack of in- meaningful claim, equivalent to the (serious) claim
ternationally applicable laws, the lack of standardiza- that the signature is forged.
tion, the lack of viable business models for fostering The described view on conventional signatures
the creation of a global PKI, problems with the inte- is in sharp contrast to what digital signatures can
gration into business processes, and, last but not least, achieve. The existence of a digital signature does not
the abstractness and complexity of the subject matter, imply the guaranteed awareness of the alleged signer

IS-32
of the act that caused the signature generation. A sig- • lower cost due to reduced security requirements,
nature could have been computed by a virus, because • improved acceptance of digital signature technol-
of another security problem, a flaw or ambiguity in ogy, and
the user interface, a flaw in the cryptographic mech-
anism, fraud or errors in the certification process, or • usability by moderately educated people.
any other of many possible reasons. Therefore a user
A cannot meaningfully deny that a signature was gen-
erated by her. Rather, a denial is equivalent to the (for REFERENCES
a judge quite useless) claim that she is not aware of Maurer, U. (2003). Intrinsic limitations of digital signatures
having issued a signature. and how to cope with them. In Boyd, C. and Mao, W.,
editors, Proceedings of the 6th Information Security
Conference (ISC’03), volume 2851 of Lecture Notes
5 DIGITAL DECLARATIONS in Computer Science, pages 180–192. Springer Ver-
lag.
Maurer, U. (2004). New approaches to digital evidence.
The dilemma can be described as follows. On one Proceedings of the IEEE, 92(6):933–947.
hand, the evidence collected to prove the execution of
a contract should be digital in order to fit smoothly
into the digitized business processes, but on the other BRIEF BIOGRAPHY
hand it should be linked to the physical reality, in par-
ticular to a user’s willful act when agreeing to d. This Ueli Maurer is professor of computer science and
appears to be a contradiction, but there exists a so- head of the Information Security and Cryptography
lution proposed in (Maurer, 2003) and called digital Research Group at the Swiss Federal Institute of
declarations. Technology (ETH), Zurich. His research interests
The user performs some willful act related to the include information security, theory of cryptogra-
relevant contract or document, and this act is recorded phy (new paradigms, security proofs), applications of
digitally and combined with characteristic informa- cryptography (e.g. digital signatures, public-key in-
tion of the digital document. In a typical implemen- frastructures, digital payment systems, e-voting), the-
tation, the digital declaration can be signed together oretical computer science, discrete mathematics, and
with the actual digital document. information theory. He is also interested in the im-
Digital declarations are purely digital and are pact of IT on the society and economy at large. Mau-
hence easy to transport and store. But they offer the rer graduated in electrical engineering (1985) and re-
same functionality as conventional hand-written sig- ceived his Ph.D. degree in Technical Sciences (1990)
natures. In particular, a user is guaranteed to be aware from ETH Zurich. From 1990 to 1991 he was DI-
of what he is doing and can hence meaningfully be MACS research fellow at the Department of Com-
forced to deny a digital declaration. Such a denial is puter Science at Princeton University, and in 1992 he
equivalent to the accusation that the digital declara- joined the CS Department at ETH Zurich where he
tion is forged. is a full professor. He has served extensively as an
Digital declarations can be embodied in many dif- editor and a member of program committees. Cur-
ferent ways. As an example, a user ordering a service rently he is Editor-in-Chief of the Journal of Cryp-
or product on-line might be asked to speak a certain tology, Editor-in-Chief of Springer Verlag’s book se-
sentence referring to the product, the price, and the ries in Information Security and Cryptography, and
date of purchase. As another possibility, the willful serves on the Board of Directors of the International
act could be documented by a digital image, a video, Association for Cryptologic Research (IACR). He is
or by any other recording device possibly invented in a Fellow of the IEEE, a Fellow of the IACR, and was
the future. the 2000 Rademacher Lecturer of the Department of
Digital declarations can be a useful feature of fu- Mathematics at the University of Pennsylvania. Mau-
ture digital transaction systems, for the following rea- rer has served as a consultant for many companies and
sons: government organisations, both at the management
• guaranteed user awareness, and the technical level. He serves on a few boards,
including the board of directors of Tamedia, a Swiss
• higher deterrence of misbehavior, hence fewer media company, and the scientific advisory board of
disputes, PricewaterhouseCoopers. He is a co-founder of the
• improved security compared to conventional sig- Zurich-based security software company Seclutions
natures, and holds several patents for cryptographic systems.

IS-33
InvisibleWhiteText.
 THE IMPORTANCE OF METROLOGY IN WIRELESS
COMMUNICATION SYSTEMS
From AM/FM to SDR Systems

Nuno Borges Carvalho

Instituto de Telecomunicacoes / Universidade de Aveiro, Portugal

Abstract: Metrology is a key point in the design and optimization of wireless communication networks, in this talk we
will make a brief presentation of the metrology chain necessities from base-band to RF wireless
communication scenarios. In this path we will address linear and nonlinear measurements strategies, from
RF figures of merit to base band figures. Finally some recent developments referred to software radio
activities will also be presented, explaining the new paradigms and envisioning the future of wireless
metrology.

BRIEF BIOGRAPHY 2000 IEE Measurement Prize. He is also the co-

inventor of three registered national patents. He is a
Nuno Borges Carvalho, was born in Luanda in 1972. reviewer for several magazines including the IEEE
He received the diploma and doctoral degrees in Transactions on Microwave Theory and Techniques,
Electronics and Telecommunications Engineering IEEE Transactions on Cicuits and Systems, IEEE
from the Universidade de Aveiro, Aveiro, Portugal Transactions on Communications and IEEE
in 1995 and 2000 respectively. From 1997 to 2000 Transactions on Intelligent Transportion Systems, as
he was an Assistant Lecturer at the same University some International journals from the Institute of
and a Professor since 2000. Currently he is an Physics. He is the webmaster and an active member
Associate Professor at the same University, and the of the IEEE MTT-11 Technical Committee, and the
scientific area coordinator of the wireless chair of the URSI-Portugal metrology committee.
communications at Instituto de Telecomunicações. Dr. Borges Carvalho is co-author of the book
He has worked as a scientist researcher at the “Intermodulation in Microwave and Wireless
Instituto de Telecomunicações, and was engaged to Circuits” from Artech House, 2003.
different projects on nonlinear CAD, circuits and
systems design and RF system integration. His main
research interests include CAD for nonlinear
circuits/systems, nonlinear distortion analysis in
microwave/wireless circuits and systems and
measurement of nonlinear phenomena, recently he
has also been involved in design of dedicated radios
and systems for newly emerging wireless
technologies. In 2007 he was a visitor researcher at
the North Carolina State University, and at the
National Institute of Standards and Technology,
NIST. Prof. Borges Carvalho is a member of the
Portuguese Engineering Association and an IEEE
Senior Member. He was the recipient of the 1995
University of Aveiro and the Portuguese
Engineering Association Prize for the best 1995
student at the Universidade de Aveiro, the 1998
Student Paper Competition (third place) presented at
the IEEE International Microwave Symposium, the

IS-35
 NEXT GENERATION CDMA TECHNOLOGIES FOR
FUTURISTIC WIRELESS COMMUNICATIONS

Hsiao-Hwa Chen
National Sun Yat-Sen University, Taiwan

Abstract: Future wireless communication systems should be operating mainly, if not completely, for burst data
services carrying multimedia traffics. The need to support high-speed burst traffic has already posed a great
challenge to all currently available air-link technologies based on either TDMA or CDMA. The current
CDMA technology has been widely used in both 2G and 3G mobile cellular standards and it has been
suggested that it is not suitable for high-speed burst-type traffic. There are many problems with current
CDMA technology, such as its low spreading efficiency, interference-limited capacity and the need for
precision power control, etc. This talk will address various important issues about the next generation
CDMA technologies as a major air-link technology for futuristic wireless applications. In particular, it will
cover two major topics: why we need the next generation CDMA technologies, and what is the next
generation CDMA technology.

BRIEF BIOGRAPHY Wiley’ Security and Communication Networks

journal
Hsiao-Hwa Chen is currently a full Professor and (www.interscience.wiley.com/journal/security). He
was the founding Director of the Institute of is also an adjunct Professor of Zhejiang University,
Communications Engineering of the National Sun China, and Shanghai Jiao Tung University, China.
Yat-Sen University, Taiwan. He received BSc and
MSc degrees from Zhejiang University, China, and
PhD degree from University of Oulu, Finland, in
1982, 1985 and 1990, respectively, all in Electrical
Engineering. He has authored or co-authored over
200 technical papers in major international journals
and conferences, five books and several book
chapters in the areas of communications, including
the books titled "Next Generation Wireless Systems
and Networks" (512 pages) and “The Next
Generation CDMA Technologies” (468 pages), both
published by John Wiley and Sons in 2005 and
2007, respectively. He has been an active volunteer
for IEEE various technical activities for over 20
years. Currently, he is serving as the Chair of IEEE
Communications Society Radio Communications
Committee. He served or is serving as symposium
chair/co-chair of many major IEEE conferences,
including VTC, ICC, Globecom and WCNC, etc. He
served or is serving as Associate Editor or/and Guest
Editor of numerous important technical journals in
communications. He is serving as the Editor (Asia
and Pacific) for Wiley's Wireless Communications
and Mobile Computing (WCMC) Journal and
Wiley's International Journal of Communication
Systems, etc. He is the founding Editor-in-Chief of

IS-37
TUTORIAL
 PATENTABILITY OF E-BUSINESS
AND COMPUTER-IMPLEMENTED INVENTIONS AT THE EPO

Falk Giemsa
European Patent Office, Germany

Abstract: The majority of patent applications in the field of e-Business are refused at the European Patent Office
(EPO). The reasons for the high number of refusals can be found in the European Patent Convention (EPC),
in the decisions of the Boards of Appeal of the EPO and in the patent applications.
This tutorial provides an overview of the general patent procedure before the EPO and explains the current
work practice at the EPO for examining patent applications in the field of e-Business and Computer
Implemented Inventions (CII).

BRIEF BIOGRAPHY
Falk Giemsa is a patent examiner at the European
Patent Office in Munich. He studied computer
sciences and now works in the field of data
processing systems or methods, specially adapted for
administrative, commercial, financial, managerial,
supervisory or forecasting purposes and in the field
of educational or demonstration appliances.

IS-41
 COMMUNICATION AND SOFTWARE
TECHNOLOGIES AND ARCHITECTURES
FULL PAPERS
 A PERFORMANCE EVALUATION OF AN ULTRA-THIN
CLIENT SYSTEM

Colin Pattinson and Tahir Siddiqui

Innovation North, Faculty of Information and Engineering Systems, Leeds Metropolitan University, Leeds LS6 3QS, U.K.
[email protected], [email protected]

Keywords: Low-energy networking, ultra thin client, simulation.

Abstract: Power consumption of individual devices is often ignored in the development of computer networks. The
traditional approach to a local areas network, such as might be deployed in a cyber-café or classroom
environment, has a number of workstations attached to a server, where the workstations have very similar
specifications (and hence power demands) to the server. Intuitively, this means there is significant over
specification of the workstation, and that much of the capability of the device is not used, with
consequential wasted energy. Alternative topologies exist, particularly those which make use of thin client
technology, and are meeting with success particularly in developing countries, where cost and power
consumption take on a much greater importance. One such design is the ndiyo project, which is delivering
thin-client based network solutions to a number of nations, allowing IT support to be deployed in places
where it would not have otherwise been achievable. In this paper, we report our work in developing a
simulation model to allow us to study the behaviour and operation of ndiyo, offering us the opportunity to
carry out some “what if” analyses of the behaviour of such systems under differing network loads.

1 INTRODUCTION technology, taking advantage of the fact that most of

the processing can be carried out by the server, so
The typical workstation / server approach to the the client’s power can be reduced accordingly.
provision of networked IT, in which a number of
“workstations”, each of which is actually a stand-
alone computer system, are networked to a “server” 2 THIN CLIENT NETWORKS
which acts as a central repository of files and as a
gateway to the wider network (the Internet) is very 2.1 Introduction
well established. However, this design, whatever its
advantages, appears inefficient from the perspective Thin client Networks have a long history; the
of resource utilisation. In particular, in situations development of terminal servers is one area which
such as the typical cyber café, and also in classroom has a long association with thin client networks .
situations where web browsing is the prevalent A Thin client (sometimes also called a lean
activity, workstations are effectively little more than client) is that part of client-server architecture
smart terminals. The use as workstations of devices networks which depends primarily on the central
similar in power to the server, suggests that server for processing activities, and mainly focuses
significant resource capacity may not be being used. on conveying input and output between the user and
While this may be acceptable from the perspective the remote server. In contrast, a thick or fat client
of resource use, since the machines are relatively does as much processing as possible and passes only
easily available; from the viewpoint of energy use, data for communications and storage to the server.
different arguments can be made. In situations where Many thin client devices run only web browsers
energy and other resources are at a premium, it or remote desktop software, meaning that all
makes sense to seek out other mechanisms to deliver significant processing occurs on the server.
the required processing power. However, recent devices marketed as thin clients
One alternative to the widespread deployment of can run complete operating systems such as Debian
full-specification systems uses thin client

5
ICE-B 2008 - International Conference on e-Business

GNU/Linux, qualifying them as diskless nodes or Ethernet networks (100Mbit or higher) are fast
hybrid clients (Wikipedia, 2007). The emphasis of enough for this approach to keep the user's screen up
this paper is on such thin client networks, In fact the to date. Nivo simply has an Ethernet (network)
devices discussed and tested in this paper are socket, a low voltage power socket, sockets for
referred as Ultra-thin client devices. (Nidyo, 2006) keyboard and mouse, and a VGA (monitor) socket.
This box replaces the entire PC in a conventional
2.2 Related Work computer workstation (Ndiyo, 2006).

Previous work in the field has addressed a variety of 3.2 The Network
topics which have direct impact upon the operation
of thin client systems. Kelly (2002) reports on the We were given the task to test the performance of a
need to gather appropriate behavioural data to use as nivo network with between 2 to 30 nivo devices
input into a simulation system. Tolia et. al. (2006) attached to a single server. Below are the major
make reference to the fact that “adequacy of thin- aspects of the nivo device’s network topology and
client computing is highly variable and depends on their operations:
both the application and the available network • The Nivo network generally runs on a switched
quality”. The increasing deployment of wireless and 100 Mb/s LAN, as Nivo devices usually
hand-held (battery powered) devices has led to supports 100 Mb/s.
consideration of whether thin-client technology is
appropriate for such technologies, and • In Nivo networks the transmitting packets have
measurements have been conducted by Yang et. al. areas of pixels, compressed in a lossless way,
(2003) reporting that such systems can operate transmitted over a simple transport layer which
successfully even with the relatively high packet loss provides very basic reliability - much simpler
rates which can be experienced with wireless than TCP. An approach of VNC transmitted
networks, and by Lai et. al, (2004) reporting lower over UDP is best approximate for this situation.
bandwidth requirements and hence better user • An update sent out from the server can be as
experience for thin client systems. large as a whole screen refresh – perhaps 5MB –
In light of these and other papers, we determined to or as small as a few bytes. The big updates will
conduct a simulation-based experiment using a be split into chunks of approximately1500 bytes
particular thin-client implementation, in part to when carried in an Ethernet UDP packet.
explore the performance of such a thin-client Keystroke and mouse events coming back are
system, but also to determine the adequacy of usually small, mouse event reports are around
simulation methods in this application. 10 bytes and reported at most about 50 times per
second.

3 NDIYO PROJECT 3.3 Prerequisite

This paper is focused on Nivo devices which are The major focus of this study was to predict the real
ultra thin client devices developed for the Ndiyo network bandwidth usage or at least some
Project by Displaylink. Ndiyo is a not for profit reasonable approximations. However the aim of the
Cambridge based project which aims to provide study is not only the network bandwidth usage but
affordable and sustainable IT networks to the world. other characteristics as well, in addition we wish to
The following section is an extract from the ndiyo measure those characteristics in different scenarios.
website (Ndiyo 2006). Some recommended scenarios are:
1. Large number of users using voice conversation
3.1 “Nivo” Ultra Thin Client and web applications.
2. Large number of users using video and web
A device called a Nivo (Network In, Video Out) is a applications.
highly optimized piece of electronics, dedicated to 3. Large number of users using only web
the purpose of displaying an interactive computer applications.
desktop over a network. The server simply sends to
the nivo - over the network using a simple
compression scheme - the pixels that need to be
displayed on the user's screen. Modern wired

6
 A PERFORMANCE EVALUATION OF AN ULTRA-THIN CLIENT SYSTEM

LAN 100/85 Mb/s

W.S
3 W.S W.S
2 1

Transmiting Node 10 bytes 2ms UDP

1500 bytes 2 ms UDP

Server Switch.
ns2 object

W.S W.S
4 5

Figure 1: The simulation model.

4 SIMULATION MODEL amount of data at different intervals towards its

client on their requests, so the transmission of data is
On the basis of the above defined structure of nivo duplex and in intervals from both sides.
networks we have developed a simulation model in The simulation model was designed in two
to predict the usage of network bandwidth in stages. First the topology of the network was
different scenarios. The details of the simulator and developed and tested (Figure 1). It is obvious from
the model developed are discussed further. the diagram that the topology here is that of a
switched LAN. The ideal bandwidth for an Ethernet
4.1 NS-2 LAN is 100 Mb/s but generally performance is
nearer to 85 Mb/s due to different losses. After
All the simulations in this project are developed with selecting User datagram Protocol (UDP) packets as
the help of Network Simulator 2 (NS-2). The latest the major traffic on the network some suitable traffic
version of NS-2 i.e. NS-2.30 was used. generators were attached to both entities i.e. The
NS-2 is an event driven network simulator Server and the Workstations. As the data generated
developed by UC Berkeley. NS-2 implements traffic from both sides is in intervals a Pareto traffic
behaviours, network protocols, routing, etc. for generator is used, generating traffic using a
simulation. Because it is open source software, probability density function i.e.
during the development many contributions have
been included from other researchers. NS-2 has
become a common tool for network researchers to
simulate and evaluate network related project.
Through the OTcl language interface users can This is sufficient to generate traffic at random
define a particular network topology, the protocols intervals but can be associated with other intervals
and applications that they want to simulate and the by its time interval parameters. The packet sizes are
form of the output that they want to obtain from the designated according to the given information i.e.
simulator quickly and clearly as a script. (Zhao and 1500 bytes size generated from the server and 10
Wu, 2006). bytes size generated from the workstations. The rate
at which these packets are generated depends upon
4.2 Thin Client Simulation Model applications used by the users at the server.

The available information about the network clearly

states that the Server should be sending a reasonable

7
ICE-B 2008 - International Conference on e-Business

5 OPERATION Different sessions of testing within our lab

provided us some more useful information such as if
the server is not able to support a large amount of
5.1 Overview
multiprocessing, this creates delay and a bottleneck
In order to describe the operation reference is again type situation. However this is not caused by the
made to Figure 1. As soon as a workstation sends a lack of network bandwidth but to the low power of
request to the server while running or initiating an the server (a Pentium4). In order to monitor traffic
application the server starts responding by sending a Ethereal was used at the server but we were able to
large amount of data in chunks of 1500 bytes with monitor only that traffic associated to the server and
assigned intervals i.e. with a delay of 200ms or connected users.
according to the application. There is a separate
traffic generator for each user therefore the volume
of data generated increases with the number of users. 6 RESULTS
Another major aspect in Traffic Generation is the
profile with which the generator is generating data. The data rates were recorded during our traffic
This option is easily available in Pareto & CBR generation, which was mainly HTTP traffic, but
traffic generators within NS-2. including some web streaming as well. These data
The workstations are connected with the server rates were used while producing the simulation
through an Ethernet switch hence the network according to our test bed as the Lab Network is
monitoring is performed at this point. In order to do almost same as our simulation model. Figure 2
so we have used a perl script, this script calculates shows that there are packet losses ranging from 2 to
the throughput at our preferred node at given time almost 3 Mbits in short time intervals. The highest
intervals. The node selected here as mentioned peak achieved is just more than 3.5 Mbits which is
earlier is our network switch because all of the well within the actual network bandwidth of around
network data will be flowing through it and the time 85Mb (eliminating different network losses from a
intervals of granularity at which the throughput of typical 100Mb LAN), but the processing of the
switch was calculated is 1 sec. systems is taken error free as a default in these
simulations. With little difference the results from
5.2 Test Bed the simulation compare with our measured
outcomes.
Before developing our simulation according to the When the same data rate was used for 30 users in
Ndiyo scenario we created a test bed within our lab. our simulation model the output that was achieved is
We wanted to compare our simulation data rates depicted in Figure 3. The throughput was almost ten
along with the data rates of that traffic which was times the 2 user throughput but yet again never
generated while using VNC (Virtual Network exceeded any problematic threshold. Another
Computing) software. This utility is widely used in interesting aspect of Figure 3 is that there were no
order to achieve remote desktop facility over major packet or throughput drops during the
different platforms i.e. it is platform independent. simulation. We believe that the small number of
The operation of VNC is almost the same as that of applications used in our test bed meant the data rates
the software used by the Ndiyo project, although were at an average level.
VNC provides a sophisticated connection setup and As mentioned earlier, our simulations were
delivery of data while using TCP. This differs from developed using data rates procured from our test
the Ndiyo project which uses a VNC over UDP bed and with minimal applications used in order to
approach in their software development. However, extend the data patterns using run length encoding to
for the purpose of exploring the activity of VNC, create our own scenarios. Doing this, we developed
this variation is not important, since we are not a simulation which produced a throughput as
measuring the relative performance or reliability. depicted in Figure 3. These steps were taken in order
We used Fedora Core4 and VNC packages to create simulations which were directly related to a
downloaded from www.Realvnc.com. VNC practical scenario.
provides a VNCVIEWER used at the user side and However in order to evaluate the bottleneck
the VNCSERVER to be run at the server. situations it was necessary to create some
VNCVIEWER provides an X11 session transmitted hypothetical scenarios where a number of
from the server. applications are used with each one having high data
generation rates. At this point it should be

8
 A PERFORMANCE EVALUATION OF AN ULTRA-THIN CLIENT SYSTEM

understood that as NIVO devices are in their test the server as the number of applications used
phase and are not produced in large quantities, we increases.
were unable to get our hands on NIVOs. To collect
such of data rates we used another network
monitoring tool OBSERVER, which provided us the 7 CONCLUSIONS
facility to record data rates for different applications
while using Remote Desktop Control. Data rates for A comparison of the four graphs reveals some very
different applications are shown in Table 1; however obvious but important points:
as these data rates were collected while using
• A LAN network can easily deliver multimedia
Remote Desktop Control the amount of TCP packets
services to its users, but in case of Nivo devices
is also included.
all of the processing is carried at the server end
and therefore data sent to the user from the
Table 1: Data rates used.
server is larger. Clearly, in the case of a video
Remote Remote Web Remote file being executed by a user while using
Video file Streaming HTTP network neighborhood facility all the processing
Traffic is done at the user end while only the file data is
500-2450 being taken from the remote location, whereas
Kb/s 727-900 Kb/s 97-150 Kb/s
in our scenarios the whole processed screen is
(25 fps)
sent over the network which ultimately results
in large amount of data transmission over the
These rates were obtained while using a single
network.
user, in order to create scenarios for large number of
user the traffic generator we can easily run-length- • Relating to our previous argument, if we
encode these rates. Also the data generation rates evaluate our thin client networks then it is
depend on the nature of the file or web page, if the obvious that large number of user could be
data set is too rich then more pixel data has to be accommodated by the network but only if we
sent over the network and this could cause an execute those applications which generate lower
increase in the data rates. Therefore different type of data rates. Multimedia applications can only be
video files, streams or web pages could generate executed if there is a small number of users (5-
different amount of data. Hence the above examples 10).Referring to our test bed results, web
can be best described as samples for such type of streaming can be achieved but not if a large
traffic. To create such a scenario where users are number of users are using it simultaneously.
using different applications at the same time, in • Thin client networks such as Ndiyo can provide
order to evaluate when a bottleneck situation occurs, affordable network solutions to those areas
we developed simulations with combination of the where a user’s requirements are more related to
above data rates. HTTP traffic rather than multimedia
Figure 4 depicts a condition where only 2 users applications.
are using the network. The amount of throughput is
almost double that recorded in previous simulations.
Whereas Figure 5 shows the situation where 30 REFERENCES
users at a time are connected to the network and
utilizing different services. Haichuan Zhao, Jianqiu Wu. 2005. Implementation and
It can be easily noticed that due to the increased simulation of HSDPA functionality with ns-2. Master
volume of data there is a significant amount of Thesis in Division of Automatic Control, Department
packet drop; Figure 5 shows that there are packet of Electrical Engineering at Linköping Institute of
Technology.
drops from around 6 Mbits to more than 10 Mbits
Kelly, T. 2002. Thin-client Web access patterns:
with in a very small period of time. However, it is Measurements from a cache-busting proxy. Computer
obvious that with a small number of users, the Communications Vol 25, Issue 1, March 2002, pp
available bandwidth is sufficient to support all of 357-366.
them. The packet drops which are noticed at Lai, A.M, Nieh, J., Bohra, B.,Nandikonda, V., Surana,
different intervals during the simulation mainly A.P, Varshneya, S. 2004. Improving Web Browsing
affect video file transmission, and could result in on Wireless PDAs Using Thin-Client Computing in
jerks in a multimedia file. These packet drops can Proc. 13th International Conference on World Wide
also be a result of reduced processing capability at Web, New York pp. 143-154

9
ICE-B 2008 - International Conference on e-Business

The Ndiyo Project. 2006 http://www.ndiyo.org/systems Pareto Distribution. .From wikipedia

<last accessed 08/04/08> http://en.wikipedia.org/wiki/Pareto_distribution <last
Tolia, N. Andersen, D. Satyanarayanan 2006. Qualifying accessed 08/04/08>
Interactive User Experience on Thin Clients IEEE Yang, S.J., Nieh J., Krishnappa, S., Mohla, A., Sajjadpour,
Computer Vol 39, No. 3 March 2006. pp 46 -52. M. 2003, Web Browsing Performance of Wireless
Wikipedia definition of Thin client. From Wikipedia, Thin-Client Computing Proceedings of 12th
2007. http://en.wikipedia.org/wiki/Thin_client <last International World Wide Web Conference Budapest,
accessed 08/04/08> pp. 68-79.

2 users (streaming)
4
3.5
3
2.5
2
1.5
1
0.5
0
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121
Time (s)

Figure 2: Data for two simultaneous streaming users.

35

30

25

20

15

10

0
0 20 40 60 80 100 120
Time (s)

Figure 3: Data for 30 simultaneous streaming users.

10
 A PERFORMANCE EVALUATION OF AN ULTRA-THIN CLIENT SYSTEM

2 users
7
6
5
4
3
2
1
0
0 20 40 60 80 100 120
Time (s)

Figure 4: Data for two simultaneous video users.

30 users
78

76

74
Throughput (Mbits/s)

72

70

68

66

64

62

60
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 115 121
Time (s)

Figure 5: Data for 30 simultaneous video users.

11
 USING CO-OCCURRENCE TO CLASSIFY UNSTRUCTURED DATA
IN TELECOMMUNICATION SERVICES

Motoi Iwashita, Ken Nishimatsu and Shinsuke Shimogawa

NTT Service Integration Laboratories, Midori-cho 3-9-11 Musashino, Tokyo, Japan
[email protected], [email protected], [email protected]

Keywords: Co-occurrence, Correspondence analysis, Telecom operation, Text mining.

Abstract: A variety of services have recently been provided according to the highly-developed networks and personal
equipment. Connecting this equipment becomes more complicated with advancement of these day by day.
Because software is often updated to keep up with advancements in services or security, problems such as
no-connection increase and determining the cause become difficult in some cases. Telecom operators must
understand the situation and act as quickly as possible when they receive customer enquiries. In this paper,
we propose one method for analyzing and classifying customer enquiries that enables quick and efficient
responses. Because customer enquiries are generally stored as unstructured textual data, this method is based
upon a co-occurrence technique to enable classification of a large amount of unstructured data into patterns.

1 INTRODUCTION tured data, i.e., it is an unstructured data. Let us take

a customer phone enquiry about no-connection as an
A conventional fixed telephone service is simply pro- example. There are several causes, such as failure of
vided by a telephone network. Because the net- the optical fiber, modem, or application. Both nor-
work structure is simple, it is easy to determine the mal and strange situations are included in text such as
cause of service problems. Furthermore, telecom gthe connection to the Internet is OK, but e-mail can-
operators with long-accumulated know-how can act not be senth. Therefore, it is important to classify the
quickly. Recently, broadband infrastructures with unstructured textual data with accuracy.
Asymmetric Digital Subscriber Line (ADSL) and op- A text mining technique, such as morphological
tical fibers have penetrated the telecommunications analysis, syntax analysis, co-occurrence relation, etc.,
industry. This trend has induced the expansion of is effective (Ohsumi, 2006),(Sato et al., 2007),(Sulli-
a variety of services, such as the exponential use of van, 2001),(Toda et al., 2005). This technique is ap-
the Internet, the provision of Voice over Internet Pro- plicable to customer questionnaire analyses in prod-
tocol (VoIP) and video distribution services, and se- uct development, word searches in portal sites such
curity software countermeasures against virus attacks as Google and Yahoo, term frequency analyses in
on PCs. Therefore, the end-to-end network structure web logs (blogs) or customer generated media, article
has become complicated, considering the connection classification by keyword in news articles, and eval-
of home equipment, such as a modem, and the setup uation indexes of a company’s image. Mainly mor-
of its software. As a result, discovering the cause of phological analysis is applied in these areas to sur-
problems is difficult. Connecting service equipment vey trends by analyzing the frequency of terms in se-
will continue to become more complicated in the near lected text. A keyword is extracted as a topic of a
future, if we consider the drive forwards ubiquitous sentence in terms of the features of the network struc-
services. ture (Masuo et al., 2001),(Ohsawa et al., 1997),(Cut-
Customer satisfaction decreases when a long time ting et al., 1992),(Ho et al., 2001),(Leuski, 2001).
is spent on restoration because discovering the cause Clustering and co-occurrence related methods have
is difficult. Therefore, it is necessary to understand been proposed classify keywords and relate them to
the features of the problem first to resolve the situa- synonymous terms, different words having the same
tion by data classification. meaning, and synonyms, which have similar mean-
The saved information is not classified as struc- ings (Uejima et al., 2004), (Rodoriguezd et al., 1998).

12
 USING CO-OCCURRENCE TO CLASSIFY UNSTRUCTURED DATA IN TELECOMMUNICATION SERVICES

An improved method was proposed for synonymous 2.2 Limitations of Morphological

term classification in fuzzy searches for the aim of Analysis
failure analysis (Naganuma et al., 2005). Simply
understanding failure trends and noting customer re- Figure 1 shows the relationship between term fre-
quirements when analyzing an enquiry then analyzing quency and its ranking for 10,000 customer enquiries
the word trends is not always effective. Understand- about telecom services. The terms were classified and
ing the meaning of sentences is essential. There are counted by morphological analysis, and those that ap-
no effective methods to semantically analyze text that pear more than 50 times are shown in Fig. 1.
are applicable to telecom management.
A text classification method from a semantic point
of view that considers the features of telecom services
and the co-occurrence of terms for classifying and an-
alyzing a large amount of unstructured data consisting
of customer enquiries is proposed. The difficulties in
analyzing textual data in telecom services when con-
ventional techniques are used is explained in Sec. 2.
Section 3 describes the features of telecom services.
Our classification method is proposed in Sec. 4, and
the results are discussed in Sec. 5.

2 DATA ANALYSIS IN TELECOM

OPERATIONS
Figure 1: Relationship between term frequency and its rank-
ing.
2.1 Necessities of Data Classification
and Text Mining ”Category 1” is text relating to equipment, such as
a modem, PC, telephone, etc., whereas ”Category 2”
In general, the telecom operator saves the customer is text relating to a failure phenomenon, such as setup
enquiry and the coping process as information. The trouble, no-connection, cable breakdown, etc. ”Cat-
aim of this is to enable the finding of similar prob- egory 0” means terms analyzed without considering
lems by searching with related keywords and to en- categories.
able quick action when such problems occur. These All the patterns show a n/1 feature, so they fol-
coping processes are effective for sharing the knowl- low a power law (Newman, 2005), (Zipf, 1949), as
edge among assigned telecom operators and for im- is normal in general sentences. A power law implies
proving their skills. Therefore, this method is useful that there are many kinds of terms in the textual data
when problems happen. However, the drawback of and means that there are many types of customer en-
this method is that it is impossible to get an overview quiries. Since the slope of Category 1 is steeper than
of all the possible patterns of a problem and to es- the other two, the terms used in that textual data are
tablish coping processes for more complex problems limited compared with the terms in the other cate-
in advance. Therefore, it is necessary to establish an gories. On the other hand, the slope of Category 2
effective coping process for customer enquiries, as- is gradual, which shows the variety of failure phe-
sign optimal operators due to advance classification nomenon terms. These results show that it is possible
of customer enquiries and survey the failure trend. to only survey the terms with high frequency, such as
Information generally consists of text. It takes a PC, telephone, internet, etc. However, it is necessary
long time to analyze text word by word and to clas- to understand the relationships among the terms.
sify large amounts of textual data. Therefore, an ef-
fective method based on a text mining technique, such 2.3 Limitations of Correspondence
as term frequency analysis, number of synonymous Analysis
words determination and related terms extraction is
needed. To classify into groups by term features, correspon-
dence analysis (Benzecri, 1992), (Hayashi, 1993),
(Takahashi, 1996) is applied. The terms with fre-
quency rankings higher than 400 were selected from

13
ICE-B 2008 - International Conference on e-Business

10,000 customer enquiries. The analysis of up to the (e.g. description of product name, abbreviation). Let
10th factor is shown in Table 1, and a graph of the 1st us consider a situation where optical and telephone
and 3rd factors is shown in Fig. 2. services are provided to a customer. It is difficult
i-th contribution rate, ri , is calculated as follows. to determine whether the cause of a problem is an
optical cable, modem failure, or application prob-
ri = (Eigenvalue(i))/Σ j (Eigenvalue( j)) (1)
lem in the case of a phone call from a customer
These results show the difficulty of describing the about no-connection. Moreover, associated factors
features. This is because the accumulated contribu- that complicate the situation include partial trouble,
tion rate is under 20%, even considering the 10th fac- e.g. the Internet works whereas e-mail does not, or
tor, and most of the information is gathered in a small no-connection because of software compatibility.
area. Therefore, it is clear that we need some prepa- To summarize, it is difficult to apply a text mining
ration before classification. technique directly to raw textual data in telecom man-
agement for semantic/structural classification. There-
Table 1: Correspondence analysis. fore, we need a modification to distinguish the fea-
tures of telecommunication services. Telecommuni-
cation service such as internet connection, VoIP, is
generally provided by an end-to-end network con-
sisting of a telephone, PC, the carrier’s network, the
provider’s server, etc. Fig. 3. There is clearly an
event feature for each component of the network. We
can predict that the component, such as the service,
telephone, PC, and network, is strongly related to the
problem, such as failure, misconfigurated of set-up,
and cable breakdown, respectively. Therefore, by des-
ignating the network factor as one event and the prob-
lem as the other event, we can construct a semantic
representation. Moreover, if a problem occurs in one
piece of equipment in a network, it is expected to lead
to problems in the other component or to other events.

Figure 2: Relationship between 1st and 3rd factors.

3 FEATURES OF TEXTUAL DATA

IN TELECOM SERVICES
The features of textual data in telecom services are
explained in this section. An operator writes down in-
formation about a customer enquiry. Therefore, the
style of the description deeply depends on the op-
erator. That is, the description style is not ordered.
There may be abbreviations of terms, wording that
only the specified operator can understand, and so
on. Moreover, there are many synonymous terms.
Taking service specifications of optical fibers as an
example, there are descriptions using general names Figure 3: Telecom features categorization.
(e.g. optical service, Fiber-To-The-Home (FTTH)),
and special/abbreviated names for customer enquiries

14
 USING CO-OCCURRENCE TO CLASSIFY UNSTRUCTURED DATA IN TELECOMMUNICATION SERVICES

4 CLASSIFICATION METHOD for any x and y such that x, y ∈ A or B.

BASED ON CO-OCCURRENCE • Procedure 4: Transition among multiple terms
(Fig. 5 (b))
4.1 Framework of Classification - Let us choose a pair of terms such that
C(x, y) ≥ α, where α is a given threshold.
The requirement for textual data classification is Step 1: Select a pair with Categories A and B
based on the ability to cover all textual data and to satisfying α ≤ C(i, j), i ∈ A, and j ∈ B.
fit the operator’s thinking. It is desirable that this Step 2: Select a pair with Category A satisfying
classification be determined from the viewpoints of α ≤ C(i, j), i, j ∈ A, and α ≤ C(i, k), i ∈ A, and
term frequency, co-occurrence, and cause-effect rela- k ∈ B.
tionship, as shown in Table 2. Term frequency can Step 3: Select a pair with Category B satisfying
tell us what kind of customer enquiries often appear, α ≤ C(i, j), i, j ∈ B, and α ≤ C(i, k), i ∈ B, and
while co-occurrence tells us which terms are strongly k ∈ A.
related. Cause-effect relationship tells us the relation-
ship among multiple terms such as network compo-
nent and porblem event.

Table 2: Criteria for classification.

Figures 4 and 5 show the frameworks in which

textual data is classified in terms of the previous three
criteria.
Classification framework: Figure 4: Term classification based on morphological anal-
ysis.
• Procedure 1: Classification by type of access (Fig.
4 (a))
- Textual data is classified in terms of access, i.e.,
type of service, such as dial-up, ADSL, FTTH,
and so on.
• Procedure 2: Classification by category (Fig. 4
(b))
- Textual data is classified in terms of categories
based on morphological analysis. For example,
Category A means component, such as a service,
telephone, PC, modem, etc., while Category B
means the problem, such as no-connection, mis-
configuration, etc.
• Procedure 3: Calculation of term frequency and
co-occurrence (Fig. 5 (a))
- The frequency of both x and y appearing in tex-
tual data is represented by f (x, y). Let us se-
lect a pair in terms of a frequency greater than
β, where β is a given threshold. Then, calculate Figure 5: Calculation based on co-occurrence.
co-occurrence as follows:
C(x,y) = f(x,y)/(f(x)+ f(y)− f(x,y)) (2)

15
ICE-B 2008 - International Conference on e-Business

4.2 Transition of Relativity Among 5.2 Co-occurrence Among Categories

Multiple Terms
Co-occurrence among categories was calculated for
The effectiveness of procedure 4 in the previous sec- the textual data from 1000 customer enquiries, as
tion is explained in this section. For example, a cus- shown in Fig. 6. We chose the pairs marked that have
tomer makes a general complaint that he/she is not a value more than the given threshold. Then, seven
able to send e-mail from his/her PC as there is no- pairs of terms were selected by co-occurrence. More-
connection with the Internet. The cause might not be over, there were pairs that had a strong transition rate
in the PC but in the modem setup in that case. There- in the same category. We merged those pairs and se-
fore as a transitional way of thinking, we need rela- lected them as the 8th and 9th choices Fig. 6.
tive keywords to suggest other causes. The relation- Let us compare the proposed method and the
ship among multiple terms should be clear. If we cal- method using only term frequency for selecting pair
culate all co-occurrence values between the pairs of of terms. The frequency of 8th choice is 70 by
terms in Categories A and B, we need a long calcula- pair of terms-frequency, while that of the 8th choice
tion time n2 × n2 = o(n4 ). If we use procedure 4, on by the proposed method is 54. Because the differ-
the other hand, we can reduce the calculation time to ence between two choices is small in that amount of
n × n = o(n2 ). This is because co-occurrence is cal- data and the 8th choice by pair of terms-frequency
culated for each category as a unit. is smaller than the given threshold, it only represents
weak relationships in the features of the text. The pro-
posed method is possible to classify and understand
complicated structure through selecting and relating
5 DISCUSSION OF RESULTS strong co-occurrence. Therefore, the choice by co-
occurrence represents the features of the text.
5.1 Co-occurrence in a Category Figure 7 shows the choice of pairs with the thresh-

The order of choice strongly depends on the thresh-

old in classification procedures. The relationship be-
tween term frequency and co-occurrence in textual
data from 1000 customer enquiries is shown in Table
3. 1000 customer enquiries are the field data saved
by telecom management for IP-based services during
the daytime at one day in 2007. Pairs of terms are
ordered by term frequency. There are three cases for
co-occurrence with the threshold as a parameter. The
selected pairs become similar in term frequency when
the threshold decreases. Because we screened and
chose pairs with high frequency by procedure 2, using
the choice decided by co-occurrence was effective for
representing the features of the text. The threshold is
given high value at first step. If the number of pairs is
small, α decreases as a second step. In this way, itera-
tion step of decreasing α can lead relationship among
terms.
Table 3: Choice of pair of terms.

Figure 6: Relationship between categories.

16
 USING CO-OCCURRENCE TO CLASSIFY UNSTRUCTURED DATA IN TELECOMMUNICATION SERVICES

old as a parameter. The number of candidates in- REFERENCES

creases when the threshold decreases. This is because
of the weakness of co-occurrence. The number of Benzecri, J.-P. (1992). Correspondence Analysis Hand-
pairs with a transition rate grows when the threshold book. Marcel Dekker.
decreases. The 6th choice is (1A and 2A) & (1B and Cutting, D., Kager, D., and Tukey, J. (1992). Scatter/gather:
2B) when α = 0.25. 1A and 2A correspond to ”Inter- A cluster-based approach to browsing large document
net” and ”VoIP” respectively, while 1B and 2B cor- colelctions. In Proc. 15th Annual International ACM
SIGIR Conference on Research and Development in
respond to ”connection is OK” and ”no connection” Information Retrieval.
respectively. We can classify the text in a semantic
Hayashi, C. (1993). Quantification -Theory and Method.
sense, e.g., ”Internet is OK, but VoIP has no connec- Asakura-shoten.
tion”.
This type of proposed data classification can get Ho, X., Ding, C., Zha, H., and Simon, H. (2001). Automatic
topic identification using webpage clustering. In Proc.
an overview of all the possible patterns of a problem 2001 IEEE International Conference on Data Mining.
and establish coping processes in advance. Further-
Leuski, A. (2001). Evaluating document clustering for in-
more, it has possibilities to mine the potential cus- teractive information retrieval. In Proc. 2001 ACM
tomer requirement that leads to new business. International Conference on Information and Knowl-
edge Management.
Masuo, Y., Ohsawa, Y., and Ishizuka, M. (2001). Document
as a small word. In Proc. JSAI 2001, International
workshop (LNAI2253), pages 444–448.
Naganuma, K., Isonishi, T., and Aikawa, T. (2005). Diamin-
ing: Text mining solution for customer relationship
management. Mitsubishi Technical Report, 79-4:259–
262.
Newman, M. (2005). Power laws, pareto distributions and
zipf’s law. Contemporary Physics, 46:323–351.
Ohsawa, Y., Benson, N., and H.Yachida (1997). Keygraph:
Automatic indexing by co-occurrence graph based on
building construction metaphor. In Proc. IEEE Forum
on Research and Technology Advances in Digital Li-
Figure 7: Threshold and pairs.
braries.
Ohsumi, N. (2006). Mining of textual data. recent trend
and its direction. http://wordminer.comquest.co.jp
6 CONCLUSIONS /wmtips/pdf/20060910 1.pdf.
Rodoriguezd, M., Gomez-Ilidalgo, J., and Diaz-Agudo, B.
A classification technique for customer enquiries is (1998). Using wordnet to complement training in-
formation in text categorization. In Proc. Recent Ad-
needed due to the increasing complexity of the con- vances in Natural Language Processing.
nections in end-to-end networks in the telecom oper-
Sato, S., Fukuda, K., Sugawara, S., and Kurihara, S. (2007).
ating field. In this paper, we proposed one method
On the relationship between word bursts in docu-
for analyzing and classifying customer enquiries that ment streams and clusters in lexical co-occurrence
enables quick and efficient responses. Because cus- networks. IPSJ, 48-SIG14:69–81.
tomer enquiries are generally stored as unstructured Sullivan, D. (2001). Document Warehousing and Text Min-
textual data, this method is based upon morphologi- ing. John Wiley.
cal analysis and co-occurrence techniques to enable Takahashi, S. (1996). Correspondence Analysis by Excel.
classification of a large amount of unstructured data Ohm-sya.
into patterns. We applied the proposed method to
Toda, H., Kataoka, R., and Kitagawa, H. (2005). Clustering
1000 customer enquiries and evaluated its effective- news articles using named entities. IPSJ SIG Techni-
ness. The method can apply not only to establish cop- cal Report, 2005-DBS-137:175–181.
ing processes in advance but also to mine potential Uejima, H., Miura, T., and Shioya, I. (2004). Improving
requirement for new business. text categorization by synonym and polysemy. Trans.
We are currently conducting further study on ap- on IECIE, J87-D-I, No. 2:137–144.
plying this method to large amounts of data and on Zipf, G. (1949). Human Behavior and the Principle of Least
determining a threshold for telecom operation. Effort. Addison-Wesley.

17
SHORT PAPERS
 FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

Toshio Kodama1, Tosiyasu L. Kunii2 and Yoichi Seki3

1
Venture Business Department, Maeda Corporation, 2-10-26 Fujimi, Chiyoda-ku, Tokyo, Japan
[email protected]
2
Morpho, Inc., The University of Tokyo,7-3-1 Hongo, Bunkyo-ku, Tokyo, Japan
[email protected], [email protected]
3
Software Consultant, 3-8-2 Sakaemachi, Hino City, Tokyo, Japan
[email protected]

Keywords: Cellular model, formula expression, topological space, cellular space, condition formula.

Abstract: Cyberworlds are distributed systems on the Web, and are constantly evolving like living things, creating
value. Currently, numerous Web business applications, such as cyberworld systems are being built, but in
the development of the systems, combinatorial explosion happens because schemas and application
programs must be modified whenever schemas change. We designed and implemented the logic of a
flexible data search function by employing a condition formula on the cellular data system. This is the
starting point to the implementation of the process graph theory, which makes a linear approach to
overcoming combinatorial explosion possible.

1 INTRODUCTION model based on the Incrementally Modular

Abstraction Hierarchy (IMAH) can model the
The system of cyberworlds is a distributed system. architecture and the changes of cyberworlds and real
One of the features of cyberworlds is that data worlds from a general level to a specific one,
dependencies are constantly changing in them. preserving invariants while preventing combinatorial
Cyberworlds are more complicated and fluid than explosion (T. L. Kunii and H. S. Kunii, 1999: 19-
any other previous worlds in human history and are 21). From the viewpoint of IMAH, existing data
constantly evolving. For example, millions of users models are positioned as special cases. For example,
manage their own blog information every day UML can model objects at levels below the
through Web services on mobile phones, like SNS in presentation level, and in the relational data model, a
Japan, which is considered one of main elements of relation is an object at the presentation level which
cyberworlds. At the same time, user requirements extends a cellular space because it has necessary
for cyberworlds also change and get more attributes in which a type is defined, while the
complicated as cyberworlds change. If you analyze processing between relations is based on the set
data using the existing technology in business theoretical level. In the object-oriented model, an
application development, you have to modify the object is also the object in the presentation level,
schema design and application programs whenever which extends a cellular space, while the relation
schemas or user requirements for output change. between Class is the tree structure, which is a special
That leads to combinatorial explosion, because user case of a topological space. An Object in XML is
requirements, and their combinations and schemas considered a special case of a cellular space which
must be specified clearly at the design stage in extends a topological space, because an attribute and
general business application development. That is a a value of it are expressed in the same tag format.
fundamental problem, so we have to reconsider In our research, one of the authors (Y. Seki)
development from the data model level. proposed an algebraic system called Formula
Is there a data model that can reflect the changes Expression as a development tool to realize the
in schemas and user requirements for output to cellular model. T. Kodama has actually implemented
analyze data in cyberworlds? We believe that the CDS using Formula Expression (Toshio Kodama,
cellular model proposed by one of authors (T. L. Tosiyasu L. Kunii and Yoichi Seki, 2006: 65-74). In
Kunii) is the most suitable model. The cellular this paper, we have introduced a new concept of a

21
ICE-B 2008 - International Conference on e-Business

condition formula and its processing maps into CDS. For a detailed explanation of each level, please refer
A condition formula search is a very effective to our earlier paper.
measure when you want to analyze data in
cyberworlds without losing consistency in the entire Formula Expression in the alphabet is the result of
system, since you can search for the data you want finite times application of the following (1)-(7).
without changing application programs, if you (1) w (∈Σ) is Formula Expression
employ a condition formula search. In addition, we (2) unit element ε is Formula Expression
put emphasis on practical use by taking up an (3) zero element φ is Formula Expression
example. First, we explain the cellular model briefly (4) when r and s are Formula Expression, addition
and add a new definition to Formula Expression. of r+s is also Formula Expression
(Section 2) Second, we design logical operation as a (5) when r and s are Formula Expression,
condition formula generalizing search conditions of multiplication of r×s is also Formula
users by Formula Expression, and design its Expression
processing maps to process a condition formula to (6) when r is Formula Expression, (r) is also
each topological space (Section 3) and implement Formula Expression
them. (Section 4) We demonstrate the effectiveness (7) when r is Formula Expression, {r} is also
of CDS by developing a business application Formula Expression
system, thereby abbreviating the process of (8) when r is Formula Expression, [r] is also
designing and implementing most application Formula Expression
programs. (Section 5) It is a bidding results data
search system where the data of files, which In this paper, we have added the 3rd bracket [] of (8)
schemas differ, are inputted without designing in the definition of Formula Expression. The
schemas. A more flexible data search is possible by algebraic structure is the following.
employing a condition formula search in the system.
[r]×(s+t) = [r]×s+[r]×t, (r+s)×[t] = r×[t]+s×[t]
In this way, if [] is added to a formula and
2 THE CELLULAR MODEL AND becomes the factor, it behaves like an identifier,
since [] is never removed by any map.
FORMULA EXPRESSION
The following list is the Incrementally Modular
Abstraction Hierarchy (IMAH) in the cellular model 3 THE DEFINITION OF
to be used for defining the architecture of LOGICAL OPERATION
cyberworlds and their modeling:
1. the homotopy (including fiber bundles) level 3.1 A Condition Formula
2. the set theoretical level
3. the topological space level If users can specify search conditions, data search
4. the adjunction space level (Fig 1) will become more functional when searching data
5. the cellular space level from data storage. Here, we introduce the function
6. the presentation (including geometry) level for specifying conditions defining a condition
7. the view/projection level formula by Formula Expression into CDS. Let
propositions P, Q be sets which include characters p,
q respectively. The conjunction, disjunction and
negation of them in logical operation are defined by
Formula Expression as follows:
1) Conjunction
P∧Q = p×q
2) Disjunction
P∨Q = p+q
3) Negation
¬P = !p
Figure 1: An example of e-manufacturing on an adjunction
space level.

22
 FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

A formula created from these is called a condition h(x, p) = i f(xi, p)

formula. Here "!" is a special factor which means h(x, p×q) = i f(f(xi, p), q)
negation. Recursivity by () in Formula Expression is h(x, p+q) = i f(xi, p)+i f(g(xi, p), q)
supported so that the recursive search condition of a h(x, !p) = i g(xi, p)
user is expressed by a condition formula. An h(x, !(p+q)) = i g(g(xi, p), q)
example is the following. h(x, !(p×q)) = i g(f(xi, p), q)+i g(xi, p)

¬(P∨Q)∧((R∧S)∨(T∧U)) = !(p+q)(r×s+t×u)

3.2 A Quotient Acquisition Map

and a Remainder Acquisition Map
A quotient acquisition map f is a map that has a term
that includes a specified factor, and a remainder
acquisition map g is a map that has a term that
doesn’t include a specified factor. These two maps
are fundamental in processing a condition formula.
(3.3.) If you assume the entire set of terms to be A,
B and the entire set of factors to be C, f:A×C h B and
g:A×C h B. Arbitrary terms r, s, t, u, v, w, x, y (A)
and an arbitrary factor p (C) follow these rules:
f: r, p h φ (when r doesn’t include p)
f: r×p×s, p h r×p×s
f: r×(s+t×p×u+v)×w, p h r×t×p×u×w
f: r×{s+t×p×u+v}×w, p h r×t×p×u×w
f: r×[s+t×p×u+v]×w, p h r×[s+t×p×u+v]×w
g: r, p h r (when r doesn’t include p)
g: r×p×s, p h φ
g: r×(s+t×p×u+v)×w, p h r×(s+v)×w
g: r×{s+t×p×u+v}×w, p h r×{s+φ+v}×w
g: r×[s+t×p×u+v]×w, p h φ Figure 2: Images from the condition formula processing
map.
If p is an identifier, f (or g) is usually repeated until p
is not enclosed in a bracket. Simple examples of Fig 2 is images of the map h. Here, f is a quotient
both maps are shown below. acquisition map and g is a remainder acquisition
f: a(b(c+d(e+f))+g)h, d h a×b×d(e+f)h map. It is obvious that any complicated condition
g: a(b(c+d(e+f))+g)h, d h a(b×c+g)h formula can be processed by the combinations of the
above four correspondences. A simple example is
3.3 A Condition Formula Processing shown below.
Map x=
"animal{color+size}(flesheating(bear{brown+big
A condition formula processing map h is a map that }+monkey{brown+small}+orangutan{darkbrown
gets a disjoint union of terms which satisfies a +big}+tiger{brown×black+big}+fox{brown×whi
condition formula from a formula. If you assume x te+small}+bear{black+big})+grasseating(horse{(
to be a formula and xi to be a term which consists of white+brown)+middle}+koala{brown+small}+g
x (namely ixi = x) and p, !p, p+q, p×q to be oat{white+small}+hamster{white+verysmall}+p
condition formulas, the images of (x, p), (x, !p), (x, anda{black×white+big}+zebra{black×white+mid
p+q), (x, p×q), (x, !(p+q)), (x, !(p×q)) by h are the dle}+giraffe{yellow×black+verybig}+elephant{g
following: ray+verybig}+mouse{gray+verysmall}))"

23
ICE-B 2008 - International Conference on e-Business

Output case 1.
User requirement: "information about a horse and a
zebra in x is required"
A condition formula = "horse+zebra"
h(x, horse+zebra)
=f(x, horse)+f(g(x, horse), zebra)
=animal{color+size}grasseating(horse{(white+bro
wn)+middle}+zebra{black×white+middle})
Output case 2.
User requirement: "information about animals
whose size is big or very big and grass-eating is
required"
A condition formula = "size (big+verybig) grasseat
ing"
h(x, size(big+verybig) grasseating)
=f(f(f(x, size), big+verybig), grasseating)
=f(f(f(x, size), big)+f(g(f(x, size), big), verybig),
grasseating)
=animal×size×grasseating(panda×big+giraffe×very
big+elephant×verybig)

4 IMPLEMENTATION
This system is a web application developed using
JSP and Tomcat 5.0 as a Web server. The client and
the server are the same machine. (OS: Windows XP; Figure 3: The flowchart of the algorithm for a quotient
CPU: Intel Core2 Duo, 3.00GHz; RAM: 3.23Gbyte; acquisition map.
HD: 240GB)
Fig 3 is the flowchart of the algorithm of a were downloaded in CSV format from the official
quotient acquisition map which is the main function website of each bureau in the Ministry of Land,
of a condition formula search. Details are Infrastructure, Transport and Tourism (MLIT,
abbreviated due to the restriction on the number of http://www.mlit.go.jp/chotatsu/kekka/kekka.html) in
pages. In this algorithm, the absolute position of the Japan. The feature of the files is that their schema
specified factor by the function of the language and changes little from month to month or from bureau
the term including the factor are acquired first. to bureau. Once you convert the CSV data files to
Next, the nearest brackets of the term are acquired formulas in CDS, you can unify them into a data
and because the term becomes a factor, a recursive storage file (.txt) by the function of a disjoint union
operation is done. +. After that, a user can search for the data she/he
wants from the data storage by creating a condition
formula. This system is actually being used in
5 CASE STUDY: A BIDDING Maeda Corp. which one of authors (T. Kodama)
belongs to.
RESULTS DATA SEARCH
SYSTEM 5.2 The Space Design
5.1 Outline We design a formula for the spaces as follows.
Σfile i×code[{Σattributei,j}](Σk[{Σvaluei,j,k}])
We have developed a business application system
using CDS for searching bidding results data for filei: a factor which expresses a file name
public construction projects. Many of the data files attributei,j: a term which expresses an attribute name
of filei

24
 FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

valuei,j,k: a term which expresses a value of an

attributei,j

Figure 5: CSV data of bidding results from May 2007 in

the Tohoku bureau.

Next, add the CSV data from June in the Kanto

bureau, which schema is slightly different from that
in formula 5.3-1, convert it to a formula (formula
5.3-2) in the same way and add it to the data storage
file using + function.

Figure 4: The data structure of the bidding results data

search system.

5.3 Data Conversion and Data Input

In this subsection, we simplify the input data without
losing generality. Let the CSV data of Figure 5 be Figure 6: CSV data of bidding results from June 2007 in
bidding results from May, 2007 in the Tohoku the Kanto bureau.
bureau. First, convert the downloaded CSV data to a
formula (formula 5.3-1) as a cell space and add it to Formula 5.3-2.
the data storage file.
(formula5.3-1)+MayOf2007InKanto×code[{bureau×name+proje
Formula 5.3-1. ct×name+place+bid×date+contract×date+delivery-month+projec
t×kind+bid×kind+bidding×company+bidding×price}](1[{kanto+
MayOf2007InTohoku×code[{bureau×name+project×name+bid×
C+p1+15/06/2007+17/07/2007+5+general×construction+general
date+contract×date+project×kind+bid×kind+bidding×company+
×bid×WTO+C1+800000000}]+2[{kanto+C+p1+15/06/2007+17/
bidding×price}](1[{tohoku+A+12/05/2007+14/05/2007+general
07/2007+5+general×construction+general×bid×WTO+C2+7800
×construction+general×bid+C1+1000000}]+2[{tohoku+A+12/05
00000}]+3[{kanto+C+p1+15/06/2007+17/07/2007+5+general×c
/2007+14/05/2007+general×construction+general×bid+C2+1100
onstruction+general×bid×WTO+C3+800000000}]+4[{kanto+C+
000}]+3[{tohoku+A+12/05/2007+14/05/2007+general×construct
p1+15/06/2007+17/07/2007+5+general×construction+general×bi
ion+general×bid+C3+1100000}]+4[{tohoku+A+12/05/2007+14/
d×WTO+C4+900000000}]+5[{kanto+C+p1+15/06/2007+17/07/
05/2007+general×construction+general×bid+C4+9800000}]+5[{
2007+5+general×construction+general×bid×WTO+C5+6800000
tohoku+A+12/05/2007+14/05/2007+general×construction+gener
00}]+6[{kanto+C+p1+15/06/2007+17/07/2007+5+general×const
al×bid+C5+1050000}]+6[{tohoku+A+12/05/2007+14/05/2007+
ruction+general×bid×WTO+C6+780000000}]+7[{kanto+C+p1+
general×construction+general×bid+C6+1100000}]+7[{tohoku+B
15/06/2007+17/07/2007+5+general×construction+general×bid×
+12/06/2007+13/06/2007+general×construction+general×biC×W
WTO+C1+820000000}]+8[{kanto+D+p2+16/06/2007+18/08/20
TO+C1+50000000}]+8[{tohoku+B+12/06/2007+13/06/2007+ge
07+12+electric×facilities+general×bid+C2+60000000}]+9[{kant
neral×construction+general×bid×WTO+C2+51000000}]+9[{toh
o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general
oku+B+12/06/2007+13/06/2007+general×construction+general×
×bid+C3+58000000}]+10[{kanto+D+p2+16/06/2007+18/08/200
bid×WTO+C3+50500000}]+10[{tohoku+B+12/06/2007+13/06/2
7+12+electric×facilities+general×bid+C4+60000000}]+11[{kant
007+general×construction+general×bid×WTO+C4+51500000}]
o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general
+11[{tohoku+B+12/06/2007+13/06/2007+general×construction+
×bid+C5+59000000}]+12[{kanto+D+p2+16/06/2007+18/08/200
general×bid×WTO+C5+49000000}]+12[{tohoku+B+12/06/2007
7+12+electric×facilities+general×bid+C6+61000000}]+13[{kant
+13/06/2007+general×construction+general×bid×WTO+C6+500
o+D+p2+16/06/2007+18/08/2007+12+electric×facilities+general
00000}]+13[{tohoku+B+12/06/2007+13/06/2007+general×const
×bid+C7+60000000}])
ruction+general×bid×WTO+C7+51000000}])

In this way, you can add data to the data storage

after converting it to a formula for a cell space using

25
ICE-B 2008 - International Conference on e-Business

+ function. In doing this, you don’t need to consider 6/2007+18/08/2007+12+electric×facilities+general×bid+C2+60

differences in schema at all. 000000}]+9[{kanto+D+p2+16/06/2007+18/08/2007+12+electri
In the same way, you can add data from another c×facilities+general×bid+C3+58000000}]+10[{kanto+D+p2+1
organization, which schema is completely different 6/06/2007+18/08/2007+12+electric×facilities+general×bid+C4
from others', as a formula for a cell space to the data +60000000}]+11[{kanto+D+p2+16/06/2007+18/08/2007+12+e
storage file. lectric×facilities+general×bid+C5+59000000}]+12[{kanto+D+
p2+16/06/2007+18/08/2007+12+electric×facilities+general×bid
+C6+61000000}]+13[{kanto+D+p2+16/06/2007+18/08/2007+
5.4 Data Conversion and Data Input
12+electric×facilities+general×bid+C7+60000000}])

When you search for data you want, you create

condition formulas according to requirements and
get an image of the formula in data storage by the
condition formula processing map h, you can get the
data you want. Examples and figures (Fig 7,8) are
shown below. Figure 8: The output result by the condition formula
If you want to search for data for "construction "kanto×!WTO".
projects of Company C1 or C2 and for WTO (World
Trade Organization) property", you make the If you want to get attribute values by specifying an
condition formula "(C1+C2)WTO", and get the attribute name, you remove "[]" once from the
image of formula 5.3-2 by the condition formula formula and get the image by the quotient
processing map h. acquisition map f. An example is shown.
If you want to search for data for "values of an
h (formula 5.3-2, (C1+C2)WTO) attribute of bid date in the Tohoku bureau", you get
= MayOf2007InTohoku×code[{bureau×name+project×name+bi the image by the composition map of f and h.
d×date+contract×date+project×kind+bid×kind+bidding×compan
Assume that formula 5.3-2' is the formula after
y+bidding×price}](7[{tohoku+B+12/06/2007+13/06/2007+gener
removing all "[]" from the formula 5.3-2.
al×construction+general×biC×WTO+C1+50000000}]+8[{tohok
u+B+12/06/2007+13/06/2007+general×construction+general×bi
f (h (formula 5.3-2', kanto), project×name)
d×WTO+C2+51000000}])+MayOf2007InKanto×code[{bureau×
=MayOf2007InKanto×code{bid×date}(1{15/06/2007}+2{15/06
name+project×name+place+bid×date+contract×date+delivery-m
/2007}+3{15/06/2007}+4{15/06/2007}+5{15/06/2007}+6{1
onth+project×kind+bid×kind+bidding×company+bidding×price}
5/06/2007}+7{15/06/2007}+8{16/06/2007}+9{16/06/2007}+
](1[{kanto+C+p1+15/06/2007+17/07/2007+5+general×construct
10{16/06/2007}+11{16/06/2007}+12{16/06/2007}+13{16/0
ion+general×bid×WTO+C1+800000000}]+2[{kanto+C+p1+15/
6/2007})
06/2007+17/07/2007+5+general×construction+general×bid×WT
O+C2+780000000}])
5.5 Considerations
Next, if you want to search for data for
"construction projects in the Kanto bureau which are When a business application system like the one
not for WTO", you create the condition formula above is developed in the existing way, user
"kanto×!WTO", and get the image of formula 5.3-2 requirements are analyzed first. Next, the system,
by the map h. schemas and application programs are designed
according to requirement analysis. Then,
implementation and testing are done. The
fundamental development process is changed if CDS
is used.
1. Schema Design and Data Input
It is almost impossible for a database designer to
design schema of this application system since
Figure 7: The output result by the condition formula she/he cannot predict the changes in schema of
"(C1+C2) WTO". MLIT bidding results data. And whenever a new file
which schema is different from that already designed
h (formula 5.3-2, kanto×!WTO) appears, it is actually impossible to modify the
= MayOf2007InKanto×code[{bureau×name+project×name+plac schema design and application programs or to
e+bid×date+contract×date+delivery-month+project×kind+bid×
develop data conversion programs. If you employ
kind+bidding×company+bidding×price}](8[{kanto+D+p2+16/0

26
 FLEXIBLE DATA SEARCHS USING CONDITION FORMULAS

CDS in the development of this application system, ER model. In the approach of (N. Routledge, L. Bird
you don’t have to worry about the above problems. and A. Goodchild, 2002: 157-166), the conceptual
This is because the concept of the disjoint union + of and logical levels are represented using a standard
the cellular model is supported in CDS, so that you UML class and the XML represents the physical
can add the data which schema are different to the level. XUML (HongXing Liu, YanSheng Lu, Qing
data storage one after another, if you only have to Yang, 2006: 973-976) is a conceptual model for
convert the data to formulas of CDS. XML schema, based on the UML2 standard. This
application research concerning XML schema is
2. Data Output
needed because there are differences in the
Data output design and application programs for expression capability of the data model between
data output have to be done during application XML and other models. On the other hand, objects
system development, and they have to be modified and their relations in XML schema and the above
when there is a new user requirement for output models can be expressed consistently by CDS,
which was not expected in the user requirement which is based on the cellular model. That is
analysis. This can be costly. But if you use CDS in because the tree structure, on which the XML model
the development, a user only has to create a is based, and the graph structure, on which the UML
condition formula according to a user requirement and ER models are based, is special cases of a
for output. This is because user requirements can be topological structure mathematically. Entity in the
generalized by condition formulas of CDS. models can be expressed as the formula for a cellular
3. Processing Speed space in CDS. Moreover, the relation between
subsets, as we showed in 3.2, cannot in general be
Detailed benchmark tests have not been conducted expressed by XML. Although CDS and the existing
yet, but when we actually tried this system, the deductive database look alike apparently, the two are
output processing speeds of 500 records and 1,000 completely different. The deductive database (Q.
records from more than 200,000 records were 3.2s Kong, G. Chen, 1995: 973-976) raises the
and 6.7s respectively. This system is considered expression capability of the relational database
practical for analyzing business data on a client PC. (RDB) by defining some rules. On the other hand,
CDS is a proposal for a new tool for data
management and has nothing to do with the RDB.
6 RELATED WORKS Plenty of CASE tools are currently available, but
they support system development according to
The distinctive features of our research are the existing data models. The differences from CDS are
application of the concept of topological process, mainly that we apply a novel model, the cellular
which deals with a subset as an element, and that the model, for building CDS, and that the customer side
cellular space extends the topological space, as seen can confirm the output by changing formulas using
in Section 2. Relational OWL as a method of data the defined maps after creating formulas as the input.
and schema representation is useful when
representing the schema and data of a database
(Takashi Washio and Hiroshi Motoda, 2003: 59-68), 7 CONCLUSIONS AND FUTURE
but it is limited to representation of an object that
has attributes. Our method can represent both
WORKS
objects: one that has attributes as a cellular space
and one that doesn’t have them as a set or a In this paper, we have developed a condition
topological space. Many works applying other formula search as an important function of CDS.
models to XML schema have been done. The Using this function of CDS, you can search for data
motives of most of them are similar to ours. The you want from formulas as data storage by creating a
approach in (Giovanna Guerrini, Marco Mesiti, condition formula according to user requirements, so
Daniele Rossi, 2005: 39-44) aims at minimizing that you don’t have to analyze user requirement for
document revalidation in an XML schema evolution, output in typical business application development.
based on a part of the graph theory. The X-Entity The point we should emphasize for future work is
model (Bernadette Farias Lósio, Ana Carolina that the search condition of a user as well as data for
Salgado and Luciano do Rêgo GalvĐo, 2005: 39-44) input/output is expressed as a formula. This certainly
is an extension of the Entity Relationship (ER) brings the system which is developing, including
model and converts XML schema to a schema of the user requirements recursively. This will be
connected to the implementation of a process graph

27
ICE-B 2008 - International Conference on e-Business

(T. L. Kunii, 2003: 86-96). It is the next step where a Toshio Kodama, Tosiyasu L. Kunii, Yoichi Seki, “A New
situation as a node is transferred to the next situation Method for Developing Business Applications: The
selecting a path as an edge. Implementation has been Cellular Data System”, In Proc of CW'06, pp. 65-74,
difficult up to the present time because there is no IEEE Computer Society Press. http://www.mlit.go.jp/
chotatsu/kekka/kekka.html (22 Apr. 2008)
tool to realize it, although one of authors (T.L. Tosiyasu L. Kunii, “What's Wrong with Wrapper
Kunii) outlined the plan many years ago. The Approaches in Modeling Information System
appearance of Formula Expression will enable it in Integration and Interoperability? “, In Proc of
the near future. If we implement the process graph DNIS2003, pp. 86-96, Lecture Notes in Computer
by developing CDS as future work, automation of Science, Springer-Verlag.
business application development will be done. We
believe that CDS brings great social impact,
changing existing development fundamentally. Our
research is still in its infancy, but it is progressing
every day. We are collaborating with companies and
universities worldwide.

REFERENCES
T. L. Kunii and H. S. Kunii, “A Cellular Model for
Information Systems on the Web - Integrating Local
and Global Information”, In Proc. of DANTE'99, IEEE
Computer Society Press, pp.19-24, 1999.
Bernadette Farias Lósio, Ana Carolina Salgado, Luciano
do Rêgo GalvĐo, “Conceptual modeling of XML
schemas”, In Proc. of WIDM'03, ACM Press, pp.102-
105, 2003.
Takashi Washio, Hiroshi Motoda, “State of the art of
graph-based data mining”, In ACM SIGKDD
Explorations Newsletter, ACM Press, pp.59-68, 2003.
David W. Embley, “Toward semantic understanding: an
approach based on information extraction ontologies”,
In Proc. of ADC'04, Australian Computer Society,
Inc., pp.3-12, 2004.
Cristian Pérez de Laborda, Stefan Conrad,
“Relational.OWL: a data and schema representation
format based on OWL”, In Proc. of APCCM '05,
Australian Computer Society, Inc., pp.89-96, 2005.
Nicholas Routledge, Linda Bird, Andrew Goodchild,
“UML and XML schema”, In Proc. of ADC'02,
Australian Computer Society, Inc., pp.157-166, 2002.
Sean K. Bechhofer, Jeremy J. Carroll, “Parsing owl dl:
trees or triples?”, In Proc. of WWW'04, ACM Press,
Inc., pp.266-275, 2004.
Giovanna Guerrini, Marco Mesiti, Daniele Rossi, “Impact
of XML schema evolution on valid documents”, In
Proc. of WIDM'05, ACM Press, pp.39-44, 2005.
Bernadette Farias Lósio, Ana Carolina Salgado, Luciano
do Rêgo GalvĐo, “Conceptual Modeling of XML
Schemas”, In Proc. of WIDM'03, ACM Press, pp.102-
105, 2003.
HongXing Liu, YanSheng Lu, Qing Yang, “XML
conceptual modeling with XUML”, In Proc. of
ICSE'06, ACM Press, pp.973-976, 2006.
Q. Kong, G. Chen, “On deductive databases with
incomplete information”, In ACM TOIS, pp.354-370,
1995.

28
 ANALYZING DECENTRALIZED GOVERNABILITY OF BUSINESS
PROCESSES BY EXTENDED PETRI NETS AND MODAL LOGICS

Takashi Hattori
NTT Communication Science Laboratories, 2-4, Hikaridai, Seika-cho, Keihanna Science City, Kyoto, Japan
takashi [email protected]

Hiroshi Kawakami, Osamu Katai, Takayuki Shiose

Graduate School of Informatics, Kyoto University, Yoshida Honmachi, Sakyo, Kyoto, Japan
{katai, kawakami, shiose}@i.kyoto-u.ac.jp

Keywords: Decentralized Control, Discrete Event System, Petri Net, Modal Logic.

Abstract: We introduce a novel notion of decentralized governance structure of event-driven processes together with
the notions of their behavioral and structural correctness. The ways for attaining correct process behavior, as
well as the notion of decentralized governability, are examined based on temporal logical analyses of process
behavior via Petri net representations of process structures. Also, the deontic and temporal logical prescrip-
tions of normative constraints (tasks) on the processes are introduced that are then translated into extended
hierarchical Petri net structures. The conflicts among these tasks are examined on this hierarchical structure.

1 INTRODUCTION They have been applied to several systems (Black-

burn et al., 2006) and their theoretical studies still
As e-businesses grow, users have gained power in progress (Nute, 2004).
obtaining information and combining various service For further development of e-business, decentral-
applications. For instance, if a user decides to go on ized governance is inevitable. Growing numbers of
a trip, s/he may use an access map, train informa- services and applications may lead to overlap of ser-
tion, a hotel search, and a defrayment service. Thus, vices, which may sometimes cause interference. Our
users are nowadays active, unlike those of the past that framework enables us to check the existence of inter-
only passively followed the line prepared by service ference among services’ control and user’s policies.
providers. In other words, a single service provider is Furthermore, it shows us how to eliminate the inter-
not as powerful now as before, and processes are not ference.
as centralized as before. The rest of this paper consists of the following sec-
This paper proposes a theoretical framework for tions. Section 2 introduces the outline of the frame-
such decentralized multi-agent (in this case, users and work. In this framework, components of a target sys-
service providers) systems that can represent agents’ tem are classified into two portions from the view-
behavior and their policies or control rules. The be- point of whether it is a prefixed structural element or
haviors of agents are represented by a Petri net (Pe- an element that may change over time like users’ poli-
terson, 1981), which offers rich mathematical analy- cies and service providers’ control rules. The former
sis, and introducing modal logics (Hughes and Cress- portion is encoded into a conventional Petri net. The
well, 1968) enables us to represent policies and con- latter half is first represented by modal logic formulae,
trol rules. It is known that a Petri net is conventional which are then translated into extended Petri nets that
and now its successors, e.g., a Unified Modeling Lan- we call “task unit graphs.” The two portions are then
guage (UML) (Saldhana and Shatz, 2000), are popu- integrated into a single extended Petri net. In section
lar tools for system modeling. Applying a Petri net is 3, the framework introduced in section 2 is applied to
still a hot topic for those researchers that place empha- the modeling of decentralized governance problems.
sis on checking the behavior of system design (Hu and Based on the model, we discuss the correctness of
Shatz, 2004). Modal logics, e.g., temporal and deon- process behavior and centralized/decentralized gov-
tic logics, are also known as conventional theories. ernability in sections 4 and 5.

29
ICE-B 2008 - International Conference on e-Business

2 MODELING SYSTEMS BY P3 : A2 is working,

PETRI NET AND MODAL P4 : cache memory of A3 stores a location,
LOGIC P5 : second memory of A3 stores a location,
P6 : A4 is idling,
This section proposes a method for modeling decen- P7 : A4 is working.
tralized systems based on a kind of Petri net and Each of the transitions τ2 and τ4 means a submission
modal logic. Hereafter, we present the procedure of of the location of a hotel or a landmark, τ5 means the
system modeling by using an example of a travel- “data transfer from the cache to the second memory”
ers’ decision on itineraries with the assistance of e- and “flushing cache,” and the firing of τ6 makes A3
applications. search for a train connection between two locations
and give train fees to A4.
Example System. Assume that in the near fu-
ture, many e-applications will work as sophisticated τ1 P2 τ2 P4 P6
agents, and help users by cooperating with each other.
Now a traveler has arrived at Porto station. S/he τ7
enters an internet cafe, and launches the following A1
P1
τ5 τ6
agents: e-landmark map agent (A1), e-hotel search A2 τ3 P3
τ4 P5
P7
(A2), e-train connection information (A3), and an e- A4
defrayment system (A4). A3
S/he has two itineraries to decide on. One is find- agent. territory
ing a hotel to stay at and a way to get there by train. A1 P1 , P2 , τ1 , τ2
The other is finding sightseeing spots and how to ac- A2 P1 , P3 , τ3 , τ4
cess them. In the former mission, s/he first searches A3 P4 , P5 , τ5 , τ6
for a hotel by using A2, which submits the hotel’s lo- A4 P6 , P7 , τ6 , τ7
cation to A3, as well as her/his current location. A3 Figure 1: Petri net representation of a decentralized discrete
determines the route between Porto and the nearest event system.
station to the hotel. After the route is decided, defray-
ment is executed by A4. In the latter mission, s/he When we correspond a place of a C/E system to
decides on two scenic sites s/he wants to visit with a proposition, we can represent the true/false value of
the help of A1, which hands over the location of the the proposition by putting/removing a token in/from
sites to A3 and defrayment is again executed by A4. the place. In this case, each transition leads the value
Anyway, A3 requires two locations and searches for alteration of the proposition. For instance, in Fig. 1,
the optimal route between them. the firing τ6 leads P4 , P5 , P6 to turning from true to
false, and P7 to turning from false to true.
2.1 Petri Net Representation of
System’s Event-Driven Aspect 2.2 Modal Logic Representation of
Tasks
A Petri net is known as a conventional representation
scheme for modeling a physical structure and event- Next, we represent the tasks that state “when the fo-
driven behavior of discrete event systems (Karatke- cused state should be true” as propositions by intro-
vich, 2007). It is also known that a k-bounded stan- ducing temporal and deontic logic.
dard Petri net can be translated into an equivalent 1-
bounded Petri net. We employ a 1-bounded one called 2.2.1 Temporal Modalities
the condition/event system (C/E system) (Reisig,
1982) where a transition can only fire if all “its output A temporal logic is given by the propositional logic,
places” are empty. modal operators, and an axiom system. This paper
For instance, the example described above is the employs the following modal operators:
case where the upper-bound of “the number of stored
locations” is two (2-bounded). Thus it can be mod-
eled as a C/E system as shown in Fig. 1, where a token
in place Pi means the following:
P1 : Both A1 and A2 are idling,
P2 : A1 is working,

30
 ANALYZING DECENTRALIZED GOVERNABILITY OF BUSINESS PROCESSES BY EXTENDED PETRI NETS
AND MODAL LOGICS

T A: A will be true at the next state S1 ,

: A holds
G A: A will be true from now on S0 , S1 , S2 , · · · ,
F A: A is true at S0 , or will be true at some time
in the future Si (i > 0),
AU B: B is true at S0 or A will be true from now GbA FbA
on until the first moment when B will be the
case,
where A, B denote logic formulae, and S0 /Si (i > 0)
mean current/future states (worlds) respectively.
Axiom systems of temporal logic vary depending
on the viewpoint of time. This paper employs one GlA FlA
of the discrete and linear axiom systems KSU (Katai
and Iwai, 1983), which is an extension of the mini- Figure 2: Alethic modes of state transitions.
mal axiom system Kt (Rescher and Urquhart, 1971).
Introducing Y (yesterday) as the mirror image of T
(tomorrow), the axiom system claims that ⊢ T ¬A ≡ A
A A
A : A holds
¬T A, ⊢ Y ¬A ≡ ¬Y A, and ⊢ T Y A ≡ Y T A ≡ A. In- A : B holds
troducing S (since) as the mirror image of U (until), A A
G A ≡ AU ⊥, where ⊥ denotes the contradiction, and A
A
F A ≡ ¬G ¬A, Kt is rewritten as A
A
AUbB AUlB
⊢ AU B ≡ B ∨ (A ∧ T (AU B)), (1)
⊢ AS B ≡ B ∨ (A ∧ Y (AS B)), Figure 3: Branching and linear modes of U .
⊢ {(A ⊃ T (A ∨ B))U C} ⊃ {A ⊃ AU (B ∨C)},
⊢ {(A ⊃ Y (A ∨ B))S C} ⊃ {A ⊃ AS (B ∨C)}, O A: A is obligatory, P A: A is permitted.
where A, B and C denote logic formulae. The axiom system we adopt here for O and P is that
Regarding state transitions in the future, there are of SDL (standard deontic logic), which defines O A ≡
two aspects, i.e., b (branching) and l (linear), thus the ¬P ¬A, and claims ⊢ O A ⊃ P A, and ⊢ O (A ⊃ B) ⊃
modes G and F are more precisely defined as (Katai, (O A ⊃ O B).
1981): Some control rules and specifications of systems
can be translated into the combinations of temporal
G b A: A will necessarily be persistent, and deontic modes by using “translation templates”
G l A: A will possibly be persistent, such as
F b A: A will possibly be the case,
F l A: A will necessarily be the case. O F A: A has to be true in the future,
P G A: A can be always true.
Figure 2 illustrates those modes where each circle de-
notes a state, each arc denotes a state transition, and They correspond to alethic modes F l A and G l A re-
the black circles mean that the state holds A. spectively.
Furthermore, U also can be branching (U b ) or lin-
ear (U l ) as shown in Fig. 3. In the figure, each black 2.3 Network Representation of Tasks
circle means that B is true in that state, and the letter
A means that A is true in that state . Among them, the We translate the task represented by modal logic
following relations are established; G b A ≡ AU b (B ∧ into an extended Petri net, which we call a “task
¬B) ≡ ¬F b ¬A, G l A ≡ AU l (B ∧ ¬B) ≡ ¬F l ¬A. unit graph,” by introducing four types of special arcs
shown in Fig. 4.
2.2.2 Deontic Modalities
(a) Prohibition of firing
Understanding the system’s behavior by temporal
(b) Compulsion of firing at the next step
logic is of an “objective” view of the focused propo-
sition. To represent our “subjective” intention or pur- (c) Compulsion of firing in the future
pose, such as how the propositions should behave, (d) Request of firing synchronization
i.e., the control rule (or task), we introduce deontic
modalities: Figure 4: Special arcs for control of transition firing.

31
ICE-B 2008 - International Conference on e-Business

These arcs are placed from a place to a transition. PP1: defrayment must not be done simultaneously
They function whenever the place holds a token and with submission of a landmark/hotel location;
the transition satisfies the firing condition, but they PP2: hotel search is always followed by a landmark
differ from regular arcs of the conventional Petri net search;
on the following points. First, they do not transfer to-
kens from places to arcs. Next, if there are multiple PP3: submission of either a hotel or a landmark is
special arcs from the same place, all of them are acti- accepted only if the first cache of the train search
vated simultaneously. As a result, simultaneous firing is empty;
of multiple transitions is permitted at the same state. PP4: the system should request a hotel submission
some time in the future before the end of a set of
A free A A free
executions.
Each task associated with each agent is activated
by the firing of the corresponding transition in its ter-
A A A
ritory, e.g., in this case we have the following tasks
represented in temporal deontic logical forms,
(a) OTA (b) OGA (c) OFA
PP1: τ6 activates O 2 (P1 U P6 )
A free B PP1: τ1 or τ3 activates O 4 (P6 U P1 )
PP2: τ4 activates O 2 (¬P3 U P2 )
PP3: τ2 or τ4 activates O 1 (P1 U (¬P4 ))
A B PP4: τ1 activates O 2 F P3
(d) O(AUB) where O i stands for the obligation for agent Ai.
Not every task corresponds to a specific transition.
Figure 5: Examples of task unit graph.
Some tasks are translated into logical forms that are
not activated by a transition but are always activated.
Figure 5 shows examples of task unit graphs, and For example, a rule
these net representations are derived by a systematic
analysis of logical representations of tasks. For in- PP0: Once the defrayment process is finished, agent
stance, consider a task O (AU B). A has to be true from A2 should not submit a hotel location until the
now on until B will be the case. If B is the case at S0 , cache memory flushes its contents to the memory,
this task is accomplished, else if ¬A ∧ ¬B at S0 , this is resident and translated into
task cannot be accepted due to Eq. (1). If A ∧ ¬B at O 2 G ({(¬P3 )U (¬P4 ∧ P5 )}S (¬P7 )).
S0 , A should be maintained and O (AU B) also has to
be the case at S1 . In each case, O (AU B) at S0 pro- Defining Q ≡ ¬P4 ∧ P5 , H ≡ ¬P3 U Q, PP0 can be
hibits the alteration from A to ¬A, so an arc of “prohi- represented as O 2 G (HS(¬P7 )), which derives an ex-
bition of firing” is placed from the place of O (AU B) tended Petri net representation as shown in the left
to the “transition of the alteration from A to ¬A.” part of Fig. 6 (Katai, 1981).
It consists of module nets reflecting its subtasks,
and they are joined with linkage relations prescrib-
ing concurrent (simulations) firing of linked relations
3 DECENTRALIZED SYSTEM shown in Fig. 4 (d). It should be noted that the transi-
STRUCTURE tive closure of these linkage relations links the task in
analysis with the target system, which is shown in the
This paper defines the characteristics of decentralized middle part of Fig. 6. In the figure, task unit graphs
systems as showing S and ∧ (conjunction) are employed. Their
• each agent has its own territory, general types are defined as shown in Fig. 7.
• each control task is given to one of these agents,
• a task can control a transition within the territory
of the agent to whom the task is given. 4 CORRECTNESS OF SYSTEM
Figure 1 is an example of decentralized system. It AND ITS BEHAVIOR
consists of four agents who are in charge of managing
the sub tasks denoted by thick broken circles in Fig. 1. 4.1 Behavioral Corrections
Assume that the traveler has her/his own policy
for using e-agents and adopts them as control rules in A system can be regarded to behave correctly iff it
order to fulfill her/his needs such as: satisfies the following conditions:

32
 ANALYZING DECENTRALIZED GOVERNABILITY OF BUSINESS PROCESSES BY EXTENDED PETRI NETS
AND MODAL LOGICS

Hierarchical Decomposition HS( P7)

of Task O2G(HS( P7))
(HS( P7)) O2G(HS( P7))

H HS( P7) P7
P7 H
P7 H
H P7
(HS( P7))
τ2
τ7
P2 P4 P6
H= P3UQ Q H P3
Target System τ1 τ5 τ6
P1
P7
Q H P3
τ4 τ3 τ3 P3 τ4 P5

P4
Q P5 free P1 free P4
Q= P4 P5 P3
or
A B
P4 Q P5 P3 P1 P4
τ6 τ5
τ5 τ2 O2FP3 O1(P1U P4)

Figure 6: Hierarchical extended Petri net representation of the system with tasks.

ASB correct. Namely, there is no need to control the sys-

B A tem.
A A B B
In terms of temporal or deontic logical expres-
sions, the above is characterized as:
B A
(ASB)
G b (¬C), G b (O i F A ⊃ F l A),
G b {O i (¬(AU B)) ⊃ ¬(AU l B)},
A B
where C is a state of contradiction.
A B We have practically more important and weaker
notions of system correctness as follows:
or
A B Definition 3 (Weak Correctness - Centralized Gov-
(A B) ernability). A system is called weakly correct iff
Figure 7: Task unit graph of S and that of “∧” (conjunction).
we can extend an arbitrary generated state transition
sequence so that it is correct by appropriately execut-
Definition 1 (Correctness of State Transition Se- ing the firing of permitted (legal) transitions This can
quence). A sequence of state transition is correct iff be characterized as
the following cases do not occur along with this infi- G b (¬C), G b (O i F A ⊃ F b A),
nite sequence:
G b {O i (¬(AU B)) ⊃ (¬(AU b B))}.
1. Any proposition A does not occur just after the
state where O i T A (O i G A, O i (AU B)) is given to In terms of the state transition diagram, the current
an agent i. notion of strong correctness can be characterized as:
2. A never happens to the case after the state (until B
is the case) where O i F A (O i (AU B)) is given. Theorem 1 (Strong Correctness of System). A
system is strongly correct iff the following hold:
4.2 System Correctness and its (a) for an arbitrary terminal state of its state transition
Characterization diagram, there is no task associated with it of the
form of O i T A, and if O i F A is present there, then
We have two kinds of system corrections as follows: A is also present on that state, and if O i (¬(AU B))
is present there, then both ¬A and ¬B are also
Definition 2 (Strong Correctness of System). A there;
system is strongly correct iff any state transition se- (b) for an arbitrary cycle (circuit) of its state transi-
quence generated by the system (system behavior) is tion diagram, the following hold:

33
ICE-B 2008 - International Conference on e-Business

(b.1) if O i F A is present at a state in the cycle, 5.2 System Correction in Terms of State
then A is also present at possibly another state Transition Diagram
in the cycle;
(b.2) if O i (¬(AU B)) is present on a state s in the It can be readily seen that the following modifications
cycle, then there is also state s′ such that ¬A on state transition diagrams are necessary for making
holds on s′ and there is no state between s and systems be weakly correct:
s′ at which B holds.
(ii.1) remove the terminal states from the diagram at
For characterizing weak correctness, we intro- which either a task
duce the notion of condensation of directed graphs by O i T A is present, or
“strong components” that are defined as their bidirec-
tionally connected maximal subgraphs (Harary et al., O i F A and ¬A are present, or
1965). O i (¬(AU B)) is present and at least one of ¬A or
¬B is absent;
Theorem 2 (Weak Correctness of System). A sys- (ii.2) remove the terminal strong components, which
tem is weakly correct iff the following hold: include

(c) the same as condition (a) in Theorem 1;

O i F A but ¬A, or
O i (¬(AU B)) is present but ¬A, or
(d) for every terminal strong component of its state
O i (¬(AU B)) at a state s and ¬A at a state s′ such
transition diagram the following hold:
that any path from s to s′ includes a state at
(d.1) if O i F A is present in a state at the compo- which B does not hold.
nent, there is a state (and possibly another) on The above operations on state transition diagrams
which A holds; need to be applied repeatedly because removal of ter-
(d.2) if O i (¬(AU B)) is present in a state s, then minal states or strong components will yield different
there is a state s′ on which ¬A holds and there terminal states and strong components. The opera-
is a path joining s and s′ along which B never tions proceed until there is no need for them. If we
holds. still have remaining states in the diagram, the system
is modified to be weakly correct.

5 DECENTRALIZED An Example of Conflict Detection. The typical

GOVERNABILITY OF conflicts are observed among tasks. Figure 8 shows
the diagram of a portion of the sequence of state tran-
PROCESSES sitions of the target system with the initial state S0,
which holds P1 and P6 , and is in charge of tasks PP0,
5.1 Method of System Correction 1, 2, 3, and 4. Table 1 shows the markings of each
state where “◦” means a normal token and “•” means
From the above results, we will have two ways of an active token, which constraints other tokens.
making an arbitrary system to behave correctly:
S7
τ7 S6
τ6 ’’
S5
τ4 ’
S4
τ4 ’
S5
(i) to make the system strongly correct,
τ1
(ii) first to make the system weakly correct and then τ2 τ5 τ3
S1 S2 S3
τ1
to control it so that its behavior (generated state τ1
transition sequence) becomes correct. S0
τ7 ’
S6
τ6 ’’’
S5
τ2 S4
τ2 S5

In the first approach, there is no more need to control Figure 8: State transition diagram of the system in table 1.
it, i.e., any state transition sequence yielded from it is
surely correct. In the latter approach, weak correct- The terminal state S5 is removed from the diagram
ness itself is merely a precondition on governability by the operation (ii.1) since O F P3 is required but ¬P3
and there is still need for supplementary control on at the state. As a result, there is no “terminal strong
permitted transitions. In other words, weak correct- component” in the diagram and the system becomes
ness guarantees the possibility of this supplementary weakly correct.
control. In this paper we will pursue the latter ap- The conflict in the state S5 can be detected by
proach, which seems to be of more practical impor- tracing synchronized firing linkages (broken lines in
tance than the former. Fig. 9) as mutual interferences among tasks. In state

34
 ANALYZING DECENTRALIZED GOVERNABILITY OF BUSINESS PROCESSES BY EXTENDED PETRI NETS
AND MODAL LOGICS

Table 1: Table of internal states and task states of the system where ◦/• stand for a normal/active tokens.
S0 S1 S2 S3 S4 S4′ S5 S5′ S5′′ S5′′′ S6 S6′ S7
P1 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
P2 ◦ ◦
P3 ◦
P4 ◦ ◦ ◦ ◦ ◦
P5 ◦ ◦ ◦ ◦ ◦ ◦ ◦
P6 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
P7 ◦ ◦
P6 • ◦ • •
P1 ◦
f ree ◦
P6 U P1 • • •
P3 ◦
f ree ◦
F P3 • • • • • • •
P1 • ◦ • • • • ◦ ◦
¬P4 ◦ ◦ ◦
f ree ◦ ◦ ◦
P1 U (¬P4 ) • • • • •
¬P3 • • • •
P2
f ree
(¬P3 )U P2 • • • •
P1 • • ◦
P6 ◦
f ree ◦
P1 U P6 • •
¬P4 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
P5 ◦ ◦ ◦ ◦ ◦ ◦ ◦
Q = ¬P4 ∧ P5 ◦ ◦ ◦
¬P3 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
H = (¬P3 )U Q ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
¬P7 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
H S (¬P7 ) ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦
G (H S (¬P7 )) ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦

S5, place O 1 P1 U (¬P4 ) has a token, which prohibits (ii.2’) prohibit the firing of transitions just before
firing of τ1 and τ3 . On the other hand, the token in (leading to) the removed components.
place O 2 F P3 requests firing of τ3 . Therefore, there is
It should be noted that we need supplementary
a conflict of firing τ3 in state S5. The only way to re-
control actions over weakly correct systems for mak-
solve this conflict is turning P4 to ¬P4 , which leads the
ing their behavior correct. More precisely, by refer-
token in P1 U (¬P4 ) to f ree. But the establishment of
ring to conditions (d.1) and (d.2) in Theorem 2, we
¬H in state S5 prohibits turning P4 to ¬P4 by tracing
need the following operations:
synchronized firing linkages from O G (H S (¬P7 )). As
a result, this conflict cannot be resolved unless ¬H (d.1’) if we arrive at a state where a task of the form
turns to H. O i F A is present, then we must eventually (surely
in the future) arrive at a state where A is realized.
5.3 Derivation of Control Rules (d.2’) if we come to a state where O i (¬(AU B)) is
present, then we must eventually arrive at a state
The above modifications on transition diagrams can where ¬A holds by going through states at which
be translated into control actions on the extended Petri B is not the case.
net systems. The removal of terminal states becomes
(ii.1’) prohibit the firing of transitions just before 5.4 Decentralized Governability
(leading to) the removed states.
The removal of terminal (strong) components is also In the above control operations, we have to consider
translated into the following: the decentralized nature of systems, i.e., each agent

35
ICE-B 2008 - International Conference on e-Business

HS( P7) The hierarchical decomposition of tasks eluci-

(HS( P7)) O2G(HS( P7)) dates the governance (control) structure of tasks over
HS( P7)
the event-driven process described by Petri net sys-
H P7
P7 H tems. The control flows descend the hierarchy, while
P7 H the flows of information reporting the changes of ob-
H P7
(HS( P7)) ject systems ascend the hierarchy. These flows go
or
τ2 along the fire synchronization arcs. Also the conflicts
τ7
Target System P2 P4 P6
τ1 τ5 τ6 between the control flows are elucidated. These anal-
P1
τ3 P3 τ4 P5
P7
yses and methods are expected as a basis for treating
complex business processes that are subject to high
free P3 P1 free P4 reliability and credibility under complicated decen-
P1
tralized governance structures.
P3 P4
O2FP3 O1(P1U P4)

Figure 9: Detection of the conflict at τ3 between tasks

O 2 F P3 and O 1 (P1 U ¬P4 ) where the latter task is trapped REFERENCES
by O 2 G (H S (¬P7 )).
Blackburn, P. et al., editors (2006). Handbook of Modal
having its own territory over which it has control. Logic. Elsevier.
Namely, we should set the following: Harary, F. et al. (1965). Structural Models: An Introduction
to the Theory of Directed Graphs. J. Wiley.
Condition (System Decentralization). Each prohi- Hu, Z. and Shatz, S. M. (2004). Mapping UML diagrams to
bition of firing of a transition (ii.1’) or (ii.2’) must be a Petri net notation for system simulation. Proc. of the
16th Int. Conf. on Software Engineering and Knowl-
caused by a task that is given to an agent whose terri- edge Engineering (SEKE), pages 213–219.
tory includes this transition.
Hughes, G. H. and Cresswell, M. J. (1968). An Introduction
of Modal Logic. Methuen.
Thus, all the prohibition operations along with the Karatkevich, A. (2007). Dynamic Analysis of Petri
course of deriving a weakly correct system should Net-Based Discrete systems, volume LINCIS356.
be subject to this condition. It should be noted that Springer-Verlag.
there may be various ways of deriving weakly correct Katai, O. (1981). Completeness and expressive power of
systems, and only a portion of them may satisfy the nexttime temporal logical system by semantic tableau
above condition. Hence, it is not easy to verify the method. Rapport de Recherche INRIA, 109.
following property of an arbitrarily given decentral- Katai, O. and Iwai, S. (1983). A design method for
ized system. concurrent systems based on step diagram and tense
logic under incompletely specified design criteria
(in Japanese). Systems, Control and Information,
Definition 4 (Decentralized Governability). A 27(6):31–40.
system is called “decentralizedly controllable” if Nute, A. L. D., editor (2004). Deontic Logic in Computer
there exists a sequence of operations (ii.1’) and (ii.2’) Science, LNAI3056. Springer-Verlag.
in which all the prohibitions of transition firing are in Peterson, J. L. (1981). Petri Net Theory and the Modeling
accordance with the above condition on system de- of Systems. Prentice Hall.
centralization. Reisig, W. (1982). Petri Nets. Springer-Verlag.
Rescher, N. and Urquhart, A. (1971). Temporal Logic.
Springer-Verlag.
6 CONCLUSIONS Saldhana, J. and Shatz, S. M. (2000). UML diagrams to
object Petri net models: An approach for modeling
We have introduced the notions of decentralized gov- and analysis. Proc. of the Int. Conf. on Software En-
ernance of event-driven processes, their behavioral gineering and Knowledge Engineering (SEKE), pages
103–110.
and structural correctness, and centralized and de-
centralized control for attaining correct behavior, as
well as that of decentralized governability. Also the
method of deriving correct behavioral processes is
shown. The basic framework we adopted was tempo-
ral and deontic logical analyses of process behavior
and Petri net representations of process structures.

36
 AN ARCHITECTURE FOR DYNAMIC INVARIANT GENERATION
IN WS-BPEL WEB SERVICE COMPOSITIONS

M. Palomo Duarte, A. Garca Domnguez and I. Medina Bulo

Department of Computer Languages and Systems, E.S.I., University of Cádiz
C/ Chile s/n. 11002 Cádiz, Spain
[email protected], [email protected], [email protected]

Keywords: Web Services, service composition, WS-BPEL, white-box testing, dynamic invariant generation.

Abstract: Web services related technologies (especially web services compositions) play now a key role in e-Business
and its future. Languages to compose web services, such as the OASIS WS-BPEL 2.0 standard, open a
vast new field for programming in the large. But they also present a challenge for traditional white-box
testing, due to the inclusion of specific instructions for concurrency, fault compensation or dynamic service
discovery and invocation. Automatic invariant generation has proved to be a successful white-box testing-
based technique to test and improve the quality of traditional imperative programs. This paper proposes a
new architecture to create a framework that dynamically generates likely invariants from the execution of web
services compositions in WS-BPEL to support white-box testing.

1 INTRODUCTION et al., 2001) we use invariant and likely invariant in its

broadest sense: properties which hold always or in a
Web Services (WS) and Service Oriented Architec- specified test suite at a certain program point, respec-
tures (SOA) are, according to many authors, one of tively. We consider that dynamically generating such
the keys to understand e-Business in the early fu- invariants, backed by a good test suite, can become an
ture (Heffner and Fulton, 2007). But as isolated interesting help in WS-BPEL white-box testing.
services by themselves are not usually what cus- We propose a new architecture for a framework to
tomers need, languages to program in the large com- dynamically generate likely invariants of a WS-BPEL
posing WS into more complex ones, like the OA- composition from actual execution logs of a test suite
SIS standard WS-BPEL 2.0 (OASIS, 2007), are be- running on a WS-BPEL engine. These invariants can
coming more and more important for e-Business be an interesting aid for white-box testing the compo-
providers (Curbera et al., 2003). sition.
There are two approaches to program test- The rest of the paper is organized as follows. First,
ing (Bertolino and Marchetti, 2005): black-box test- we explain the particularities of WS compositions
built with WS-BPEL. The next section shows how dy-
ing is only concerned about program inputs and out-
puts, while white-box testing takes into account the namic generation of invariants can succesfully help in
WS composition white-box testing. In the following
internal logic of the program. The latter produces
more refined results, but it requires access to the main section we introduce our proposed architecture,
discuss how to solve some technical problems that
source code. However, WS-BPEL presents a chal-
lenge for traditional white-box testing, due to the in- might arise during its implementation, and show an
example of the expected results. Finally, we compare
clusion of WS-specific instructions to handle concur-
rency, fault compensation or dynamic service discov- our proposal with other alternatives and present some
conclusions and an outline of our future work.
ery and invocation (Bucchiarone et al., 2007).
Automatic invariant generation (Ernst et al., 2001)
has proved to be a successful technique to assist in
white-box testing of programs written in traditional
imperative languages. Let us note that, throughout
this work, to be consistent with related work (Ernst

37
ICE-B 2008 - International Conference on e-Business

2 WS COMPOSITIONS AND But the inherent dynamic nature of these technolo-

WS-BPEL gies also implies new challenges for program testing,
as most traditional white-box testing methodologies
The need to compose several WS to offer higher level cannot be directly applied to this language.
and more complex services to suit customers require-
ments was detected and satisfied by the leading com-
panies in the IT industry with the non-standard spec- 3 WS-BPEL WHITE-BOX
ification BPEL4WS, which was submitted to OASIS
TESTING
in 2003 (OASIS, 2003). Soon after, OASIS created
the WS Business Process Execution Language Techni-
WS composition testing is one of the challenges for
cal Committee to work in its standardization, releas-
ing the first standard specification, WS-BPEL 2.0, in its full adoption in industry in the forthcoming years.
The dynamic nature of WS poses a challenge for test-
2007 (OASIS, 2007).
Standardization has been an important mile- ing, having to cope with aspects like run-time discov-
ery and invocation of new services, concurrency and
stone for WS-BPEL wide adoption by SOA lead-
ing tools, being a key interoperability feature of- fault compensation.
fered by many e-Business systems (Domnguez Jim- Little research has been done on applying white-
nez et al., 2007). This way, companies providing box testing directly on WS-BPEL code. Main ap-
services can take advantage of the new possibilities proaches (Bucchiarone et al., 2007) create simulation
that this programming-in-the-large technology allows models in testing-oriented environments. But simu-
for: concurrency, fault recovery and compensation or lating a WS-BPEL engine is very complex, as there
dynamic composition using loosely coupled services is a wide array of non-trivial features to be imple-
from different providers selected through several cri- mented, such as fault compensation, concurrency or
teria (such as cost, reliability or response time). event handlers. In addition, it is sometimes necessary
WS-BPEL is an XML-based language using XML to translate the code of the WS-BPEL composition to
Schema as its type system. WS-BPEL specifies ser- a second language to check its internal logic.
vice composition logic through XML tags defining In case any of these features is not properly imple-
activities like assignments, loops, message passing or mented, compositions would not be accurately tested.
synchronization. It is independent of the implemen- So we consider that this is an error-prone process, as it
tation and platform of both the service composition is not based on the actual execution of the WS-BPEL
system and the different services used in it. There code in a real environment, that is a WS-BPEL engine
are other W3C-standardized XML-related technolo- invoking actual services.
gies (W3C, 2008) at its foundation: Therefore, we propose using dynamically gener-
ated invariants from actual execution traces as a more
XPath. allows us to query XML documents in a flex- suitable approach.
ible and concise way. Although the latest version
of the language is XPath 2.0, WS-BPEL uses by 3.1 Using Invariants
default XPath 1.0.
SOAP. is an information exchange protocol com- An invariant is a property that holds at a certain
monly used in WS. It is platform independent, point in a program. Classical examples are function
flexible and easy to extend. pre-conditions and post-conditions, that is, assertions
WSDL. is a language for WS interface description, which hold at the beginning and end of a sequence of
detailing the structure every message to be ex- statements. There are also loop invariants, which are
changed, the interfaces and locations of the ser- properties that hold before every iteration and after
vices offered, their bindings to specific protocols, the last one.
etc. Let us illustrate these concepts with a simple ex-
ample. Suppose that we want to sum all the integers
There are several technologies that can extend WS-
from 0 to a given natural n. In this case, we could
BPEL. They are usually referred to as the WS-
define this simple algorithm:
Stack (Papazoglou, 2007). One of the most interesting
is UDDI, a protocol that allows maintaining reposito- 1. r ← 0
ries of WSDL specifications, easing the dynamic dis- 2. From i ← 1 to n:
covery and invocation of WS and the access to their
specifications. UDDI 3.0 is an OASIS Standard (OA- r←i+r
SIS, 2008). 3. Return r

38
 AN ARCHITECTURE FOR DYNAMIC INVARIANT GENERATION IN WS-BPEL WEB SERVICE COMPOSITIONS

This algorithm has the pre-condition n > 0 in step 1, its number and quality is limited due to the inner lim-
as n is natural by definition. Looking at the loop in itations of the formal machinery which analyzes the
step 2, we can tell that the loop invariant r = ∑i−1
j=0 j code, specially in unusual languages like WS-BPEL.
holds before every iteration. Since in step 3 we have Conversely, a dynamic invariant generator (Ernst
exited the loop, we will have i = n + 1, which we can et al., 2001) is a system that reports likely program
substitute in the previous loop invariant to formulate invariants observed on a set of execution log files. It
the post-condition for step 3 and the whole algorithm, includes formal machinery to analyze the information
r = ∑nj=0 j. From this post-condition we can tell that in the logs about the values held by variables at differ-
the algorithm really does what it is intended to do. ent program points, such as the entry and exit points
Manually generated invariants have been success- for functions or loops.
fully used as above to prove the correctness of many The process to generate dynamic invariants is di-
popular algorithms to this day. Nonetheless, their vided into three main steps:
generation can be automated. Automatic invari- 1. An instrumentation step where the original pro-
ant generation has proved to be a successful tech- gram is set up so that, during the later execution
nique to test and improve programs written in tra- step, it generates the execution log files. This step
ditional structured and object-oriented programming is called instrumentation step because the usual
languages (Ernst et al., 2001). way to do it is by adding, at the desired program
Invariants generated from a program have many points, logging instructions. These instructions
applications: write to a file the name and value of the variables
Debugging. An unexpected invariant can highlight a that we want to observe at those points and other-
bug in the code which otherwise might have been wise have no effect on the control and data flow
missed altogether. This includes, for instance, of the process. Sometimes it is also necessary to
function calls with invalid or unexpected param- modify the environment where the program is go-
eter values. ing to be executed.
Program Upgrade Support. Invariants can help de- 2. An execution step in which the instrumented pro-
velopers while upgrading a program. After check- gram will be executed under a test suite. During
ing which invariants should hold in the next ver- each test case an execution log is generated with
sion of the program and which should not, they all the necessary data and program flow informa-
could compare the invariants of the new version tion for later processing.
with those of the original one. Any unexpected 3. An analysis step where formal methods tech-
difference would indicate that a new bug had been niques are applied to obtain invariants of the vari-
introduced. ables logged at the different program points.
Documentation. Important invariants can be added Thus, the dynamic generation of invariants does not
to the documentation of the program, so develop- analyze the code, but a set of samples of the values
ers will be able to read them while working on it. held by variables in certain points of the program.
Verification. We can compare the specification of the Wrong invariants do not necessarily mean bugs in the
program with the actual invariants obtained to see tested program, but rather they might come from an
if they satisfy. incomplete test suite. If the input x is a signed integer
Test Suite Improvement. A wrong likely invariant and we only used positive values as test inputs, we
dynamically generated, as we’ll see in next sec- will probably obtain the invariant x > 0 at some pro-
tion, can demonstrate a deficiency in the test suite gram point. Upon inspection, we would notice that
used to infer it. invariant and improve our test suite to including cases
with x < 0.
3.2 Automatic Invariant Generation
3.3 Dynamic Invariant Generation in
Basically, we there are two approaches when generat- WS-BPEL Compositions
ing invariants automatically: static and dynamic.
Static invariant generators (Bjørner et al., 1997) We consider the dynamic generation of invariants to
are most common: invariants are deduced statically, be a suitable technique to support WS-BPEL compo-
that is, without running the program. To deduce in- sition white-box testing. If we use a good test suite,
variants, its source code is analyzed (specially data all of the complex internal logic of our BPEL compo-
and control flows). On one hand, invariants generated sition (compensation, dynamic discovery of services,
this way are always correct. But, on the other hand, etc.) will be reflected in the log files of the different

39
ICE-B 2008 - International Conference on e-Business

executions, and the generator will infer significant in- 1. Firstly, we could modify the execution environ-
variants. ment itself. We could use an existing open-source
Generally, due to its dynamic nature, the more WS-BPEL engine and modify it in order to pro-
logs we provide the generator, the better results it will duce the log files that we need for any process ex-
produce. In case we obtain seemingly false invariants ecuted in it.
in a first run, we will be able to certify if they were The degree of effort involved would depend on the
due to an incomplete test suite or to actual bugs in a logging capabilities already implemented in the
second run with an improved test suite including ad- engine. Most engines include facilities to track
ditional suitable test cases. process execution flow, but tracking variable val-
Another important benefit is that all the informa- ues is not widely implemented. Of course, modi-
tion in the logs is collected from direct execution of fying the code would take a considerable effort.
the composition code, using no intermediate language
2. Secondly, we could modify the source code of the
of any sort. This way we avoid errors that could arise
WS-BPEL composition to be tested, adding calls
in any translation of the WS-BPEL code or the simu-
to certain logging instructions at the desired pro-
lation of the real-world environment, that is, the WS-
gram points. These instructions would not change
BPEL engine and invoked services.
the behavior of the process, being limited to trans-
An important problem to solve is that usually, all parently inspect and log variable values.
external services will not be available for testing, due
to access restrictions, reliability issues or resource In a similar way to the previous method, we might
constraints. Or it could also just be that we wanted be able to use any existing engine-specific WS-
to define several what-if scenarios with specific re- BPEL logging extension. It could append mes-
sponses from several external WS. Thus, we will also sages to a log file, access a database or invoke an
have to allow for replacing some external services external logging web service, for instance.
with mockups, that is, dummy services which will re- Using this approach we would have to instrument
ply our requests with predefined messages. the two different languages used by the WS-BPEL
standard: the WS-BPEL language itself, which is
XML-based, and the XPath language that is used
to construct complex expressions for conditions,
4 PROPOSED ARCHITECTURE assignments, and so on.
3. And finally, we could implement our own logging
An outline of our proposed architecture for a WS-
XPath extension functions.
BPEL dynamic invariant generator is shown in the
figure 1. These new functions would be called from an
We are using a classical pipeline-based architec- instrumented version of the original WS-BPEL
ture which has, in general terms, three coarse-grained composition source code, and included in exter-
steps corresponding to the three general steps of the nal modules which are reasonably easy to auto-
dynamic invariant generation process which we de- matically plug into most WS-BPEL engines. This
scribed above. is a hybrid approach, as both the composition and
We detail further their WS-BPEL specific issues the engine need to be modified.
below. It is quite likely that using both internal (that is,
inside the WS-BPEL engine) and external mod-
4.1 Instrumentation Step ifications (using new XPath extension functions)
will allow us to obtain more detailed logs with less
In this is the step where we take the original program effort than any of the previous approaches.
and perform the necessary changes on it in order to We also take into account the fact that there are
produce the information that the invariant generator many WS-BPEL engines currently available. Engine-
needs. specific files for the deployment of the composition
In our case, we will take the original WS-BPEL under the selected engine may have to be generated
process composition specification with its dependen- automatically on the fly, to abstract the user from the
cies and automatically instrument it. If necessary we technical details involved.
will create any additional files needed for its execu-
tion in the specific WS-BPEL engine which we will 4.2 Execution Step
be using.
Additional logic to generate the logs that we need In this step, we will take the previously instrumented
can be added in three ways: program and run it under a test suite to obtain the logs

40
 AN ARCHITECTURE FOR DYNAMIC INVARIANT GENERATION IN WS-BPEL WEB SERVICE COMPOSITIONS

Inputs
Test case suite
WS−BPEL process
specification

Instrumentation step
Instrumenter

WS−BPEL
instrumented process

Execution step

WS−BPEL Unit Test Library

SOAP
Director Mockup messages WS−BPEL engine
process server

Process instance
execution logs

Analysis step

Preprocessor

Generator−specific
input files

Invariant
generator

Outputs

Inferred
invariants

Figure 1: Proposed architecture for dynamic invariant generation.

and other information that we need for the next step. in the test suite specification. Each of them will de-
Specifically, we will take our instrumented WS-BPEL fine the initial input message that will cause a new in-
composition and run it under each test case detailed stance of the WS-BPEL process to be started, as well

41
ICE-B 2008 - International Conference on e-Business

as the outputs of the external services required by the 4.3 Analysis Step
WS-BPEL process that we wish to model as mock-
ups. After all the test cases have been executed and logs
The ability to model none, some or even all of the have been collected, it would seem at first glance to
external services as mockups will enable us to obtain be a matter of just handing them to our invariant gen-
invariants reflecting different situations. On one hand, erator.
if we do not use mockups at all, but only invoke ac- However, it will not be so simple in most cases,
tual services, we will be studying the complete WS- because the invariant generator could require certain
BPEL composition in the real-world environment. On additional information about the data to analyze to
the other, if every external service is replaced with a work properly. All information would have to be re-
mockup with predefined responses, we will be focus- formatted according to the input format expected by
ing on the internal logic of the composition itself and the invariant generator. This reformatting could range
how it behaves in certain scenarios. We can also settle from a simple translation to a thorough transforma-
for a middle point in a hybrid approach. tion of the XML data structures to those available to
After this step is done, every test case will have the invariant generator.
generated its own execution log, which we will pass The invariant generator may even accept not only
on to the next step. We will need roughly two compo- logs, but also a list of constraints already known, and
nents to make this possible: which thus do not need to be generated again as in-
• A WS-BPEL engine for running the process itself, variants. This would reduce its output size and make
which will invoke the external services as needed. it easier to understand for its users. To generate this
Making the engine use a mockup for an external constraint list, we would have to analyze the XML
service could be achieved in basically two ways: Schema files contained in the WS-BPEL process and
by modifying the service address included in the all of its dependencies.
WSDL source files, or by creating (or modifying) All these obstacles can be overcome through a
the engine-specific files with the proper values. preprocessor. It could even call the invariant gener-
ator with the generated files to finally obtain the in-
• A WS-BPEL unit test library which will deploy variants of the WS-BPEL composition.
and act both as a client, invoking our composi-
Depending on the number and complexity of the
tion with the desired parameters, and as the exter-
invariants produced by the generator, we could even
nal mockup services for the WS-BPEL process.
need to pass them later to a simplifier. It is a program
These services will behave according to the exter-
based on formal methods that receives a set of invari-
nal test case specification described above.
ants and removes those logically inferred by others.
This unit test library has to be quite more complex
than similar libraries for other languages. It can be 4.4 Example
divided once more into the following subcomponents:
• A director which will prepare and monitor the ex- We comment briefly an example of the invariants we
ecution of the whole test suite according to each could infer in the classical WS-BPEL example of the
test case specification. Ideally, it should also be Loan Approval Service included in the WS-BPEL 2.0
able to deploy and undeploy the WS-BPEL pro- specification (OASIS, 2007).
cess from our selected engine. This WS-BPEL composition receives loan re-
quests from costumers. Each request includes the
• A mockup server, properly set up by the director,
amount and certain personal information. The
which will handle incoming requests and act as
WS-BPEL composition simply notifies the costumer
the external mockup services required by the WS-
whether his loan request has been approved or re-
BPEL process that we choose to model.
jected. The approval of the loan is based on the
Mockups have no internal logic of their own, be- amount requested and the risk that a risk assessment
ing limited to either replying with a predefined WS determines for the costumer according to his per-
XML SOAP message or failing as indicated in the sonal information. If the amount is below $10,000
test case specification. and the risk assessment WS considers the applicant a
There are several lightweight Java-based web low-risk costumer the loan is automatically approved.
servers available for this role, but, if necessary, In case the amount is below the threshold but risk
we believe it would also be feasible to develop it is considered medium or high, the composition in-
from scratch, being a simple URL → SOAP mes- vokes an external loan approval WS, and its answer
sage matching system. is passed to the costumer as the response of the com-

42
 AN ARCHITECTURE FOR DYNAMIC INVARIANT GENERATION IN WS-BPEL WEB SERVICE COMPOSITIONS

position. Finally, in case the requested amount is over Level Agreement. Therefore, it constitutes a black-
the threshold no risk checking is done, and the answer box testing technique. In contrast, our architecture
of the composition is also that of the external loan ap- follows a white-box testing approach, oriented to the
proval WS. generation of the invariants from the internal logic of
Our architecture could infer the following invari- a WS-BPEL composition.
ants for this example, it were backed by an exhaustive The relation between the test cases used for dy-
and high-quality test suite: namically generating invariants and the quality of the
invariants derived is studied in (Gupta, 2003). Aug-
menting a test suite with suitable test cases can be an
$request.amount < 10000 ∧ $risk.level = ’low’ interesting way to increase the accuracy of the invari-
=⇒ $approval.accept = ’yes’ ants inferred by our architecture.
Dynamo (Baresi and Guinea, 2005) is a proxy-
based system to monitor if a WS-BPEL composition
$request.amount < 10000 ∧ $risk.level 6= ’low’ holds several restrictions during its execution. We
=⇒ $approval.accept = response(approver) think it might be useful as a way to check if the in-
variants obtained from our architecture hold while it
$request.amount ≥ 10000 is running in a real-world environment.
Test cases for a WS-BPEL composition are auto-
=⇒ $approval.accept = response(approver) matically generated in (Zheng et al., 2007) according
to state and transition coverage criteria. We could as-
invoke(customer) = $approval sure the quality of the invariants generated by using
them as inputs for our architecture.
We can clearly see that the system could infer that
approval depends on the amount requested and the re-
sponse provided by the approver when invoked. The
system could also detect that the response we provide
6 CONCLUSIONS AND FUTURE
the costumer is always the value of the variable ap- WORK
proval.
Of course, to get results this fine-grained we will The fact that WS are the future of e-Business is al-
need a good test suite. For our example, the test suite ready a given, thanks to their platform independence
would have to contain test cases with amounts over and the abstractions that they provide. The need
and under the threshold (including the limit values to orchestrate them to provide more advanced ser-
$9,999, $10,000 and $10,001). Personal information vices suiting costumer’s requirement has been satis-
causing the risk assessment WS answers both affir- fied through the WS-BPEL 2.0 standard.
matively and negatively must also be provided, spe- However, WS-BPEL compositions are difficult to
cially for those under the threshold. At any rate, as test, since traditional white-box testing techniques are
discussed before, in case we do not obtain the de- difficult to apply to them. This is because of the un-
sired invariants in the first run, we could extend the usual mix of features present in WS-BPEL, such as
test suite with more test cases refining the invariants concurrency support, event handling or fault compen-
obtained. This way, in the next run, we would obtain sation. We have showed how dynamically generated
more accurate invariants. likely invariants backed by a good test suite can be-
come a suitable and successful help to solve these dif-
ficulties, thanks to their being based on actual execu-
tion logs.
5 RELATED WORK In this work, we have proposed a pipelined archi-
tecture for dynamically generating invariants from a
In this section we present some related works. There WS-BPEL composition. Requirements on every of
is a wide variety of topics related to our architecture, the components have been identified, leaving as our
mainly dynamic invariant generation, WS composi- next future work finding suitable systems for each
tion testing and test case generation: one.
An interesting proposal to use dynamically invari- Once the architecture is completely implemented we
ants for WS quality testing is (SeCSE, 2007). It will perform an experimental evaluation of the frame-
collects several invocations and replies from a WS- work under several compositions. This way we will
BPEL composition to an external WS and dynami- test its reliability and evaluate its results through met-
cally generates likely invariants to check its Service rics such as quality of the invariants generated or time

43
ICE-B 2008 - International Conference on e-Business

taken to infer them. Heffner, R. and Fulton, L. (2007). Topic overview: Service-
Looking further ahead, we will later study the re- oriented architecture. Forrester Research, Inc.
lation between the quality of the invariants generated OASIS (2003). OASIS members form web ser-
and the test case suite used to infer them. For this vices business process execution language (WS-
we can use different WS-BPEL compositions with BPEL) technical committee. http://www.oasis-
open.org/news/oasis news 04 29 03.php.
their specifications and different test suites providing
certain coverage criteria (branch coverage, statement OASIS (2007). WS-BPEL 2.0 standard. http://docs.oasis-
coverage, . . . ) of them. open.org/wsbpel/2.0/OS/wsbpel-v2.0-OS.html.
Finally, we could use the invariants generated by OASIS (2008). OASIS standards. http://www.oasis-
our proposal to support WS-BPEL white-box testing open.org/specs/.
and check if it improves its results. Papazoglou, M. (2007). Web services technologies and
standards. computing surveys (enviado para revisin).
SeCSE (2007). A1.D3.3: Testing method
definition V3. http://secse.eng.it/wp-
ACKNOWLEDGEMENTS content/uploads/2007/08/a1d33-testing-method-
definition-v3.pdf.
This work has been financed by the Programa Na- W3C (2008). W3C technical reports and publications.
cional de I+D+I of the Spanish Ministerio de Edu- http://www.w3.org/TR/.
cacin y Ciencia and FEDER funds through SOAQSim Zheng, Y., Zhou, J., and Krause, P. (2007). An automatic
project (TIN2007-67843-C06-04). test case generation framework for web services. Jour-
nal of Software, 2(3):64–77.

REFERENCES
Baresi, L. and Guinea, S. (2005). Dynamo: Dynamic mon-
itoring of WS-BPEL processes. In Benatallah, B.,
Casati, F., and Traverso, P., editors, ICSOC, volume
3826, pages 478–483. Lecture Notes in Computer Sci-
ence, Springer.
Bertolino, A. and Marchetti, E. (2005). A brief essay on
software testing. In Thayer, R. H. and Christensen,
M., editors, Software Engineering, The Development
Process. Wiley-IEEE Computer Society Pr, 3 edition.
Bjørner, N., Browne, A., and Manna, Z. (1997). Automatic
generation of invariants and intermediate assertions.
Theoretical Computer Science, 173(1):49–87.
Bucchiarone, A., Melgratti, H., and Severoni, F. (2007).
Testing service composition. In ASSE: Proceedings of
the 8th Argentine Symposium on Software Engineer-
ing.
Curbera, F., Khalaf, R., Mukhi, N., Tai, S., and Weer-
awarana, S. (2003). The next step in Web Services.
Communications of the ACM, 46(10):29–34.
Domnguez Jimnez, J. J., Estero Botaro, A., Medina Bulo,
I., Palomo Duarte, M., and Palomo Lozano, F. (2007).
El reto de los servicios web para el software libre. In
Proceedings of the FLOSS International Conference
2007, pages 117–132, Jerez de la Frontera. Servicio
de Publicaciones de la Universidad de Cdiz.
Ernst, M. D., Cockrell, J., Griswold, W. G., and Notkin,
D. (2001). Dynamically discovering likely program
invariants to support program evolution. IEEE Trans-
actions on Software Engineering, 27(2):99–123.
Gupta, N. (2003). Generating test data for dynamically dis-
covering likely program invariants. In ICSE, Work-
shop on Dynamic Analysis.

44
 A NEW REINFORCEMENT SCHEME FOR STOCHASTIC
LEARNING AUTOMATA
Application to Automatic Control

Florin Stoica, Emil M. Popa

Computer Science Department, “Lucian Blaga” University, Str. Dr. Ion Ratiu 5-7, Sibiu, Romania
[email protected], [email protected]

Iulian Pah
Department of Sociology, “Babes-Bolyai” University, Bd.21 decembrie 1989, no.128-130, Cluj-Napoca, Romania
[email protected]

Keywords: Stochastic Learning Automata, Reinforcement Learning, Intelligent Vehicle Control, agents.

Abstract: A Learning Automaton is a learning entity that learns the optimal action to use from its set of possible
actions. It does this by performing actions toward an environment and analyzes the resulting response. The
response, being both good and bad, results in behaviour change to the automaton (the automaton will learn
based on this response). This behaviour change is often called reinforcement algorithm. The term stochastic
emphasizes the adaptive nature of the automaton: environment output is stochastically related to the
automaton action. The reinforcement scheme presented in this paper is shown to satisfy all necessary and
sufficient conditions for absolute expediency for a stationary environment. An automaton using this scheme
is guaranteed to „do better” at every time step than at the previous step. Some simulation results are
presented, which prove that our algorithm converges to a solution faster than one previously defined in
(Ünsal, 1999). Using Stochastic Learning Automata techniques, we introduce a decision/control method for
intelligent vehicles, in infrastructure managed architecture. The aim is to design an automata system that can
learn the best possible action based on the data received from on-board sensors or from the localization
system of highway infrastructure. A multi-agent approach is used for effective implementation. Each
vehicle has associated a “driver” agent, hosted on a JADE platform.

1 INTRODUCTION learning system should therefore have the ability to

improve its behavior with time, toward a final goal.
The past and present research on vehicle control The stochastic automaton attempts a solution of
emphasizes the importance of new methodologies in the problem without any information on the optimal
order to obtain stable longitudinal and lateral action (initially, equal probabilities are attached to
control. In this paper, we consider stochastic all the actions). One action is selected at random, the
learning automata as intelligent controller within our response from the environment is observed, action
model for an Intelligent Vehicle Control System. probabilities are updated based on that response, and
An automaton is a machine or control the procedure is repeated. A stochastic automaton
mechanism designed to automatically follow a acting as described to improve its performance is
predetermined sequence of operations or respond to called a learning automaton. The algorithm that
encoded instructions. The term stochastic guarantees the desired learning process is called a
emphasizes the adaptive nature of the automaton we reinforcement scheme (Moody, 2004).
describe here. The automaton described here does Mathematically, the environment is defined by a
not follow predetermined rules, but adapts to triple {α , c, β } where α = {α1 , α 2 ,..., α r } represents
changes in its environment. This adaptation is the a finite set of actions being the input to the
result of the learning process (Barto, 2003). environment, β = {β1 , β 2 } represents a binary
Learning is defined as any permanent change in response set, and c = {c1 , c2 ,..., cr } is a set of penalty
behavior as a result of past experience, and a

45
ICE-B 2008 - International Conference on e-Business

probabilities, where ci is the probability that action 2.2 Absolutely Expedient

α i will result in an unfavorable response. Given that Reinforcement Schemes
β (n) = 0 is a favorable outcome and β (n) = 1 is an
The reinforcement scheme is the basis of the
unfavorable outcome at time instant learning process for learning automata. The general
n (n = 0, 1, 2, ...) , the element ci of c is defined solution for absolutely expedient schemes was found
mathematically by: by (Lakshmivarahan, 1973).
ci = P ( β (n) = 1 | α (n) = α i ) i = 1, 2, ..., r A learning automaton may send its action to
The environment can further be split up in two multiple environments at the same time. In that case,
types, stationary and nonstationary. In a stationary the action of the automaton results in a vector of
environment the penalty probabilities will never responses from environments (or “teachers”). In a
change. In a nonstationary environment the penalties stationary N-teacher environment, if an automaton
will change over time. produced the action α i and the environment
In order to describe the reinforcement schemes, responses are β i j j = 1,..., N at time instant n , then
is defined p (n) , a vector of action probabilities: the vector of action probabilities p (n) is updated as
pi (n) = P (α (n) = α i ), i = 1, r follows (Ünsal, 1999):
Updating action probabilities can be represented
⎡1 N
⎤ r
as follows: pi (n + 1) = pi (n) + ⎢
⎣N
∑β i
k
⎥ ∗ ∑ φ j ( p (n)) −
p(n + 1) = T [ p(n), α (n), β (n)] k =1 ⎦ j =1
j ≠i
where T is a mapping. This formula says the next
⎡ 1 N
⎤ r
action probability p(n + 1) is updated based on the − ⎢1 −
⎣ N
∑β
k =1
i
k
⎥ ∗ ∑ψ j ( p(n))
⎦ j =1 (1)
current probability p(n) , the input from the j ≠i

environment and the resulting action. If p(n + 1) is a ⎡1 N

⎤
linear function of p (n) , the reinforcement scheme is p j (n + 1) = p j (n) − ⎢
⎣N
∑β
k =1
i
k
⎥ ∗ φ j ( p (n)) +
⎦
said to be linear; otherwise it is termed nonlinear.
⎡ 1 N
⎤
+ ⎢1 −
⎣ N
∑β
k =1
i
k
⎥ ∗ψ j ( p (n))
⎦
2 REINFORCEMENT SCHEMES for all j ≠ i where the functions φi and ψ i satisfy
the following conditions:
2.1 Performance Evaluation
φ1 ( p(n)) φ r ( p(n))
= ... = = λ ( p (n)) (2)
Consider a stationary random environment with p1 (n) p r ( n)
penalty probabilities {c1 , c 2 ,..., cr } defined above. ψ 1 ( p (n)) ψ r ( p(n))
= ... = = μ ( p (n))
We define a quantity M (n) as the average p1 (n) p r ( n)
penalty for a given action probability vector: r
r pi (n) + ∑ φ j ( p(n)) > 0 (3)
M ( n) = ∑ c i p i ( n) j =1
j ≠i
i =1
r
An automaton is absolutely expedient if the
pi (n) − ∑ψ j ( p (n)) < 1 (4)
expected value of the average penalty at one j =1
iteration step is less than it was at the previous step j ≠i

for all steps: M (n + 1) < M (n) for all n (Rivero, p j (n) + ψ j ( p (n)) > 0 (5)
2003). p j (n) − φ j ( p(n)) < 1 (6)
The algorithm which we will present in this
for all j ∈ {1,..., r} \ {i}
paper is derived from a nonlinear absolutely
expedient reinforcement scheme presented by The conditions (3)-(6) ensure that
(Ünsal, 1999). 0 < p k < 1, k = 1, r (Stoica, 2007).

Theorem. If the functions λ ( p (n)) and μ ( p (n))

satisfy the following conditions:

46
 A NEW REINFORCEMENT SCHEME FOR STOCHASTIC LEARNING AUTOMATA - Application to Automatic
Control

λ ( p (n)) ≤ 0 4 EXPERIMENTAL RESULTS

μ ( p (n)) ≤ 0 (7)
λ ( p (n)) + μ ( p (n)) < 0 4.1 Problem Formulation
then the automaton with the reinforcement scheme To show that our algorithm converges to a solution
in (1) is absolutely expedient in a stationary faster than the one given in (Ünsal, 1999), let us
environment. consider a simple example. Figure 1 illustrates a grid
The proof of this theorem can be found in (Baba, world in which a robot navigates. Shaded cells
1984). represent barriers.

3 A NEW NONLINEAR
REINFORCEMENT SCHEME
Because the above theorem is also valid for a single- Figure 1: A grid world for robot navigation.
teacher model, we can define a single environment
response that is a function f of many teacher The current position of the robot is marked by a
outputs. circle. Navigation is done using four actions
Thus, we can update the above algorithm as α = {N , S , E ,W } , the actions denoting the four
follows: possible movements along the coordinate directions.
p i (n + 1) = p i (n) + f ∗ (−θ ∗ δ ∗ H (n)) ∗ [1 − p i (n)] − Because in given situation there is a single
optimal action, we stop the execution when the
− (1 − f ) ∗ (−θ ) ∗ [1 − p i (n)]
probability of the optimal action reaches a certain
p j (n + 1) = p j (n) − f ∗ (−θ ∗ δ ∗ H (n)) ∗ value (0.9999).
(8)
∗ p j (n) + (1 − f ) ∗ (−θ ) ∗ p j (n)
4.2 Comparative Results
for all j ≠ i , i.e.:
ψ k ( p(n)) = −θ ∗ p k (n) We compared two reinforcement schemes using
φ k ( p(n)) = −θ ∗ δ ∗ H (n) ∗ p k (n) these four actions and two different initial
conditions.
where learning parameters θ and δ are real values
which satisfy:
Table 1: Convergence rates for a single optimal action of a
0 < θ < 1 and 0 < θ ∗ δ < 1 . 4-action automaton (200 runs for each parameter set).
The function H is defined as follows:
Average number of steps to reach
⎧ p i ( n)
H ( n) = min{1; max{min ⎨ − ε, popt=0.9999
⎩θ ∗ δ ∗ (1 − pi (n)) 4 actions with 4 actions with
pi (0) = 1 / 4, popt (0) = 0.0005,
⎫
⎛ 1 − p j ( n) ⎞ ⎪ pi ≠ opt = 0.9995 / 3
⎜ −ε ⎟ ; 0}} i = 1,4
⎜ θ ∗ δ ∗ p j ( n) ⎟ j =1, r ⎬
⎝ ⎠ j ≠i ⎪
⎭ θ δ Ünsal’s New Ünsal’s New
Alg. alg. Alg. alg.
Parameter ε is an arbitrarily small positive real
number. 1 644.84 633.96 921.20 905.18
Our reinforcement scheme differs from the one 0.01 25 62.23 56.64 205.56 194.08
given in (Ünsal, 1999) by the definition of these 50 11.13 8.73 351.67 340.27
two functions: H and φ k . 1 136.99 130.41 202.96 198.25
0.05 5 74.05 63.93 88.39 79.19
The proof that all the conditions of the 10 24.74 20.09 103.21 92.83
reinforcement scheme (1) and theorem (7) are 1 70.81 63.09 105.12 99.20
satisfied can be found in (Stoica, 2007). 0.1 2.5 59.48 50.52 71.77 65.49
In conclusion, we state the algorithm given in 5 23.05 19.51 59.06 54.08
equations (8) is absolutely expedient in a stationary
environment. The data shown in Table 1 are the results of two
different initial conditions where in first case all

47
ICE-B 2008 - International Conference on e-Business

probabilities are initially the same and in second

SLA
case the optimal action initially has a small
Frontal Environment
probability value (0.0005), with only one action Detection β1
receiving reward (i.e., optimal action). F1 Longitudina α1
Comparing values from corresponding columns, l
we conclude that our algorithm converges to a Left
solution faster than the one given in (Ünsal, 1999). Detection

Right
α2
5 USING STOCHASTIC Detection β2 Lateral
F2 Automaton
LEARNING AUTOMATA FOR
INTELLIGENT VEHICLE Speed
Detection
CONTROL Planning Layer

Localization
The task of creating intelligent systems that we can System Highway
rely on is not trivial. In this section, we present a
method for intelligent vehicle control, having as
theoretical background Stochastic Learning Auto
Automata. We visualize the planning layer of an yes α1 vehicle

action
α1
intelligent vehicle as an automaton (or automata Regulation
group) in a nonstationary environment. We attempt β1 = 0 Buffer
to find a way to make intelligent decisions here,
having as objectives conformance with traffic no ─
parameters imposed by the highway infrastructure
(management system and global control), and yes α2
improved safety by minimizing crash risk. α2
Regulation
The aim here is to design an automata system β2 = 0 Buffer
that can learn the best possible action based on the
data received from on-board sensors, of from no ─
roadside-to-vehicle communications. For our model,
we assume that an intelligent vehicle is capable of
two sets of lateral and longitudinal actions. Lateral
actions are LEFT (shift to left lane), RIGHT (shift to Physical Environment
right lane) and LINE_OK (stay in current lane).
Figure 2: The model of the Intelligent Vehicle Control
Longitudinal actions are ACC (accelerate), DEC System.
(decelerate) and SPEED_OK (keep current speed).
An autonomous vehicle must be able to “sense” the
The response from physical environment is a
environment around itself. Therefore, we assume
combination of outputs from the sensor modules.
that there are four different sensors modules on
Because an input parameter for the decision blocks
board the vehicle (the headway module, two side
is the action chosen by the stochastic automaton, is
modules and a speed module), in order to detect the
necessary to use two distinct functions F1 and F2
presence of a vehicle traveling in front of the vehicle
or in the immediately adjacent lane and to know the for mapping the outputs of decision blocks in inputs
current speed of the vehicle. for the two learning automata, namely the
These sensor modules evaluate the information longitudinal automaton and respectively the lateral
received from the on-board sensors or from the automaton.
highway infrastructure in the light of the current After updating the action probability vectors in
automata actions, and send a response to the both learning automata, using the nonlinear
automata. Our basic model for planning and reinforcement scheme presented in section 3, the
coordination of lane changing and speed control is outputs from stochastic automata are transmitted to
shown in Figure 2. the regulation layer. The regulation layer handles
the actions received from the two automata in a
distinct manner, using for each of them a regulation

48
 A NEW REINFORCEMENT SCHEME FOR STOCHASTIC LEARNING AUTOMATA - Application to Automatic
Control

buffer. If an action received was rewarded, it will be distance (< admissible distance), a penalty response
introduced in the regulation buffer of the is sent to the automaton for actions LINE_OK,
corresponding automaton, else in buffer will be SPEED_OK and ACC. All other actions (LEFT,
introduced a certain value which denotes a penalized RIGHT, DEC) are encouraged, because they may
action by the physical environment. The regulation serve to avoid a collision.
layer does not carry out the action chosen The Speed Module compares the actual speed
immediately; instead, it carries out an action only if with the desired speed, and based on the action
it is recommended k times consecutively by the choosed send a feedback to the longitudinal
automaton, where k is the length of the regulation automaton.
buffer. After an action is executed, the action
Table 4: Outputs from the Speed Module.
1
probability vector is initialized to , where r is the
r Speed Sensor Module
number of actions. When an action is executed, Speed: Acceptable Speed:
Actions
regulation buffer is initialized also. too slow speed too fast
SPEED_OK 1 0 1
ACC 0 0 1
DEC 1 0 0
6 SENSOR MODULES
The reward response indicated by 0* (from the
The four teacher modules mentioned above are Headway Sensor Module) is different than the
decision blocks that calculate the response normal reward response, indicated by 0: this reward
(reward/penalty), based on the last chosen action of response has a higher priority and must override a
automaton. Table 2 describes the output of decision possible penalty from other modules.
blocks for side sensors.

Table 2: Outputs from the Left/Right Sensor Module. 7 A MULTI-AGENT SYSTEM

Left/Right Sensor Module FOR INTELLIGENT VEHICLE
Vehicle in sensor
No vehicle in CONTROL
sensor range
Actions range or no adjacent
and adjacent
lane
lane exists
In this section is described an implementation of a
LINE_OK 0/0 0/0 simulator for the Intelligent Vehicle Control System,
LEFT 1/0 0/0 in a multi-agent approach. The entire system was
RIGHT 0/1 0/0 implemented in Java, and is based on JADE
platform (Bigus, 2001).
Table 3: Outputs from the Headway Module. In figure 3 is showed the class diagram of the
simulator. Each vehicle has associated a JADE agent
Headway Sensor Module (DriverAgent), responsible for the intelligent control.
Vehicle in range “Driving” means a continuous learning process,
No vehicle
Actions (at a close frontal
in range sustained by the two stochastic learning automata,
distance)
namely the longitudinal automaton and respectively
LINE_OK 1 0
LEFT 0 0 the lateral automaton.
RIGHT 0 0 The response of the physical environment is a
SPEED_OK 1 0 combination of the outputs of all four sensor
ACC 1 0 modules. The implementation of this combination
DEC 0* 0 for each automaton (longitudinal respectively
lateral) is showed in figure 4 (the value 0* was
As seen in Table 2, a penalty response is substituted by 2).
received from the left sensor module when the
action is LEFT and there is a vehicle in the left or
the vehicle is already traveling on the leftmost lane.
There is a similar situation for the right sensor
module.
The Headway (Frontal) Module is defined as
shown in Table 3. If there is a vehicle at a close

49
ICE-B 2008 - International Conference on e-Business

found to converge to the ”optimal” action faster than

nonlinear schemes previously defined in (Ünsal,
1999).
Using this new reinforcement scheme was
developed a simulator for an Intelligent Vehicle
Control System, in a multi-agent approach. The
entire system was implemented in Java, and is based
on JADE platform

REFERENCES
Baba, N., 1984. New Topics in Learning Automata:
Theory and Applications, Lecture Notes in Control
and Information Sciences Berlin, Germany: Springer-
Verlag.
Barto, A., Mahadevan, S., 2003. Recent advances in
hierarchical reinforcement learning, Discrete-Event
Systems journal, Special issue on Reinforcement
Figure 3: The class diagram of the simulator. Learning.
Bigus, J. P., Bigus, J., 2001. Constructing Intelligent
// Longitudinal Automaton Agents using Java, 2nd ed., John Wiley & Sons, Inc.
public double reward(int action){ Buffet, O., Dutech, A., Charpillet, F., 2001. Incremental
int combine; reinforcement learning for designing multi-agent
combine=Math.max(speedModule(action), systems, In J. P. Müller, E. Andre, S. Sen, and C.
frontModule(action)); Frasson, editors, Proceedings of the Fifth
if (combine = = 2) combine = 0; International Conference onAutonomous Agents, pp.
return combine; 31–32, Montreal, Canada, ACM Press.
} Lakshmivarahan, S., Thathachar, M.A.L., 1973.
// Lateral Automaton Absolutely Expedient Learning Algorithms for
public double reward(int action){ Stochastic Automata, IEEE Transactions on Systems,
int combine; Man and Cybernetics, vol. SMC-6, pp. 281-286.
combine=Math.max( Moody, J., Liu, Y., Saffell, M., Youn, K., 2004. Stochastic
leftRightModule(action), direct reinforcement: Application to simple games
frontModule(action)); with recurrence, In Proceedings of Artificial
return combine; Multiagent Learning. Papers from the 2004 AAAI Fall
} Symposium,Technical Report FS-04-02.
Figure 4: The physical environment response. Narendra, K. S., Thathachar, M. A. L., 1989. Learning
Automata: an introduction, Prentice-Hall.
Rivero, C., 2003. Characterization of the absolutely
expedient learning algorithms for stochastic automata
8 CONCLUSIONS in a non-discrete space of actions, ESANN'2003
proceedings - European Symposium on Artificial
Reinforcement learning has attracted rapidly Neural Networks Bruges (Belgium), ISBN 2-930307-
03-X, pp. 307-312
increasing interest in the machine learning and Stoica, F., Popa, E. M., 2007. An Absolutely Expedient
artificial intelligence communities. Its promise is Learning Algorithm for Stochastic Automata, WSEAS
beguiling - a way of programming agents by reward Transactions on Computers, Issue 2, Volume 6, ISSN
and punishment without needing to specify how the 1109-2750, pp. 229-235.
task (i.e., behavior) is to be achieved. Reinforcement Sutton, R., Barto, A., 1998. Reinforcement learning: An
learning allows, at least in principle, to bypass the introduction, MIT-press, Cambridge, MA.
problems of building an explicit model of the Ünsal, C., Kachroo, P., Bay, J. S., 1999. Multiple
behavior to be synthesized and its counterpart, a Stochastic Learning Automata for Vehicle Path
meaningful learning base (supervised learning). Control in an Automated Highway System, IEEE
Transactions on Systems, Man, and Cybernetics -part
The reinforcement scheme presented in this A: systems and humans, vol. 29, no. 1, january 1999.
paper satisfies all necessary and sufficient conditions
for absolute expediency in a stationary environment
and the nonlinear algorithm based on this scheme is

50
 BUSINESS INTELLIGENCE THROUGH REAL-TIME TRACKING
Using a Location System Towards Behaviour Pattern Extraction

Pedro Abreu, Vasco Vinhas

FEUP - Faculty of Engineering of the University of Porto, R. Campo Alegre 1021, Porto, Portugal
DEI - Department of Informatics Engineering, R. Campo Alegre 1021, Porto, Portugal
LIACC - Artificial Intelligence and Computer Science Laboratory, R. Campo Alegre 1021, Porto, Portugal
[email protected], [email protected]

Pedro Mendes
FEUP - Faculty of Engineering of the University of Porto, R. Campo Alegre 1021, Porto, Portugal
DEI - Department of Informatics Engineering, R. Campo Alegre 1021, Porto, Portugal
INESC Porto - Institute for Systems and Computer Engineering of Porto, Rua Dr. Roberto Frias, 378, Porto, Portugal
[email protected]

Keywords: Location System, Real-Time, Knowledge Extraction, Business Intelligence.

Abstract: Nowadays, tracking systems constitute an important knowledge support in order to compute important mea-
surements in companys processes efficiency. As consequence of that, this project proposes a methodology and
an application, based on a tracking system to obtain, by automatic means, dynamic location data on items.
This solution assumes that the client carries or drives an item of some kind. In each item there is an identify-
ing tag attached and hidden in order to make the item at hand detectable by all the sensors that are scattered
around the area. Because of the fact that the tag is light and hidden and also has no information regarding
the specific person/agent this process is completely transparent to the client or robot that is being implicitly
tracked. This system produces real-time shop floor visualization maps with intelligible data on online item
localization; individual item complete path routes; online and historical population density rates and path
routes concentration; and also item vision enabled concentration maps as emulation for item omnidirectional
vision considering occlusions. This proposed system might be useful in many different areas, for instance in
a traditional retail environment tracing clients through a commercial area or enabling item tracking and route
analysis in a hospital.

1 INTRODUCTION presents a methodology and a tool, based on a track-

ing system to obtain by automatic means, movements
Detecting behavioral patterns is a challenging task data on these elements.
that marketing and distribution companies face. The The presented solution assumes that the client car-
issue has been addressed through the past years on ries or drives an item of some kind, inside the space.
several perspectives like deterministic psychology It is also assumed that these items are outside the en-
(Luce, 1999; Choustova, 2007). However due to their try of the traceable space, although already inside of
active consciousness, human beings are extremely un- its admissible space. In each of the items there is an
predictable and so these methods failed to provide any identifying tag attached and hidden in order to make
accurate data that could be used for industrial pur- the item at hand detectable by all the radars that are
poses. Having these approaches been unsuccessful, scattered around the area. By detecting it, the client
statistical inference with large data sets is still one of or robot is being implicitly tracked in a completely
the most powerful tools available. transparent way for him/it since the tag is light and
Nowadays, tracking systems can represent a pow- hidden and also has no information regarding the spe-
erful tool to support monitoring activity. With these cific person/agent.
systems, some performance measurements on com- Several benefits can be withdrawn from using a
pany process efficiency can be obtained regardless system such as this. Instantaneously one could use it
of the specific tracked target. This research work to monitor the traceable area in a more effective way

51
ICE-B 2008 - International Conference on e-Business

than looking at dozens of screens with images from this goal, several technologies have been used. By
security cameras. Live monitoring of the elements’ doing a brief comparison, it is observable that all of
positions on a specific area allows managers to iden- them have their strengths and flaws concerning char-
tify congested sectors. It also allows him to identify acteristics like the cost in terms of initial investment
hot and cold zones which may be related to the inter- and maintenance. There are others related to environ-
ests points at hand or, for instance, to a local security ment specificities. Among these last, other parame-
issue or other type of event such a medical emergence. ters such as coverable area, tracking detection errors
Regarding long term data analysis the advantages and occlusion problems should also be considered.
of using such a system are several and may be more One of the most effective technologies is also one
than the ones presented next. By analyzing all the of the most expensive ones and concerns detection of
paths taken by the elements, it is possible to obtain the thermal signatures. This technique is appropriate to
hot zones on any time frame and thus evaluate the suc- living organisms which emit particular heat waves.
cess rate of a given promotional campaign and among One particular application of this technology is the
other things. Erroneous and random movements may monitoring of the fauna in the ocean (Raizer, 2003).
also be correlated with a security issue, and thus this Another interesting technology is Bluetooth because
system could also represent an interesting addition to most modern mobile equipments are prepared to send
conventional security systems. The results later ob- a receive data though this protocol. Although the
tained would clearly point out the success of the lay- initial investment is low the coverable area is not
out redesign. Of course these last measures imply cor- very wide and battery consumption is high, in relative
relating client positions with goods bought by him. terms (Jappinen P., 2007).
The paper is structured as follows: section 2 de- The cheapest solution is infrared based. Even
scribes the current state of the art regarding the sev- though its price attractiveness, infrared systems tend
eral areas of knowledge involved in the development to fail on most real environments because the signal
of the system described in this paper. Section 3 de- is unable to reach the target if there is an opaque ob-
scribes the tool that simulates the environment as well ject between the receiver and the target (Krotosky J.,
as its architecture and core functionalities. Section 4 2007).
discusses the results obtained so far and the next sec-
Two of the most emergent technologies for track-
tion concludes the paper by summarizing the focus
ing are RFID and Wi-Fi based. The first one still lacks
of this research work and pointing out future lines of
standardization which is somehow reflected in the
discussion.
pricing of both receivers and transmitters. It is based
on high frequency radio waves having the detectable
tags a passive or an active response. Passive tags are
2 STATE-OF-THE-ART only detectable on a 13 meter radius and are used
for instance on the new USA passports. Active tags,
Nowadays, tracking systems represent an important alternatively, are detectable on a much wider range
research area as their applications are transversal to but are more expensive (around 400%) essentially be-
several areas of knowledge (e.g computer science, cause the tags require an independent power supply
medicine, simulation, robotics as well as industrial (Chao C., 2007). Wi-Fi may also be considered as a
tracks). In the past few years, technology has evolved tracking technology. This approach is mainly used for
in order to provide more accurate measurements. In creating wireless computer networks but in this case
the robotics area, for better modeling the world, it is the involved tracking only requires the usage of the
extremely relevant to accurately process the signals low level protocols. This type of solution is interest-
received by the multiple sensors involved. Locating ing because it makes possible reusing existent com-
objects of the real world to the modeled one is a criti- puter networks for other proposes and takes advan-
cal task for the appliance of the navigation algorithms tages from possible simple detection with at least one
and methodologies. Following these advances the access point. With only one access point the system’s
work published by Hyunwoong Park (Park H., 2006) precision may not be very high but there is no need
presents a new kind of sensor system with multiple for triangulation. Occlusion problems and signals
rotating range sensors. Such system allows a robot losses, with the use of this technique will be reduced
to guide itself on a priori unknown world. On the to a residual level in both open spaces and indoors
other hand these tracking systems also find interesting – considering that indoor spaces do not have signifi-
applications on scenarios where the context environ- cant metal structures within the walls) (Mingkhwan,
ment is already known. Regarding this last system, 2006).
locating elements assumes a crucial role. To achieve Another area that suffered several developments

52
 BUSINESS INTELLIGENCE THROUGH REAL-TIME TRACKING - Using a Location System Towards Behaviour
Pattern Extraction

in the past few years concerns world modeling. In

this research area there are relevant research topics,
most of them related to computer graphics. Most
of the current advances focus on three dimensional
(3D) worlds. In this scope, the evolution on com-
puter graphics is the most notorious. Nowadays, sim-
ple systems are able to represent complex 3D world
including high resolution textures, detail animation
(Vazquez, 2007) and weather condition (Grudzinski,
2007). It is even possible to recreate a 3D world from
textual specifications (Moura J., 2004). On top of
3D world, many algorithms are applicable in other to
optimize rendering performance and obtaining world
data such as visible objects of a certain point.
Figure 1: System’s Architecture.
These algorithms are too complex for most real-
time tracking systems and therefore for this research In the first stage, using a location system that might
work the world is assumed to be a two and a half di- be instanciated in a RFID or Wi-Fi based solution,
mension one (2,5D). In this scenario, a map is rep- covering an area of for instance a large open space or
resented considering a bird’s eye view and assuming building, with a maximum error of one meter, move-
that the height of the objects has no maximum value. ment data is collected regarding the tagged elements
In these conceptions it is simpler to obtain the set present in the given specific floor. In order to gather
of potentially visible objects from a certain point us- the location data one agent has been developed to col-
ing a portal culling algorithms; that consider walls as lect all the positions. This action has a given periodic-
complete occluders and assigns a vision probability to ity and is dependent on the location engine. Typically,
each region in the map (Pires, 2001). Such is achieved this collected data is guaranteed to be obtained every
by dividing the indoor space into separate sectors and second at most for every single tag, although this fig-
then portals which represent the breaches between the ure might be decreased depending on the number of
sectors. By drawing cones that connect the observer simultaneously trackable objects.
point and both extremities of a given portal, one can
After the collection process the agent sends the in-
obtain the areas where all the objects are potentially
formation to the server application. Prior to this ac-
visible. Some caution is required when performing
tion, the agent executes a simple, yet efficient, data
such an operation, since the lines that represent the vi-
validation that is based on the map of the structure
sion cone cannot intersect the ones representing zone
that is sent by the server before the collector agent
divisions. It is also relevant to state that one must as-
boot process.
sume the observer’s vision direction is the center of a
given ”vision cone”. The server application, before being able to re-
ceive any position data, must load the floor map that
will contain the trackable objects, from a XML file.
The maps are modeled as 2,5 D worlds and include
3 PROJECT DESCRIPTION several structures that can be easily adapted to many
types of spaces. Before using it, the server validates
The project description is divided of three distinct the map against a XML Schema.
subsections. In the first denominated as Project Ar- For instance, a XML file representing a traditional
chitecture a description of the system’s architecture retail shop includes entry areas and an exit one that
is depicted. In the Real Time Tracking Visualization in this case is designated as a payment area. There
and Concentration Maps subsection the principles of are also walls that have infinitive height positioned
the system features is explained and in the last sec- around the map. In the supermarket example these
tion, Client Vision Module the used vision algorithm last are named as shelves as can be seen on Figure
is exposed. 2. The proposed XML structure allows specifying a
color and a designation for each half of the wall. The
3.1 Project Architecture half is determined by the largest dimension, and in
case of a square it is assumed to be vertically alligned
This research work proposes a decentralized architec- as can be seen on Figure 2.
ture and prototype tool following that same principle For each set of received coordinates, the server
as detailed in Figure 1. stores them in a database for universality sake. The

53
ICE-B 2008 - International Conference on e-Business

sitioning tags on the map this transformation is also

applied to their centre positioning.
The simplest view allows representing online the
tags in their actual positions. Other views are ob-
tained through the server’s knowledgment extraction
features. The Zone Matrix consists in determining
in real-time which are the most and least populated
zones, denominated as hot and cold zones.
It is also possible to consult this data on a wider
time frame considering the same space with or with-
out the same layout. This last feature requires
database access. The zones are automatically ob-
tained by dividing the space into a grid with flexi-
ble dynamic resolution. This dynamic division allows
both a more in depth study of the hot and cold zones
and also a less detailed one in order to study, for ex-
ample, the spaces quadrants occupation.
Several other results are obtainable by accessing
memory-based data structures for limited time frame
analysis and by querying the database in similar
modes. It is possible to obtain historical client paths,
the shop areas walls, shelves or objects that were
more observed by the clients.
Figure 2: Drawn Map with Corresponding XML File Struc- In order to reproduce historical client path recog-
ture.
nition all the clients coordinate are stored into the
system executes it in a completely completely agnos- database with a timestamp that is related to a given
tic way concerning the database provider. At this map. Figure 3 exposes the paths taken by several
point it is also relevant to state that the server is a clients in a given time frame.
multi-threaded application – having a thread per tag.
Each thread writes into a reserved memory location
the tag’s current position. The previous position is
not overwritten; instead it is stored in the database.
Each thread includes some recovery proceedings such
as deleting all the records of a specific tag in a given
time frame if it stops transmitting its location for a
long period of time regarding the context at hand.
As each thread executes, the server GUI is able to
display online data. This data represents where the
tags are located on the map. Other types of views in-
volve processing the data in real-time, using computer
memory and/or by consulting the database registers
for further in depth reports (section 3.2).

3.2 Real-Time Tracking Visualization Figure 3: Paths taken by the Elements.

and Concentration Maps
The Server GUI includes several different views of 3.3 Client Vision Module
both the online and historical data. In all these views
there is a visual representation of the map. In order The client vision module uses a simplified version of
to draw a map the server requires to systematically the occlusion portal culling detection algorithm. This
perform a scale transformation involving real world simplified version discards all zones that have a low
coordinates and pixel coordinates. This kind of trans- probability of being seen and also disregards orienta-
formation must be dynamic because, in any instance, tion and assumes that the observer can see simultane-
the GUI can be resized. When the view involves po- ously north, south, east and west directions. For each

54
 BUSINESS INTELLIGENCE THROUGH REAL-TIME TRACKING - Using a Location System Towards Behaviour
Pattern Extraction

of these directions the observer throws a vision cone reported zero seconds. The results showed that for ten
having its center coincident with the direction at hand. items, the differences between low-end and high-end
The first intersected walls are considered to be vis- computers is absolutely negligible. For one hundred
ible, and the others invisible. The Figure 4 summa- items, configuration B needs twice the time of con-
rizes with descriptive colors which were the shelves figuration A but real-time visualization is not jeop-
that were most observed by the moving targets. erdized in anyway. In both cases, for this scale there
was not registered any flicker effect and the process
time was compatible with a real-time system. Only if
the scale is pushed to one thousand, configuration B
takes three point seven seconds to compute and even
configuration A takes two point three seconds. These
figures show that for this kind of scale it is needed
a high-end computer system – even if one consider
the traditional consumer market products – and hard
real-time requirements are not met but one might still
assume near real-time features that are perfectly ad-
equated for this kind of management/monitoring sys-
tems.

Configuration A Configuration B
CPU Time (ms) N/A N/A
1
Flicker N N
Number of Items

CPU Time (ms) 15 23
10
Figure 4: Visible Walls with Historical Data. Flicker N N
CPU Time (ms) 125 215
100
Flicker N N
CPU Time (ms) 2300 3700
1000
Flicker Y Y
4 RESULTS
Figure 5: Simulation Performance Benchmarking.
In this section, the project’s results will be de-
picted taking into account three generic levels: sys- Regarding the first aspect, all the enunciated pre-
tem’s main features; simulation statistics extraction dicted functionalities, thecnically described in the
and global aplication; and architecture stability and previuos section, were successfully implemented and
feasability. fully tested. As illustrated in Figure 6, it is possi-
For simulation purposes, there had been consid- ble to visualise in real-time the location of up to one
ered two standard computer configurations: a high- thousand items overlapped with the shop floor layout.
end machine with 4GB of RAM, an Intel Core 2 Duo This number of items can be increased but it is de-
E820 CPU and a SATA II 320 GB 16 MB cache hard pendent of the external location system’s features. In
disc denominated as configuration A, and a low-end the same illustration, it is visible the tool’s flexibility
configuration, denominated as B, equiped with a 2 GB in what concerns to layout management and design
of RAM, an Intel Pentium D 3.00GHz and a SATA II as all shop floor static structures are fully defined and
250 GB 8MB cache hard disc. Both systems were described through a simple, yet flexible XML config-
equiped with Windows Vista Ultimate and the simu- uration file. With this approach, it is possible to model
lations were performed with similar workload condi- heterogeneous environments and, therefore, apply the
tions. In Figure 5 the experiment’s results are fully proposed system to several domains.
depicted as for both hardware configurations four dif- In spite of the importance of the mentioned func-
ferent scenarios were simulated. For each one it was tionalities, the greated added value resides in the
recorded the CPU time needed to perform the most knowledge extraction extendend features. Having in
demanding task – real-time dynamic grid concentra- consideration the online item location gathering, the
tion levels with memory-based historical data – and system is able of real-time item path reconstruction
the presence of absence of image flicker, with a dif- and visualization operating both in memmory-based
ferent number of tracked items ranging from a single or database access, depending on data dimension.
one to one thousand. Conducted experiments showed that real-time mem-
As previous note, one shall point that for single mory access is feasible using a low-end computer –
item tracking, the measured CPU time for both con- with 2 GB of RAM – for tracking one hundred items
figurations is not available as the benchmarking tool for a period of an hour at a medium pace.

55
ICE-B 2008 - International Conference on e-Business

R.'"-T..... srwFW[z....,M..:r~ICle<-N~~L _

" -
----

Figure 8: Dynamic Concentration Grid Example.

This application requirement performs the emulation

of an omnidirectional vision of each trackable item.
The described algorithm is able of identifying the
visible objects, having into consideration both single
instantaneous data and historical information, previ-
ously collected and stored – in direct memmory ac-
cess or in a database. In the conducted experiments,
Figure 6: Real-Time Item Location. this system’s module also showed high levels of effi-
ciency and correctness; much similar to the ones al-
As illustrated in Figure 7 the location data is used to ready described in the above paragraph.
extract more significant information about item dis- Regarding the system’s global architecture def-
persion/concentration both in the present and also inition and implementation, the undertaken simula-
considering historical data. It was used a gradient tions demonstrated its adaptative capability through
scale where concentration levels vary through the its flexibility in what concerns to both the database
RGB scale where red means high levels of concen- provider and, perhaps more important, to the loca-
tration and blue low levels. tion system tecnology. These caracteristics greatly
enhance the whole system’s applicability in several
scenarios. Still in this domain, the distribuited sys-
tem’s design enables the usage of low-end computers.
Therefore it constitutes an incentive to client’s IT in-
frastructure reusage while minimizing the solution’s
economical impact. Simultaneously, this approach
enables greater site manager’s empowerment through
real-time information access to all system’s features
visualization. These actions can be triggered for both
partial and global organization providing more and
deeper analysis points of view.

5 CONCLUSIONS

Figure 7: Historical Concentration Matrix. Considering the project’s simulation environment and
the achieved results, one shall state that, although the
One significant functionality of this model is the pos- location engine had been implemented in order to re-
sibility to perform calculations based on completely alistically simulate traceable items, all concept has
flexible and dynamic projection grid. This option been demonstrated. The developed prototype proved
proved to be efficient on online data processing for to be efficient and effective in large scale distributed
a significant number of tracked items - approximately data gathering and real-time item location visualiza-
one hundred - without database access by using a tem- tion.
poral location matrix. This feature enables a full de- Taking into account the system’s architecture, it
tailed concentration analysis in real time when recur- was verified the concern in allowing multi-store man-
ring to a high definition grid that divides the layout in agement with both distribuited modules and central
small areas; and enables swift high big-picture studies integration concerns that enable consistent and online
when using a less tight net. This capability is appro- knowledge extraction and visualization. Having in
priately described in Figure 8. mind the different application modules, one ought to
Finally, considering the features results description, refer that the integration with the XML-based layout
one shall paint the relevance of the vision module. manager proved to be extremely flexible to accom-

56
 BUSINESS INTELLIGENCE THROUGH REAL-TIME TRACKING - Using a Location System Towards Behaviour
Pattern Extraction

modate distinct real scenarios and yet realizable and ACKNOWLEDGEMENTS

realistic.
Considering the most noble project’s slice, one The authors would like to thank Professor Eugnio
ought refer to the previously enunciated knowledge Oliveira for inspiration and the contribution regard-
extraction features. Having as support the results de- ing behavior pattern recognition and artificial intel-
picted in the last section, it is secure to state that the ligence guidelines, Professor Augusto Sousa for en-
system is able to produce real-time shop-floor visual- lightenment in the portal cells algorithms.
ization maps with intelligible data on online item lo-
calization; individual item complete path routes; on-
line and historical population density rates and path REFERENCES
routes concentration; and also item-vision enabled
concentration maps – as emulation for item omnidi- Chao C., Yang J., J. W. (2007). Determining technology
rectional vision, yet considering occlusions. All of trends and forecasts of rfid by a historical review and
these features are allowed for graphical user interface bibliometric analysis from 1991 to 2005. In Techno-
through different grid dimensions for distinct analy- vation 27, Issue 5, pages 268–279. Elvisier Ltd.
sis granularity. Bearing in mind the project as whole, Choustova, O. (2007). Price-dynamics of shares and
one shall state that the developed knowledge extrac- bohmian mechanics: Deterministic or stochastic
tion platform with online and diversified visualization model? In Foundations of Probability and Physics,
tools constitutes a solid ground for online item track- pages 274–282.
ing and heterogeneous space management with dis- Grudzinski, J. (2007). Real-time clouds and weather sim-
tributed capabilities. One final major advantage of the ulation in computer games. In Learning with Games
2007.
proposed system resides in the total independence re-
garding the external position engine both in terms of Jappinen P., P. J. (2007). Preference-aware ubiquitous ad-
vertisement screen. In eCommerce 2007 Proceedings,
suppliers and even more important in terms of base pages 99–105. IADIS Press.
technology.
In spite of the enunciated project’s accomplish- Krotosky J., T. M. (2007). A comparison of color and in-
frared stereo approaches to pedestrian detection. In
ment, an even by being in prototype stage – yet re- Intelligent Vehicles Symposium, pages 1226–1231.
liable and fully functional – there are several future
Luce, R. (1999). Where is mathematical modeling in psy-
work areas that are able to greatly enhance the sys- chology headed? In Theory and Psychology, pages
tem’s appliance and success. From these, the most 723–737.
relevant ones are believed to be centered in eccen- Mingkhwan, A. (2006). Wi-fi tracker: An organization wi-
tric shop-floor layouts both in terms of shape and fi tracking system. In IEEE CCECE/CCGEI, Ottawa,,
multi-level buildings; complete path routes analysis pages 1353–1356.
enabling common node fusion for global paths proba- Moura J., Coelho A., L. P. e. a. (2004). Readings in Infor-
bilistic construction; flexible and dynamic report def- mation Society. Universidade Fernando Pessoa.
inition tool with configurable alarm triggering; and Park H., Lee S., C. W. (2006). Obstacle detection and fea-
perhaps the most interesting would be the character- ture extraction using 2.5d range sensor system. In
ization of ’what-if’ scenarios with simulated traffic SICE-ICA SE, 2006. International Joint Conference,
based on real historical data. Considering both the pages 2000–2004.
project current achievements and the depicted future Pires, P. (2001). Dynamic algorithm binding for virtual
work areas, one might identify the most desirable im- walkthroughs. Master’s thesis, Technical University
plementation domains. Although there are not limited of Lisbon.
to, the proposed system might be useful for traditional Raizer, V. (2003). Validation of two-dimensional ocean
retail environment for shopping cart tracking; tracing microwave signatures. In Geoscience and Remote
clients through a commercial area such as shopping Sensing Symposium, 2003. IGARSS ’03., pages 2694–
2696.
centers; enabling item tracking and route analysis in
an hospital; producing activity reports and analysis in Vazquez, P. (2007). Real time falling leaves. In Second In-
controlled areas such as penal complexes, mental in- ternacional Conference on Computer Graphics The-
ory and Applications, pages 244–251.
stitutions or closed educational organizations.

57
A FRAMEWORK FOR DYNAMIC KNOWLEDGE ACQUISITION

Ana Aguilera and Alberto Subero

Facultad de Ciencias y Tecnología, Universidad de Carabobo, Valencia, Venezuela
[email protected], [email protected]

Keywords: Collaborative Work, Multi-agents, Medical diagnosis.

Abstract: This article proposes a conceptual model based on agents in medical workgroups for the resolution of
clinical cases. Our objective is to provide a suitable conceptual framework for the dynamic acquisition of
expert knowledge from different sources. This framework is sufficiently flexible to allow its transformation
to the symbolic level that finally characterizes the implementation approach; it is a design model prior to
implementation. We considered an interface design to be used for the integration and combination of
different sources of expertise. The subject which interests us is the knowledge produced in the workgroup
processes.

1 INTRODUCTION On the other hand, the evolution of the

application domain in traditional Artificial
There are many applications related to the Intelligence (AI) to cover other complex and
workgroup such as decision-making, monitoring, heterogeneous fields such as aid to decision-making,
transmission, planning, idea generation, problem form recognition and comprehension, process
resolution, resulting discussion, negotiation, conflict control, etc., shows the limits of traditional AI
resolution, analysis and system design and approaches. All this has contributed to the birth of a
collaborative group activities such as the preparation new discipline: Distributed Artificial Intelligence
and distribution of documents, to mention just a few. (DAI), which is interested in the intelligent
This work style is illustrated by real-life behaviours (and their modelling) that are the
applications, where their nature is clearly outcome of cooperative activity between several
collaborative. Several of these applications imply agents. Nevertheless, the passage from individual
various specialized fields, such as medical diagnosis. behaviour to collective behaviour is not only
The field of medicine has become so vast that it is considered as an extension but as an enrichment of
divided into several specialities and many medical AI, as new properties and new behaviours emerge
cases are considered by group analysis. Among the from it. The purpose of DAI is to provide a remedy
various workgroup activities, complex-problem for the shortcomings of the traditional AI approach
resolution is highlighted as being an increasingly by proposing expertise distribution for an agent
important subject today. group capable of working and acting in a common
Different information technologies which environment, and solving possible conflicts. New
consider these group modalities have been concepts in AI have appeared, such as cooperation,
developed in software engineering. Thus, we found action coordination, negotiation and emergence.
a broader field entitled group support systems or Three fundamental research areas of DAI are multi-
electronic meeting systems which include other wide agent systems (MAS) (Weiss, 2000), distributed
areas like Group Decision Support Systems (GDSS), problem solving (DPS) and Parallel Artificial
Distributed Group Support Systems (DDSS), Intelligence (PAI).
Computer Support Collaborative Work (CSCW),
Groupware, where the common denominator of such
technologies covers, but is not limited to, distributed 2 THE PROBLEM
facilities, computer hardware and software, audio
and video technology, procedures, methodologies, Everyday medical work (implying inter-consultation
facilitation, and applicable group data (Turban, between specialists, case conferences, and hospital
1995). morning rounds) includes exchange and cognitive

58
 A FRAMEWORK FOR DYNAMIC KNOWLEDGE ACQUISITION

processes within groups. It is possible to observe This architecture determines the structure on which
different physicians (specialists and non-specialists) the framework will be defined and permits us to
working together, at the same time or at the different clarify the semantic definition of the problem under
time, on the same patient (clinical examination often study.
requires the cooperation of several specialists). The
contribution of these exchanges to solve complex 3.2 Methodology
problems is becoming more and more essential.
These acts of collaboration are important for clinical Different methodologies for the development of
decision-making concerning diagnosis and treatment knowledge-based systems (where knowledge
both for students still in training and recently acquisition and modelling play a leading role) have
qualified physicians, without forgetting the emerged during the past 15 years. We can mention
continuous experience which helps to develop expert in particular methodologies such as KADS
knowledge. There are several examples: cases of (Schreiber, 1993), VITAL (Domingue, 1993),
multi-system illnesses where physiopathology and REFLECT (Reinders, 1991), (van Harmelen, 1992),
the nature of origin of the illness make it necessary ACKnowledge (van Heijst, 1992), CommonKADS
to examine the results of several diagnostic (Schreiber, 2000) and KATEMES-MultiExp
procedures; patients with chronic disorders such as (Ladibi, 1995). KADS and VITAL focus on the
diabetes mellitus, obstructive pulmonary illnesses, development cycle of a knowledge-based system,
cardiological illnesses; or patients receiving REFLECT proposes theories and architectures for
palliative care at home. reflexive systems, ACKnowledge introduces
On the whole, the most frequent scenarios where directive models for knowledge modelling.
it is possible to observe this collaborative pattern CommonKADS offers methodology for knowledge-
are: intensive system development and for knowledge
Inter-consultations. This is a process where a management and KATEMES-MultiExp is a
physician needs specialized consultation with prototype tool for multi-expert knowledge
several other physicians. There are two possibilities acquisition.
for this process: 1) between specialists with the same The approach used is based on an extension of the
speciality (consultation between radiologists who KATEMES-MultiExp methodology. This method
study an image to decide on a diagnosis) and 2) integrates a set of models for modelling multi-
between specialists with different specialties (an expertise and its objective is to help the knowledge
obstetrician who refers his/her patient to a engineer in the collecting and modelling phase of
cardiologist for a coronary problem). human behaviours in cooperative problem solving.
Case Conferences and morning rounds. There are These models are: agent, organization, cooperation,
cases where it is possible to find several physicians task and communication models (Aguilera, 2003b).
(specialist physicians interconnected with general We added a coordination model (Aguilera, 2003a)
practitioners) exchanging information on several because originally KATEMES was designed for
medical cases in order to make patient evaluations, knowledge acquisition before system development
to work on the publication of cases, etc. These are and we considered the dynamic knowledge
considered as staff meetings. acquisition approach. One of the important features
A clinical case illustrating this type of work is considered is the ability to extract expertise in a non-
shown in figure 1 (A more detailed explication is intrusive way. This minimizes biased behaviour by
found in (Quintero, 2003). In this case, we can monitoring expert decision makers directly (through
observe the interaction between a generalist, a the information system) during their daily tasks,
radiologist and a pathologist. without changing anything about their choices or
strategies. The approach is based on the concept of
the agent, with the idea of modelling a group of
3 PROPOSITION experts via a community of agents in interaction.
This gives us the additional advantage of multi-agent
3.1 Architecture systems which provide a way to relax the constraints
of centralized, planned, sequential control and to
The problem follows a structure on three levels make systems that are decentralized, emergent and
(Figure 2): the collaborator level or users (human concurrent available. We are convinced that, the
agents), computer systems agents (software agents) modularity of a multi-agent architecture facilitates
and the data level (databases and knowledge bases). knowledge acquisition and the parallel design of
each expertise. In fact, the multi-agent systems are

59
ICE-B 2008 - International Conference on e-Business

P atient data
A ge: 40 years old.
Sex: m asculine
C ontex t: H e is a non-smoker and he does not present any obvious particular antecedents in his past m edical
history.
Sy mptom s: H e went to his general practitioner with a non-productive cough of three months.
P hysical T est: normal
T reatm ent: palliative treatm ent
C om plem entary T ests: laboratory and paraclinical tests (Postero-anterior chest x-ray)

Circumscribed 2-3 cm N odule located in the

right upper lobe of the lung with the presence of
interior calcifications of non-specified type.
Scanner recom m ended. H eart and rest of study
without obvious m odifications.

C AT Lung scanner: It indicates a 2 x 3 cm mass with non-epim acular aspect located in the right upper lobe of
the lung with non-specified calcifications.
There is no affectation of M ediastinum lym phatic ganglia. There are no other m asses in the thorax.

Pulm onary biopsy: m acro and m icroscopic analysis of post-operative piece.

D iagnosis: E N D O BR O N C H IA L H AM AR TH O M A.
The patient leaves the hospital and considering the benign origin of the pathology, the doctor recom m ends an
annual check up with his general practitioner.

Figure 1: A clinical case.

I Distributor
Agent
Physician
(in charge N
of the case)
T Data
Agent
Knowledge
E Planning base
Agent
R
M edical M anager
F
Staff Agent
Interface
A Agent M edical
Past History
C
App. Speciality-
E oriented

Collaborator Level Computer Systems Level Data Level

Figure 2: Problem Structure.

suitable for highly dynamic domains. They belong to 3.3 The Framework
the complex open systems category and make
incremental application development possible. It is clear that the installation of a knowledge
Our choice is justified because agent modelling acquisition environment coming from multiple
adapts well to the distributed nature of the problem, experts requires a multi-field approach which, at the
as well as, to the preliminary phases of the same time, integrates social, cognitive and data-
knowledge acquisition and to the development of processing dimensions. It is for this reason that the
collaborative activities between several experts. objective of this model is to integrate social aspects

60
 A FRAMEWORK FOR DYNAMIC KNOWLEDGE ACQUISITION

coming from the human activities, cognitive aspects asks for information. He/she makes it possible to
held in the databases and knowledge bases of the establish the clinical context.
system, and the aspects related to computer systems. - The planning agent: An artificial agent which
This proposal considers important aspects of multi- manages everything related to time in the system:
agent systems and software engineering. assigns the tasks with a time context, checks the
We consider the importance of the conceptual time of their activation, execution and ending. It
model phase in the development of any technology, returns the requests of task assignments if it has not
since it provides a good definition and a clear received an opportune response. It centralizes
analysis of the problem treated. The model helps the information about task executions and their different
knowledge engineer to consider the presence of all states of development.
significant elements and their correct interrelations - The distributor agent: An artificial agent which
and it is also an instrument which enables him to locates the agents that will execute the tasks,
exchange opinions with colleagues and experts, etc. according to their specialities and availability for
The advantage of this framework is that it offers us a executing them. For this, it takes into account the
high level tool of abstraction. This tool enables us to urgency of tasks required. It centralizes information
discuss viewpoints with others and is therefore a
about group members.
facilitator for conceptual analysis. Additionally, this
framework is a set of models that gives a wide - The manager agent: An artificial agent which
vision including all actors, their interrelations, their controls the requests for tasks. It sends and receives
organization, their communication mechanisms, and results.
more. - Interface agent: Artificial agent who controls
The framework definition is given, the security accesses to systems and data. It
simultaneously in the structuring and adequately adapts the appropriate interface
conceptualizing of each model in the extended according to each user with his/her different levels.
KATEMES-multiexpert methodology. At the It centralizes general access information.
conceptual level we can define the following - The data agent: An artificial agent which
models: manages data in the databases and knowledge bases.
• Agent model It manages metadata, user view definitions and
This model enables us to identify and define all authorizations to data access. It also controls
human and artificial individuals interacting and anything related to data integrity and recovery
taking part in cooperative problem solving. These mechanisms.
agents are (figure 2): Subsequently and for reasons of brevity we will
- The medical meeting agent: This is considered define only the radiologist agent. This is a human
as a high level hierarchical agent. It is a set of agents agent, an expert and a specialist belonging to the
and it is defined to identify the group of physicians. radiology department. He may be or not be
The communication protocol between agents follows subordinate according to his status. He is part of a
human communication procedure based on personal community including radiologists or other
conversations; that is, somebody intervenes and the colleagues of different specialities (general
others remain silent and listen, and when one person practitioners, pathologists, etc). All of these people
has finished, another intervenes and so on. The rank equally and take part in the development of a
leadership position can be assumed by the doctor diagnostic solution. This agent also has a
sitting opposite the patient (Physician in charge of hierarchical relationship with the department head.
the case). With respect to inter-agent relationships, the
- Medical speciality agents. They define a radiologist could influence the reasoning line of an
classification based on the different medical agent with another speciality, who requires his
specialities, i.e., human agents such as radiologists, service for a diagnostic examination, or he could
neurologists, pathologists and others. The directly influence another radiologist who discusses
knowledge of agents is heterogeneous. The agents a particular case with him. In the radiology
divide the tasks amongst themselves and share data department, software systems for image processing
about the patient. Each one of these specialists can are frequently used. The execution of these systems
observe only one part of the “outside” (i.e., the can be initiated or stopped at any time by
patient). radiologists. Concerning cooperative relationships,
- The patient: A human agent who as an active the agent can assist another radiologist in the
being takes part in his/her treatment. He/she requests establishment of a diagnostic hypothesis or in other
medical appointments, takes part in his/her cure, tasks. In the medical group context, his task does not
interfere with other tasks but sometimes its

61
ICE-B 2008 - International Conference on e-Business

accomplishment is important for the continuation of • Task model

other tasks. Other inter-agent relationships can be To arrive at a diagnosis, doctors carry out several
established, but in order to simplicity, only those tasks. In particular we mention: hypothesis
related to the example given (figure 1) will be generation, selection and test of hypotheses and
mentioned. diagnostic conclusion. In the case of hypothesis
• Expertise model generation, which is the example presented here, it is
not clear if the same process is followed equally in
Knowledge based systems are founded on a
every medical speciality. Undoubtedly, there are
separation between the knowledge necessary for
different contexts and expertise models can vary
problem solving (domain knowledge) and the
within the same speciality. However, it is not our
reasoning mechanisms needed to exploit this
objective to present a discussion of the diagnostic
knowledge (inferential process knowledge for
process.
problem solving). In the research field, this subject
remains open and there is a great diversity of • Communication model
analyses and different viewpoints about it. This topic There are several communication mechanisms
is related to reasoning mechanisms used by doctors between agents given by the type of their
in diagnostic problem solving. Many variables are interrelations. For example, in the communication
considered: patient, doctor experience, general between human agents, there are communications
medical knowledge. The discussion of this theme is based on spoken conversations, written
extended and goes beyond the domain of this article. conversations (chats) or by data (medical file).
We suggest readers refer to specialised literature Between human and artificial agents, the
(Torasso 2001, Long, 2001), in particular, we refer communication is established by an interface agent.
to (Park, 2003). Among artificial agents, there are mechanisms based
on message passing, and finally, between artificial
• Organisation model
agents and data, through a data agent.
In the architecture proposed (Figure 2), we find
Communication modelling is described by inter-
artificial and human agents. From this, we identify
agent relationships via R-intervention relationships
the medical group as a hierarchical structure with
which do not contradict the organisational structures
two levels; headed by the physician treating the
described. Some of them are shown in the definition
patient, and then the rest of the doctors are
of the radiologist agent.
considered to be collaborators. This structure can
change by establishing as group leader another • Coordination model
doctor in any other speciality. Coordination is supported by artificial agents: the
It is also possible to observe other hierarchical manager, the planning and the distributor agent.
structures, for example in a medical group within the They centralize everything related to the
same department (radiology department, pathology coordination tasks required. Coordination will also
department, etc.), with the department head at the depend on the dynamics of the cases exposed, their
highest level. characteristics, their emergency requirements, their
Within a group of artificial agents, some have resources and the interrelations established between
only coordination functions. A hierarchical structure the active agents..
is present and centralized in the manager agent. It is Therefore services offered by this model depend
possible to observe delegation of tasks. For example, on: the session context, the resources provided by
the manager agent requests that the planning agent participants and the social behaviour of participants
specify the localization of a task and then the latter (actions carried out by human agents) (Ossowski,
asks to the distributor agent which finally assigns the 1999). The combination of these three elements
task. creates the dynamics of coordination and the basic
actions for this collaboration; the context fixes the
• Cooperation model
limits of these dynamics.
The model presupposes the cooperative
motivation of its participants. In the case of a
medical group, the doctors can be assisted within the 3.4 System Design and Implementation
same speciality, when they request consultation at
any step of the diagnostic process (doubts, At implementation level, we design a multi-agent
confirmations, etc.). In different specialities, no one architecture based on the framework proposed. The
interferes directly with the others but their effective different medical specialities are modelled well with
participation, in the total establishment of diagnosis, a modular structure. The current project is
is decisive. progressing. So far, we have designed and
implemented a groupware application that facilitates
the management of clinical cases (Cárdenas, 2001)
and a groupware application for radiological

62
 A FRAMEWORK FOR DYNAMIC KNOWLEDGE ACQUISITION

teleconsultations. The conceptual data model of our J. Domingue, E. Motta Et S. Watt, “The Emerging VITAL
application uses an entity relationship model. At the Workbench”, Lecture Notes in Computer Science,
moment, we are developing a workflow application volume 723, pp. 320-339. Springer-Verlag, Toulouse
for the coordination of medical diagnostic work. and Caylus, France, 7th European Workshop, EKAW,
1993.
S. Ladibi, “Ingénierie de la Connaissance dans le cadre de
Projets Multi-Experts : Méthode, Techniques et
4 CONCLUSIONS Outil”. Thèse Doctorale, Université de Nice – Sophia
Antipolis, 1995.
In this article we defined an agent-based conceptual W. J. Long, “Medical informatics: reasoning methods”,
framework that integrates the social aspects of Artificial Intelligence in Medicine, Volume 23, Issue
human activities, the cognitive aspects structured in 1, August 2001, Pp. 71-87.
S. Ossowski, “Coordination in Artificial Agent Societies.
databases and knowledge bases of the system, as
Social Structure and Its Implications for Autonomous
well as, the computer system aspects which are
Problem-Solving Agents”, Number 1202 in LNAI.
present in the software applications and information Springer Verlag, 1999.
technologies, in the medical diagnostic context. This W. Park, “Causalité et Raisonnements dans le
approach is model-based; thus six models are diagnostique médical”, proceedings Setit, Sousse-
defined: the agent model, the organization model, Tunisia, 2003.
the cooperation model, the task model, the J. Quintero, “Décision collaborative dans le domaine
communication model and the coordination model. médical analyse d’un cas de diagnostic”, proceedings
It is possible to instantiate these models for Setit, Sousse- Tunisia, 2003.
particular situations, for example, clinic and hospital M. Reinders, H. Akkermans, et al., “A conceptual
frameworks. We have considered the extension of modelling framework for knowledge-level reflection”,
the metamodel proposed in (Ladibi, 1995), the AI-Communications, 4(2/3) :74-87, 1991.
annexing of the coordination model and we provided G. Schreiberakkermans, H., Anjewierden, A., et al.,
general templates for medical groupwork. The “Knowledge Engineering and Management. The
selection of the model-based method and the design CommonKADS Methodology”. The MIT Press, 2000.
of the model were directed towards the dynamic P. Torasso, “Multiple representations and multi-modal
reasoning in medical diagnostic systems”, Artificial
knowledge acquisition produced in collaborative
Intelligence in Medicine, Volume 23, Issue 1, August
medical interactions, which is the next the step in
2001, Pp 49-69.
this research. E. Turban, “Decision support and Expert Systems.
Management Support Systems”, Prentice Hall, 4°
Edition, 1995.
ACKNOWLEDGEMENTS R. Van Harmelen, “Knowledge-level reflection :
Specifications and architectures”, Technical Report
R.2 RFL/UvA/III.2, ESPRIT Basic Research Action
We are grateful to FONACIT, Venezuela, under P3178 REFLECT, REFLECT Consortium, 1992.
Project No G-2005000278, which has sponsored this G. Van Heijst, P Terpstra, et al., “Using generalized
work. directive models in knowledge acquisition”, In Proc.
of the 6th European Knowledge Acquisition
Workshop, pp. 112-132, Heidelberg and
Kaiserslautern, Germany, Lecture Notes in Artificial
REFERENCES Intelligence, 1992.
G. Weiss, “Multiagent Systems: A Modern Approach to
Aguilera A.(a), “An agent model for medical diagnostic Distributed Artificial Intelligent”, The MIT Press,
support with multi-experts", Sciences of Electronic, 2000.
Technology of Information and Telecommunications,
SETIT 2003, Sousse, Tunisia, Mars 2003, paper
R234A002.
A. Aguilera (b), « Télé-consultation en radiologie
médicale », Département ITI, ENSTB. Colloque Les
Nouvelles Technologies dans la cité, Rennes- France,
Déc. 2003 WWW : http://tic-cite.univ-
rennes1.fr/resume.php?lien=42.txt&mail=ana.aguilera
@enst-bretagne.fr.
F. Cárdenas, J. Pacheco, “Desarrollo del software para un
sistema colaborativo de diagnosis médica”. Thesis in
electrical engineering, Universidad de Carabobo,
2001.

63
 PRES – PERSONALIZED EVALUATION SYSTEM IN A WEB
COMMUNITY
A Conceptual Model Designed to Evaluate Reputation in Order to Achive a
Personalised View on the System for Each User

Lenuţa Alboaie
Department of Computer Science, “A.I. Cuza” University of Iaşi – 16, Berthelot, 700483 Iaşi, Romania
[email protected]

Keywords: Reputation systems, evaluation, resource, Web community.

Abstract: The purpose of the PRES model is to build a flexible and easy way to manage resources in a personalized
manner. Our proposed model assures for every user that his preferences are important and permits the
formation of some homogenous groups on the basis of these preferences. The homogeneity is due by the
relations resulted from the explicit and implicit evaluations of resources. The purpose of the proposed model
is to build a flexible way to filter irrelevant resources for users. In this way, a user which is member to a
community based on the PRES model will dynamically see information that he/she is most interested in.

1 PREAMBLE model. In the fourth section we analyze the benefits

of the proposed system. The article ends with an
In this moment the WWW space stores large overview on the discussed domain, mentioning the
amounts of data which are continuously growing. future directions.
The main problem that appears is to find solution to
use efficiently the existent resources.
A first step to solve this problem is to associate 2 ACTUAL SITUATION
metadata to resources. As a fact, it is a manual
classification process performed by the user (E.g. At this moment there are many sites that collect
delicious, digg.com). This direction is a part of various information about thousands or even
explicit Web that is realized through explicit millions of people on the Web. This information is
activities as tagging or digging. obtained often without even interrupt user actions
An important direction, using the above solution, is with questions. His behavior and profile can be
to obtain data/information by observing and obtained from this information using different
analyzing the user actions. Thus, we enter the space techniques like machine learning and statistical
that is known as implicit Web (O’Reilly, 2005). An methods
important drift of it is collective intelligence domain In the collective intelligence spectrum we have
(T. Segaran, 2007). two different approaches, one exists due the
In this context, this work proposes the analysis information furnished by users (e.g. Wikipedia). The
and projection of a prototype of a reputation’s other part of the spectrum is based on different
personalized evaluation system in a Web community algorithms which allow obtaining new information
(PRES - Personalized Resource Evaluation System). that enhance the user experience. An important
The originality of this approach consists in the example in this sense is Google, which uses links to
chosen perspective to accomplish the evaluation. rank web pages, but also collects and process data
This work is structured as follows: in section 2 obtained from situations when advertisements are
we describe a short survey on the present situation clicked.
(O’Reilly, 2005, H.Zhuge, 2008). In the next section Other examples consist of web communities that
we present the problem, we explain why such a use recommendation systems (Massa,
system is necessary and we present the proposed B.Bhattacharjee, 2004). In this cases there are

64
 PRES – PERSONALIZED EVALUATION SYSTEM IN A WEB COMMUNITY - A Conceptual Model Designed to
Evaluate Reputation in Order to Achive a Personalised View on the System for Each User

collected information like purchasing history and Another system we referred here before is
user characteristics, and the system make proper Wikipedia that represents an online community
recommendations based on them (e.g. Amazon, containing a great number of users, but not using a
Netflix). formal reputation computation mechanism.
Other examples consist of web systems which As in the previously discussed systems, a less
use reputation systems (Golbeck, Hendler, 2005). visible user hierarchy exists. All users, on the basis
Reputation systems are extremely useful in those of their contribution, may receive the so-called
communities where the users have to interact with barnstar acknowledgement. Although one can
some resources posted by other users or they have to follow each user posting history, it does not exist a
interact with other users. (E.g. YouTube, Slashdot, particular rating system.
Flicker). In these situations, using experience of
other users would be very useful. Also, reputation
systems are useful in setting some evaluation levels 3 PRES MODEL PROPOSAL
for users and resources (e.g. more or less interesting
resources). There are a variety of reputation systems.
A well-known system, mentioned before, is Google
3.1 Context
Page Rank (A. Langville, C. Meyer, 2006) that is
In section 2 we have discussed a set of reputation
based on complex algorithms that assure the web
systems (R. Falcone, S.Barber, L. Korba, M. Singh,
page ranking.
2002), but in all these related approaches we do not
Another reputation system is that used by eBay.
find a personalized evaluation. In this section we
The system assures a feedback profile for each
explain what a personalized evaluation means, from
member.
our point of view.
Each feedback consists of a positive, negative or
In a Web community there exist a lot of
neutral value (these values are obtained from the
resources. There are human resources and other
ratings of the transaction partners) and a short
types of resources. The people have either different
comment.
or similar profiles. Therefore, they are interested in
Everything2 is a knowledge base that contains
either different or similar resources.
reputations system both for users and their posted
We quantize this interest with values which are
articles. The system is based on anonymous votes of
provided by the user for other users or resources.
other users which determine positive or negative
Also, this interest will have an indirectly computed
ratings. Negative evaluated articles are deleted. The
component. We give a simple example here, the
users are evaluated on the basis of the number of
other cases being analyzed in section 3.2. We have
their submitted articles (and not deleted) and on the
the situation when a user evaluates favorably one or
average of their associated values.
more users. These users evaluate favorably a given
Such a system implies some problems: new users
resource. Even if the user does not evaluate directly
posting articles that receive negative feedbacks may
that resource we will consider an implicit favorable
appear. These articles will be deleted, thus
evaluation (J.Golbeck, J. Hendler, 2006). Thus, the
discouraging new postings by such users. Even the
user has the chance to access more relevant
experienced users hesitate to post new articles which
resources for him.
they consider as being not very good, because the
In our system there is no it does not exist an
received negative feedbacks are not deleted. Also, in
absolute value of good or bad resource
this kind of system the re-actualization of older
characteristic. A resource can be good for a set of
articles is less appreciated.
users but not useful for other set of users.
Slashdot has a reputation system named karma.
In section 3.2 we establish a set of metrics (J. L. Mui,
In this system there are moderators that can make
2002), taken into account by the evaluation
the evaluations in a similar way to the system
mechanism, for the purpose of measuring the
Everything2. Every user may become moderator if
usefulness of a resource for a given user.
he has a good karma obtained on the basis of the
Whenever new users become community
ratings associated to their comments. But this
members they can interact with the users
moderator state is temporary until he uses the
corresponding to their preferences. Also, they will
available votes. This evaluation system is criticized
be able to access much faster the proper resource set.
because it is weak on issues like Anonymous
This represents the general direction our system is
Coward or sock puppets (R. Falcone, S.Barber, L.
based on.
Korba, M. Singh, 2002).

65
ICE-B 2008 - International Conference on e-Business

3.2 The Proposed Model ƒ Explicit worth of a resource: WE_UR

(user, resource) – explicit worth, represents
First we define the vocabulary used in the the rating for a resource, this rating being
developing model. We also specify the used given manual by a user
notations and their semantics. The system will ƒ Implicit worth of a resource: WI_UR (user,
contain: resource) – implicit worth, represents a
ƒ Users which know other users. rating inferred from the set of existing
ƒ The list of the users considered to be ratings from the known person list of a user
interesting for a user. ƒ Explicit worth of a user: WE_UU (user,
ƒ Users nominated by a community as user) – explicit worth, represents the rating
evaluators. We use notations E1...En to for a user, and the rating is given manual by
indicate the community evaluators. These the user to another user
evaluators are in fact some reviewers. They ƒ Implicit (deducted) worth of a user:
will be useful for the new users which have WI_UU (user, user) – measure how close
not established their own knowledge list are his preferences to the others preferences
yet. (The preference can be considered: the
ƒ Known person list of a user. Initially, it accepting degree of a point of view or the
contains the community reviewers list only. appreciation degree of a piece of art).
ƒ Resources – their definition is made Implicit we consider that we have:
accordingly to the definition given by (T. WI_UU (user, evaluator) = MaxWorth;
Berners-Lee, 1998).
If an user evaluates an evaluator in an explicit
So, in our system one considers as resources manner, then this evaluation - WE_UU (user,
everything having an identity (e.g. electronic evaluator) – will have priority.
document, an image, a service and eventually a
ƒ we consider the function WU(user, user)
collection of other resources). There are considered
for every pair of (user, user) from a Web
as resources those that cannot be accessed directly
community
via Internet (e.g. human beings, organizations)
Its value will be WE_UU (user, user) if there is
ƒ Worth – this parameter is a metric. This an explicit evaluation (different from 0), otherwise
metric signifies a given rating, according by its value will be WI_UU (user, user). So, let us
a user to a resource or a user. Also, the consider the users: Ux and Uy.
worth can be obtained (quantized) If the user Ux evaluates explicitly the user Uy
indirectly. then the function WE_UU has a value different of 0
This parameter – Worth – takes the following and the value of WU(Ux,Uy) will be WE_UU(Ux,Uy).
values: If Ux does not make an explicit evaluation for
ƒ 1 = useless/spam user Uy then WU (Ux,Uy) value will be the inferred
ƒ 2 = poor value which is actually the value of WI_UU(Ux,Uy).
ƒ 3 = worth attention ƒ we consider a function WR(user, resource)
ƒ 4 = good for every pair (user, resource)
ƒ 5 = exceptional. We note this limit with WR (user, resource) value will be WE_UR if the
MaxWorth. user evaluates explicitly the resource, thus the value
We think of using the 1-10 interval for possible of WE_UR exists. Otherwise WR value will be the
values for Worth metric, this approach assuring value of WI_UR.
higher granularity in resource evaluation. We prefer Therefore, let us consider the user Ux and the
the above specified selection to simplify the model. resource Ry. WR (Ux, Ry) value will be WE_UR
In future works we will analyze if this aspect has a (Ux,Ry) if user Ux has explicitly evaluated the
major influence on the resource evaluation manner. resource Ry. Otherwise WR value will be WI_UR
We will use a set of constructions which have the (Ux, Ry) if the user Ux did not evaluate the resource.
following associated semantics. In fact, these This value is based on the ratings to Ry made by
constructions can be mathematically considered as users which are in known list of the user Ux.
functions (eventually partial functions) or, from the We will define the manner of computation of the
implementation point of view, they are considered implicit values introduced above.
associative tables:

66
 PRES – PERSONALIZED EVALUATION SYSTEM IN A WEB COMMUNITY - A Conceptual Model Designed to
Evaluate Reputation in Order to Achive a Personalised View on the System for Each User

Implicit WI_UU Value Computation The need of partial function WIR_UU when the set
of users used for defining of WIU_UU is the empty
Let us consider two users Ux, Uy from the Web set. This means that we do not have a set of users
community. In order to define WI_UU(Ux,Uy) we U1, .., Uk whom we have explicit ratings from User1
introduce the following partial functions: to Ui, 1 ≤ i ≤ k and also we do not have explicit
ƒ WIU_UU(Ux,Uy) – its value indicates the ratings from Ui to User2. In this case we can obtain
deducted worth based on explicit information on the basis of the worth of a set of
evaluations made by users to each other resources evaluated by users. These resources are
ƒ WIR_UU(Ux, Uy) – its value signifies the required to be evaluated by both users. Thus, on the
deducted worth based on evaluations that basis of the evaluations of the same resource, one
users do to the same resources can obtain a mutual evaluation of two users.
Defining WIU_UU value on the basis of the Let us consider: Ux, Uy and the resources
explicit values R1,...,Rn. If there exists WE_UR(Ux,Ri) and
Let the users, whom we have explicit ratings WE_UR(Uy,Ri) , for all 1 ≤ i ≤ n , then the value of
from user User1 to be Ui, 1 ≤ i ≤ k , be U1,…Uk. WIR_UU(Ux,Uy) will exist and it will be equal with
Therefore we have the definition WE_UU WIR_UU(Uy,Ux). We define WIR_UU (Ux,Uy) as
(User1, Ui). Also we have explicit ratings from Ui to follows:
User2, so we have defined WE_UU (Ui, User2) (see WIR _ UU (U x , U y ) = MaxWorth −
Figure 1). n (3)
∑ | WE _ UR (U
i =1
x , Ri ) − WE _ UR (U y , Ri ) |

where 1 ≤ i ≤ n . The demonstration of the assertion:

WIR _ UU (U x ,U y ) = WIR _ UU (U y ,U x ) (4)

Figure 1: WI_UU computation based on explicit
evaluations. is obvious. Therefore, in the case when we want to
obtain WI_UU on the basis of resource evaluation
To evaluate the WIU_UU value we must WI_UU has the value of WIR_UU (U1, U2).
consider which is the value of the weight Finally the worth of WI_UU (U1, U2) will be
corresponding to the explicit ratings. WIU_UU (U1, U2), if defined, or WIR_UU, if
We denote this weight with PE (User1, Uj). It defined, or it will be a default value fixed in the
represents an explicit rating weight, in our case the system configuration.
weight of the rating provided by User1. In a future work, we will present a mechanism to
The value of this weight is computed as a ratio of obtain a complete function WIU_UU without this
the explicitly user established value and the sum of implicit value. In addition, if we have both user-user
all explicit ratings provided by him. We will have: and user-resource evaluations, then we can foresee a
given priority between them.
WE _ UU (User1 , U j ) (1)
PE (User1 , U j ) = k Implicit WI_UR Value Computation
∑ WE_UU(User1 ,U i )
i =1 We will define the manner to obtain the worth of the
implicit evaluation - WI_UR(Ux,Rz) – whom a user
where 1 ≤ i ≤ k , 1 ≤ i ≤ k . Ux provides to a resource Rz, 1<z<n, where n is the
In this moment we prepare the context to resource number of the system. We consider that the
compute implicit rating whom user Ux provided to user Ux has in his known person list the following
Uy: users: U1,...,Uk. These users have evaluated the
k
(2) considered resource. This means we have defined
WIU _ UU (U x , U y ) = ∑ PE (U x , U j ) * WE _ UU (U j , U y )
j =1
the following relations: WE_UU(Ux,Ui), 1 ≤ i ≤ k
and also WE_UR(Ui,Rz), 1 ≤ i ≤ k
where 1 ≤ j ≤ k . The implicit rating provide by Ux to resource Rz
Defining WIR_UU value on the basis of resource is represented by the proportion between: sum of the
evaluation product of the rating weights of the user Ux for each
user from his list and the value provided by him to

67
ICE-B 2008 - International Conference on e-Business

resource Rz, and the number of users (which is k in In other case when user Uy will evaluate users Uj
our situation) with worth metric with a higher value than the sum
k
k

∑ PE(U x ,U i ) * WE _ UR(U i , Rz ) ∑ PE (U
i =1
x ,U i ) *WE _ UR (U i , Rz ) has a higher value
WI _ UR(U x , Rz ) = i =1 (5)
k and in this case the resource Rx will be automatically
consider useful for user Uy. We argue with this
where 1 ≤ i ≤ k . We introduce the worth average example one case from a set of possible use-cases.
provided to a resource and we denote it with WA. We will discuss in detail these cases and the used
The value of WA (Resource) represents relevant algorithms in our next paper.
statistical average provided to a resource by all ƒ The resources which are relevant for the
users. WA for a resource inside a Web community user are on top of the list of visible
plays the same role that page rank plays in Web resources. In this moment we know that
page evaluation. This metric is necessary in case we Google uses Page Rank system. The new
do not have enough trustworthy evaluators in the resources, even valuable, will reach hardly
community. on the top, because it takes long until they
receive links. And worse it is the fact that if
they are not on the top, they do not receive
4 ASPECTS REGARDING THE links.
PRES BENEFITS Therefore there exists a very high probability
that a good resource is not used.
In this section we discuss shortly a set of In our system the new valuable resources appear
consequences, due to the way the system has been quickly on top when they are evaluated first by an
modeled. We will argue our assertions through few honest community member (one who tries always to
examples and in a next paper we will give the evaluate correctly). If somebody over evaluate his
appropriate algorithms used for these cases. own resource and the others rate it with low marks,
the mark WI_UU will drop, therefore those who add
ƒ The system assures the property to see the
resources are required to give right marks.
things prioritized the same way as similar
users. ƒ The users will be required to do a fair
ƒ The spammers will see more spam because evaluation.
the system groups the users by their It will not happen like in the eBay system. In this
preferences. system, one assures a feedback for each user. The
Let us consider a web community with users feedback value is obtaining from other users
U1,…,Uk. We can consider that a new user Ux joins evaluations. One observed that the users are afraid of
the community and posts a new resource - Rx. obtaining a negative feedback. For this reason they
The resource posted by Ux will be evaluated by post positive feedbacks in a high proportion, hoping
the users from community with worth values ( that they will obtain positive feedbacks.
implicitly WE_UR(Ux,Rx)=5). ƒ The system can be easily integrated in
If Rx is a spam resource, it will be explicitly different Web communities.
evaluated by users Ui which are not interested in Let us consider a real community like LinkedIn.
spam resources with WE_UR(Ui,Rx)=1 or it will be There exist in this moment some posted
explicitly evaluated by users Uj which are interested announcements which offer jobs for this community
by this kind of resources with WE_UR(Uj,Rx) = 5, only. Our system would give the possibility that this
where 1 ≤ i ≤ k , 1 ≤ j ≤ k , i ≠ j . announcements to be visualized only by the users
Also, let us consider the case when a user Uy with a given profile, the announcement being not
evaluates the users Ui. Because users Ui have useful for other users types.
evaluated resource Rx with low worth than the sum Thus our system makes it more efficient the
k information management that is visible to the user.
∑ PE (U
i =1
x ,U i ) *WE _ UR (U i , Rz ) has a low value.

Than the value of WI_UR(Uy,Rx) will be low and in

this case the spam resource Rx will not be considered
interesting for the user Uy.

68
 PRES – PERSONALIZED EVALUATION SYSTEM IN A WEB COMMUNITY - A Conceptual Model Designed to
Evaluate Reputation in Order to Achive a Personalised View on the System for Each User

5 CONCLUSIONS Ross A. Malaga, 2004. Web-Based Reputation

Management Systems: Problems and Suggested
Solutions, Springer
Reputation system gives people information about
L. Mui, PhD thesis: Computational models of trust and
past activities of the other users. It can enhance an reputation: agents, evolutionary, games and social
on-line interaction environment by: helping people networks, MIT, 2002
decide who to interact, encouraging people to be Jennifer Golbeck, James Hendler: Reputation network
more honest, discouraging those who are not analysis for email filtering, for the 1st Conf. on Email
responsible from participating. The actions, the and Anti-Spam, 2004
behavior, the user preferences can be regarded as T. Berners-Lee et al. (eds.), Uniform Resource Identifiers
resources on which one can initiate interpretation (URI): General Syntax, RFC 2396, IETF, 1998
and processing mechanisms. PRES system allows Toby Segaran, 2007. Programming Collective
Intelligence, O’Reilly.
each user to have its own evaluation of the resources
Web-o-Trust effort by Russell Nelson, http://www.web-o-
and of the other users. The proposed metrics can be trust.org/.
used for implementation in real Web communities. del.icio.us. http://del.icio.us/.
In this work we have presented the basic dig.com. http://digg.com/.
elements of PRES model. For the future Slashdot. http://slashdot.org/.
development of the prototype we will perform a eBay. http://www.ebay.com/.
detailed analyzes of the system properties. LinkedIn. http://www.linkedin.com
In a real system the resources are changing in time. Everything2.http://everything2.com/.
This problem will be studied in our system thus Wikipedia. http://en.wikipedia.org.
foreseeing the possibility that the users can see and
change the given ratings.
Another problem related to the reputation
computation that will be studied is a complexity of
the algorithm of performing the entire calculus in the
system. The computation of WR and WU can be
easily performed for a proper number of resources
and users. For hundreds thousands of users and
resources we need a parallel algorithm to compute
periodically the WR and WU values.

REFERENCES
Amy N. Langville, Carl D. Meyer, 2006. Google's
PageRank and Beyond: The Science of Search Engine
Rankings , Princeton University Press
Hai Zhuge, The Web Resource Space Model, 2008,
Springer.
J. Golbeck, J. Hendler. 2006. Inferring Trust Relationships
in Web-Based Social Networks, ACM Transactions on
Internet Technology, 6(4).
J. Golbeck, J. Hendler. 2004. Accuracy of Metrics for
Inferring Trust and Reputation in Semantic Web-based
Social Networks, Proceedings of 14th International
Conference on Knowledge Engineering and
Knowledge Management, Northamptonshire, U.K.
Paolo Massa, Bobby Bhattacharjee, 2004, Using trust in
recommender systems: an extol analysis, in
Proceedings of iTrust 2004
O’Reilly, T., 2005. What is Web 2.0 – Design Patterns
and Business Models for the Next Generation of
Software. O’Reilly.
R. Falcone, S.Barber, L. Korba, M. Singh (Eds.), 2002,
Trust, reputation, and Security: Theories and Practice.
AAMAS 2002 International Workshop, Springer

69
POSTERS
 A STUDY OF THE EFFECTIVENESS OF “WAKE UP ON LAN”
AS A MEANS OF POWER MANAGEMENT

Colin Pattinson and Linton Robinson

Innovation North Faculty of Information and Technology, Leeds Metropolitan University, Leeds U.K.
{c.pattinson, l.a.robinson} @leedsmet.ac.uk

Keywords: Power management, wake on LAN, Energy savings in corporate IT.

Abstract: The growing awareness of the fragile nature of our environment, and of the damage that humankind is
causing to that environment, makes it necessary to review the environmental impact of all aspects of human
activity. One such area is the operation of corporate information technology (IT) systems. The growth in the
number and complexity of such systems over recent decades has led to a consequent increase in their power
requirements, to the point where for organisations of any size, “corporate IT” will be a major contributor to
the organisation’s overall energy consumption. A drive for overall reductions in consumption will, in the
words of the chair of the newly formed UK Environmental IT Leadership Team, mean “the IT department
becomes the focus of carbon reduction policies”. It is therefore timely to consider the present situation in
respect of power consumption within corporate IT systems, and to explore the potential avenues for
reducing that consumption. In this paper, we will briefly overview the area of power usage in IT, before
reporting on the outcomes of a specific project in which we explored the operation of the “wake on LAN”
method in a real situation.

1 INTRODUCTION them in a “dormant” state, like the “stand by” mode

of many consumer electronics devices. Note: in a
The rapid growth in computer use is placing a strain further parallel with consumer electronics,
on the UK’s ageing electrical infrastructure. The “dormant” is not “off”, and some power is still
growth of networked systems within organisations is required. The generic term for this activity is power
a major factor, with an increase in the number of management (PM). PM can bring about energy
servers in a typical organisation from around 4 in the savings “…of the 74 TWh/yr of electricity
late 1990s to 10 to 15 times that number (Watson, (approximately $6 billion per year) … consumed by
2006). In October 2006, there were 2,507,278 the Internet in the USA alone … 32% could be
registered companies in the UK (Companies House, saved with effective PM on desktops” (Christensen
2006). While we do not know exactly how many of et. al. 2007), and the equivalent of 250M litres of
these companies operated with the 60 servers in gasoline per day could be saved if 1M PC users
Watson’s analysis or even as many computers, it is employed PM (Long 2006).
not unreasonable to suggest that a good proportion Many current energy reduction activities
will have PCs, Servers and other network devices. (including the one reported here) aim at the desktop
Data for energy consumption of IT systems vary (user PC) element of IT systems, reasonably enough
according to the exact definition of what is included, in view of the numbers of such devices. However,
and there is variation in reporting methods. server and network device (switches, hubs and
Kawamoto et. al. (2000) found “devices connected routers) must not be overlooked. Allowing such
to the Internet” for “commercial use” were devices to enter a dormant state is problematic, as
responsible for 2% of the total U.S. power typical network performance and security
consumption. In Germany over 7% of domestic monitoring techniques expect networked devices to
energy consumption is “computer-related” respond to a regular sequence of polling packets;
(Loerincik 2006). indeed some network management tools generate an
One attempt to save energy has been to limit the alarm if a device does not react to a probe request;
power taken by devices when not in use by placing therefore there is a contradiction. Furthermore,

73
ICE-B 2008 - International Conference on e-Business

“[t]he immense increase in networking … is an Table 1: Alternative PM approaches.

important factor [limiting the use of power saving]
Dynamic Power Algorithms designed to shut
because network activity can keep computers from down a device only when an idle
Management
entering low-power modes” Webber et. al. (2001). period is of sufficient length to
(DPM)
justify performance degradation
Lu et. al. (2000) and state-transition energy.
2 APPROACHES TO PM A free tool which allows network
EZ Group Policy administrators to centrally
In the opening parts of this paper, we argued that Objects (GPO) control Power Management
there is a contradiction between the conventional settings using GPOs.
way networks operate, PM and cost savings. Let us
SMS 2003 currently lacks the
therefore take an overview of the different potential means for waking up machines, a
strategies in this regard. We start by giving a brief plug-in from 1E, Inc., called
explanation of each of these techniques (Table 1). SMS Wakeup, provides this
To some extent, this table is rather simplistic; SMS Wakeup functionality. SMS Wakeup
e.g. overlaps between some techniques could (Hobbs 2006), integrates with SMS and uses
combine to effectively provide a PM solution. In data from the SMS asset
this study, WOL has been chosen primarily because collection to obtain MAC
of its ease of use and simplicity in implementation. addresses for each client
computer.
Also, a previous research project successfully used
WOL in both wired and wireless scenarios, so we By adding a second low-power
have confidence in the underlying method. channel, it is possible to shut the
system off and reduce idle
Using a low-power power. Out-of-band control
channel information can be sent
3 QUESTIONS TO ADDRESS Shih et. al. (2002), simultaneously, to maintain
connectivity and wakeup the
The objective of this research was to determine how universal communicator (UCoM)
WOL operates; hence we begin by exploring how device when necessary.
WOL will be delivered in a typical networked Wake on LAN Wake on LAN is a Layer 2-based
scenario. Crucial to the deployment of WOL across (WOL) means for waking up machines
a network are questions of the nature and operation from sleep states such as system
of the communication process; for instance, are (Korn et. al., standby, hibernate and shutdown
WOL packets broadcast, unicast or multicast? What 2006). – and for remote access to them.
are the issues with regard to a ‘connectionless or
connection-oriented approach?’ and finally, what are
WOL encapsulates such Magic Packets inside a
the issues in terms of efficiency and security.
broadcast UDP packet. This gives benefits over raw
Ethernet frames including Operating System (OS)
interoperability.
4 THE MAGIC PACKET (MP)
The complete UDP packet, sent over an Ethernet
‘Magic Packet’ (MP) and ‘Wake up On LAN’ interface, looks like this: (Spurgeon 2006)
(WOL), are the same thing. The Magic Packet is [Ethernet header][IP header]
broadcast via port 7 or 9 which can be sent using a [UDP header][Magic sequence][CRCS]
variety of connectionless protocols, with UDP the
most common. (CapaInstaller, 2006). The Magic
Packet structure is similar to an Ethernet packet, but
the destination Medium Access Control (MAC) 5 TO BROADCAST, UNICAST OR
address is repeated 16 times within the packet. MULTICAST?
Unicast packets are sent from host to host, a one to
one relationship. Broadcast packets are one to many,
one host communicating with all other hosts.
Multicast is a restricted one to many relationship; a

74
 A STUDY OF THE EFFECTIVENESS OF “WAKE UP ON LAN” AS A MEANS OF POWER MANAGEMENT

Figure 1: Routed Network – MP Implementation.

single device can communicate with a specific set of traffic, reducing the potential benefits of PM.
hosts. Our previous work with the MP showed that Therefore one challenge for developers is to bridge
it can be transferred by broadcast, unicast and the gap between the command and the response.
multicast, however for the purposes of this paper, we
consider a ‘broadcast’ approach, minimising effort 6.3 Security
(just one message needs to be created), while
maximising coverage. By design, most modern There are a number of security issues within WOL
routers will block IP broadcast traffic and restrict it and its implementation (Robinson, 2007). For
to the local subnet, but this can be changed by the instance, applying the ‘layered approach’ of most
network administrator. network designs as in the following model:

((PM + current security) + (LAN security) +

6 WOL QUESTIONS (WAN security) + internet security))
= Total Vulnerabilities
6.1 Efficiency means that vulnerabilities are compounded. We also
found that “more advanced systems” do not
We conducted a series of tests to send the MP over a necessarily mean “more advanced PM” a further
wired network with unicast and multicast indication of the need for research and development
instructions to wake machines. In these tests, the into PM. Min and Chandrasakan (2003) argue that
packet successfully woke up and shut down computational algorithms and low power digital
workstation(s). When this was carried out using a hardware trade energy for quality, because digital
wireless link, we found that in 2 cases, the processing occurs more slowly and uses less energy
instruction to sleep succeeded, but the instruction to when circuit voltage is reduced.
wake failed if the target machine was in another
subnet. We believe that a correctly configured
wireless AP in that foreign subnet will rectify this 7 BROADCASTING THE MP
and conclude from these results that the MP is able
to produce satisfactory results and to do so reliably. OVER ROUTED NETWORK

6.2 Reliability In section 5, we identified different ways of sending

the MP. Most networks will be routed in some way.
It is well known that UDP is not reliable (Bhatti So we first consider if we can broadcast the MP.
2007); therefore the immediate question might be Figure 1 shows a routed network with two broadcast
why use it? Why not invoke a connection oriented domains (10.10.40.0 /24 and 172.16.0.0 /16),
protocol with reliable delivery? Once a command is randomly selected and intentionally kept distant
issued to sleep, there is currently no feedback from each other, (Class A and Class B respectively)
mechanism to indicate that command was received for ease in configuring the router.
and obeyed. Instead, the network administrator may It was possible to send the MP from A to B and
need to utilise other tools creating extra work and C, to shutdown and wake-up. We already knew that

75
ICE-B 2008 - International Conference on e-Business

A to B was possible, from results obtained in the Long C. (2006) Energy cost of PCs on standby. British
previous test; however, here we successfully showed Broadcasting Corporation (BBC). news.bbc.co.uk/
that the MP does route to another domain. Though a go/pr/fr/-/2/hi/programmes/click_online/4929594.stm
single router is used, in a network comprising of Lu Yung-Hsiang, Chung Eui-Young, Simuni’c Tajana,
Benini Luca (2000) Quantitative Comparison of
many routers, the MP should also be able to achieve Power Management Algorithms. Proceedings of the
the ends of Power Management, provided that conference on Design, Automation and Test in
routers and firewalls are configured to allow Europe, March 2000 pp. 20-26.
broadcast packets to travel between subnets (Korn Min R. Chandrakasan A (2003), Top Five Myths about the
et. al., 2006). (The tool used in these experiments is Energy Consumption of Wireless Communication
Prof Shutdown 3.2 evaluation version Mobile Computing and Communications Review Vol.
(profshutdown, 2006)). 7 No. 1 pp. 65 - 67 January 2003
Shih E., Bahl, P., Sinclair, M.J.(2002) Wake on wireless:
An event driven energy saving strategy for battery
operated devices. Proceedings of the Eighth Annual
8 CONCLUSIONS International Conference on Mobile Computing and
Networking (MOBICOM'02), September 2002.
This paper shows that PM can be supported by MP Robinson L. (2007) An Investigation into Power
without detriment to network performance. Our own Management and Power Savings: Is this achievable?
MSc Mobile and Distributed Computer Networks
research indicates that the adoption of PM take-up is
Dissertation, Leeds Metropolitan University July
at best ad-hoc, sparse and often a small project after 2007.
business critical objectives have been realised. We Rtsecurity (2006) www.rtsecurity.com/kb/tech/
believe this to be counterproductive, as buy-in by all wakeonlan.php (Accessed: 23/11/07)
stakeholders will itself contribute to the bottom line Spurgeon C. (2006) Wake on LAN. http://gsd.di.
of businesses and our world. Because the MP piggy- uminho.pt/jpo/software/wakeonlan/mini-howto/wol-
backs Ethernet, we further believe that the case for mini-howto-2.html
its deployment and the potential return on Watson J. (2006) Power shortage hits IT
investment (ROI) is significant. www.computing.co.uk/computing/news/2150433/pow
er-shortage-hits
Webber C. A., Roberson J. A., Brown R. E., Payne C. T.,
Nordman B., Koomey J. G., (2001) Field Surveys of
REFERENCES Office Equipment Operating Patterns. Energy Analysis
Program. Lawrence Berkeley National Laboratory.
CapaInstaller (2006) www.capainstaller.com/index.asp http://enduse.lbl.gov/info/LBNL-46930.pdf.
?page=157&page2=212&page3=269
Christensen, K., George, A.D. Nordman, B. (2007) The
Energy Efficient Internet Project. University of South
Florida. www.csee.usf.edu/~christen/energy/main.html
Companies House (2006) Verbal conversation
Hobbs C. (2006) Does SMS 2003 Support Wake-on-LAN?
http://www.faqshop.com/sms2003/sms2k3swd.htm
Profshutdown (2006) www.profshutdown.com/
(Accessed: December 2006)
Kawamoto, K., Koomey, J. G., Nordman, B., Brown, R.
E., Piette, M.A. Meier, A.K. (2000) Electricity Used
by Office Equipment and Network Equipment in the
U.S. Lawrence Berkeley National Lab document
LBNL-44917 August 2000
Korn D., Huang R., Bolioli T., Walker M. (2006)
Computer Power Management for Enterprises - A
Practical Guide. Proceedings of the 2006 IEEE
International Symposium on Electronics and the
Environment, May 2006 .pp. 161- 166
Loerincik, Y (2006) Environmental impacts and benefits
of information and communication technology
infrastructure and services, using process and input-
output life cycle assessment. PhD Thesis Ecole
Polytechnique Federale de Lausanne. September 2006.

76
 ALGORITHM AND AN ELEVATOR CONTROL SYSTEM
EXAMPLE FOR CTL MODEL UPDATE

Laura Florentina Cacovean

Department of Computer Science, Lucian Blaga University of Sibiu, Faculty of Sciences
Str. Dr. Ion Ratiu 5-7, 550012, Sibiu, Romania
[email protected]

Iulian Pah
Department of Sociology, Babes-Bolyai University Cluj-Napoca
Bd. 21 decembrie 1989, no. 120-130, 400604, Cluj-Napoca, Romania
[email protected]

Emil Marin Popa and Cristina Ioana Brumar

Department of Computer Science, Lucian Blaga University of Sibiu, Faculty of Sciences
Str. Dr. Ion Ratiu 5-7, 550012, Sibiu, Romania
[email protected], [email protected]

Keywords: CTL Kripke model, model update, algorithm, directed graph, implementation.

Abstract: In this paper is presented an update of the Computational Tree Logic (CTL) model checker. The minimal
modifications which appear represent the fundamental concept for model the dynamic system. In the paper
we use five primitive operations discompose from the operation of a CTL update used already by (Baral,
2005) which presented their approach of knowledge updated on the structures of single agent S5 Kripke.
Then we will define the criteria of minimal change for the CTL model update based on these primitive
operations. In the final section of this paper are presented the steps of implement the CTL model updated
and are described some details of algorithm implementation by applying the model update to the elevator
control scenario. The paper (Ding, 2006) is base of results obtained.

1 INTRODUCTION algorithms so that they can handle system

modification as well. The idea of integrating model
The verification tools o automated formal, such as checking and automatic modification has been
model checkers, shows delivered diagnosis to investigate in recent years. In work (Harris, 2003)
provide through automatic error diagnosis in the model checking is formalized often with an
complex designs, examples in (Wing, 1995). The updating operator satisfied the axioms U1-U8 what
current state of the model checkers technique, as represent the classical proposition knowledge of
Symbolic Model Verification (SMV) example updated Katsuno-Mendelzon postulates for belief
(Clarke, 2000), Cadence SMV (McMillan, 2002), update (Baral, 2005). They discussed knowledge
uses SMV as specification language for both CTL update and its minimal change, based on modal logic
(Computational Tree Logic) and LTL (Lineal S5. Both the update of the knowledge base and the
Temporal Logic) model checking. Progressing knowledge update are currently at the theoretical
update of the method of the model checkers, begun research stage. Their approach of knowledge update
to employ a formal method for repair approximate could integrate with model checking technology
error. Since model, checking can handle verification towards a more general automatic system
problems complex system and as it may, modification. In this paper, we considered the
implemented via fast algorithms, it is quite natural to problem of the update of CTL model from both
consider whether we can develop associated theories. In substance, as the traditional knowledge

77
ICE-B 2008 - International Conference on e-Business

based on the update (Winslett, 1990) consider an model M' = (S',R',F ') such that M'= (M',s0'),
update of CTL model subdue a principle of (M',s0')╞ f where s0'∈ S'. We use Upd(M, f) to
minimum inferior change. More, this minimal denote the result M' and Upd(M,f) =M if M ╞ f.
change are defined be as well to is definite as a
process based on of some operational processes
which a concrete algorithm for the update of CTL
model could be implemented. In the final section of
3 PRIMITIVE OPERATORS
this work, we present a study case where we shown
P1. Add an only relation. Given M = (S, R, F), its
how the system prototype (Ding, 2006) could be
applied for the system modified. updated model M' = (S', R', F ') is the result of M
having only added one new relation. That is S'= S,
F '=F, and R' = R∪{(sadd,sadd2)} where (sadd, sadd2)∉R
2 SYNTAX AND SEMANTICS for one pair of sadd,sadd2∈S.
P2. Remove an only relation. Given M = (S, R, F),
CTL is a branching time temporal logic meaning its updated model M' = (S', R', F ') is the result of M
that its formulas interpreted over all paths beginning having only removed one existing relation. That is,
in a given state of the Kripke structure. A Kripke S'= S, F ' = F, and R' = R-{(srem,srem2)} where (srem,
model M over AP is a triple M = (S,R, F:SÆ2AP) srem2)∈R for one pair of srem, srem2∈S.
where S is a finite set of states, R⊆S×S is a P3. Substitute a state and its associated with an only
transition relation, F:SÆ2AP is a function that relations. Given M = (S, R, F), its updated model M'
assigns each state with a set of atomic proposition. =(S',R',F ') is the result of M having only substituted
Syntax definition of a CTL model checker (Huth, one existing state and its associated relations. That
2000). A CTL has the following syntax given in is, S' = S[s/ssubst], R' = R∪{(si, ssubst), (ssubst,sj)|for
Backus near form: f ::⊤ |⊥ |p|(¬f1)|f1∧ f2| f1∨ f2| f1⊂ some si, sj∈S}-(si,s),(s,sj)|(si,s),(s,sj)∈R} and F '(s) =
f2| AX f1| EX f1| AG f1| EG f1| AF f1| EF f1|A[f1∪f2]| F (s) for all s∈S ∩ S' and F '(ssubst) = τ (ssubst), where
E[f1∪f2] where ‫׊‬p∈AP. τ is a truth assignment on ssubst.
A CTL formula is evaluate on a Kripke model M. P4. Add a state and it associated with an only
A path in M from a state s is an infinite sequence of relations. Given M = (S, R, F), its updated model M'
states from definition π = [s0,s1,… , si-1, si, si+1,…] = (S',R',F ') is the result of M having only added one
such that s0=s and (si,si+1) ∈ R holds for all i ≥ 0. We new state and it associated relations. That is, S' =
write (si, si+1)⊆ π and si∈ π. If we express a path as π S∪{saddst}, R'=R∪{(si, saddst),(saddst,sj)|si,sj∈S∩S'}
=[s0,s1,…,si,…,sj,…] and i<j, we say that si is a state and F '(s)=F(s) for all s∈S∩S' and F '(saddst)=τ (saddst),
earlier than sj in π as si < sj.
where τ is a truth assignment on saddst.
Semantics definition of a CTL model checker
P5. Remove a state and it associated with an only
(Huth, 2000). Let M = (S,R,F:SÆ2AP) be a Kripke
relations. Given M = (S,R,F), its updated model M'
model for CTL. Given any s in S, we define if a CTL
= (S',R',F ') is the result of M having only added one
formula f holds in state s. We denote this by (M,s) ╞
f. The satisfaction relation ╞ define by structural existing state and its associated relations. That is, S'
induction on all fourteen CTL formulas (Ding, =S-{sremst|sremst∈S}, R'=R-{(si, sremst),(sremst,sj) | for
2006). We assume all the five formulas CTL some si,sj∈S} and F '(s) = F (s) for all s∈S ∩ S'.
presented in the contextually as the paths are All the changes on CTL model can be in terms of
satisfied. Be a CTL Kripke model which satisfies the all five operations. It can be arguing P3 can be
CTL formulas and we considered as a model that defined in terms of P4 and P5. Anyway, we treat state
can be updated satisfying given formulas. The substitution differently from a combination of state
minimal change should define, based on some addition and state removed. That is the context,
operational process, a concrete algorithm for CTL whenever it substitutes a state needed, applied P3
model update that can be implemented. directly more than P4 followed of P5. This thing will
The CTL update definition: Be a CTL Kripke simplify definition of minimal change of the CTL
model M=(S,R,F) and a CTL formula f. An update model.
of M=(M,s0), where s0∈ S with f is a CTL Kripke For defined the criteria of minimal change of

78
 ALGORITHM AND AN ELEVATOR CONTROL SYSTEM EXAMPLE FOR CTL MODEL UPDATE

update CTL model, it needs to consider the changes control which including in first case, a process for
for both states and relations for the underlying CTL normal moving of lift cabin and in second case, for a
models. We achieve these specifying the differences faulty process. In first case for the normal moving
among states and relations on the models CTL using the elevator cabin process don’t appear with errors,
the primitive operations. Be any two sets X and Y, so the door is closed and the passenger going up or
symmetrical difference among X and Y be denoted down when the button is pressed. For the second
as Diff(X, Y) = (X - Y) ∪ (Y - X). Be two CTL process, the faulty process appears when the lift
models, M = (S, R, F), and M' = (S', R’, F ') for each cabin isn’t moving when the button is pressed for
primitive operation Pi with i = 1,…,5, Diff Pi(M,M') start the moving. The aim of the model is where the
indicates the differences between one of two the faulty process appears. The objective of model
CTL models where M' is a resulting model from M, updating, on other word, is to correct the original
that make clear this difference between this model, which contains the faulty process. Starting
operations the types may occur. Since P1 and P2 only from the original CTL structure for our propose EC
changes relations, we define DiffPi(M,M')=(R - R') system presented in the figure 1 with eight states
∪ (R'-R) where i = 1, 2. For the operations P3, P4 denoted with s1, s2,…, s7, and sd state we added for
and P5, we define DiffPi(M,M')=(S-S')∪(S'-S) with checking if the elevator is required of another
i=3,4,5. Although any state changes caused by P3, passenger.
P4, P5 will imply also correspondence changes on The Kripke model has eight states and the
relations, we only count the modifications states and propositional variables are from the set {Start,
take the state change as the primitive factor in order Close, Move, Error}. Start (St) represented the start
to measure difference between and M'. For the button for start moving up or down the elevator,
operations P3, we should consider the case, which a Close (Cl) represent the close door to the lift cabin,
state is substitute with a new state. For this is Move (Mv) is moving up or down the elevator and
necessary difference between these two states to be Error (Er) means occur some error.
minimal before the condition of formulated update. The formal definition of the Kripke structure of
A formal algorithm for the proposed CTL model EC is given by M=(S,R,F), where S={s1,s2,…, s7},
update approach is described in (Ding, 2006) and R={(s1,s2), (s2,s3), (s3,s2), (s3,s4), (s4,s3), (s4,s5),
(Cacovean, 2007). (s5,s6), (s6,s7), (s7,s7), (s7,s4), (s4,s1), (s1,sd), (sd,s4),
(sd,s1)}, AP={St, Cl, Mv, Er}. The F assigns state s1
in M with not start, not close, not move and not
4 ELEVATOR EXAMPLE error, write this as {¬St,¬Cl,¬Mv, ¬Er}. State
s2={St,¬Cl,¬Mv,Er}, s3={St,Cl, ¬Mv, Er},
In this section we present a study of case where it is s4={¬St,Cl,¬Mv,¬Er}, s5={St,Cl,¬Mv,¬Er},
illustrated the features of CTL model updated
s6={St,Cl,Mv,¬Er} and s7 ={¬St,Cl,Mv,¬Er}.
approaches.
The model shown hereinbefore:
As example, we shall present a scenario for an
elevator control system. The designer analyzes the
state-transition diagram for the only control
transformation, Elevator Controller (EC), finds eight
locked-state events (Gomma, 1993). These locked-
state events occur because the EC, in most instances,
takes one action and then awaits a response before
moving on a new state. In fact, have only two event
flow, Up and Down Request, when we denote with
Move state when the request exist and is not a
locked-state event. This event flows is not qualify
because each of them can arrive any time a client
presses a floor button or when the scheduler Figure 1: The CTL structure of Elevator Controller.
schedules an elevator. The remaining events can
only arrive when the EC is expecting them. In figure 1 START represented the start elevator,
We assume that we have an elevator system Open and Close represent the open door and close
the door, RESET is for a new initialization and

79
ICE-B 2008 - International Conference on e-Business

DONE represents the done moving of elevator. and the state s3′ is set {¬St,Cl,¬Mv, Er}.
The faulty process from this graph is the path [s1, The algorithm will generate one of the three
s2, s3, s4]. The interpretation is: start elevator {s1,s2}. resulting models without specific indication, because
In the state s2 we observed that have not close, that criteria used are satisfying all the minimally changes
is the door and it isn’t close, and the moving is out from the original model. We consider that our
of order and it pointed some error. Passed from the elevator model propose is a model much more
state s2 in the state s3 where the door elevator shall simple for understandable and for implemented,
be close. In the state s3 has error and the movement because we used a steps method to illustrate this
of elevator don’t start so it shall push the reset elevator controller. In our case we used the CTL
button for the reestablishment. That is, from s3 model checker update, verifying all five properties
passed to the state s4. Observed that the process with mentioned above which are accomplished also in our
normal move in the case view from the original CTL case of study.
Kripke structure through [s1, s4, s5, s6, s7]. Noticed
that this model do not satisfies the property f =
¬EF(St∧EG¬Mv) (Harris, 2003). The CTL model 5 CONCLUSIONS
updated brings a minimum modification of the
Kripke model which satisfies the property f. Firstly, In this paper, we presented a formal approach for the
it should analyze f in AG(¬(St∧EG¬Mv)) for update the CTL models. Specification of five
remove the symbol ¬. The translation is doing with primitives on the CTL Kripke models (Ding, 2006),
the function Upd¬. Then is necessary to check each define the minimal change criteria of the CTL model
state whether it satisfies ¬(St∧EG ¬Mv). This string updated. Also in this paper are presented semantics
shall be parsing before it is checked. Selecting the and the computing property of approach that we
EG¬Mv to elevator through the model checking used. The proposed case study is an update principle
of minimal change with maximal reachable states,
function for EG.
which can significantly improve the update results in
In this model, any path has any state when ¬Mv modification scenarios of complex system.
is selected. Here are searched the paths in the form
[s1,s2,s3,s4,s1,…] and [s1,s4,s1,…] which represent the
connected components loops satisfy EG¬Mv. Then
are identified all states with St, these are {s2,s3,s5,s6}.
REFERENCES
Then are selected the states with St and ¬Mv, these Baral C. and Y. Zhang, 2005, “Knowledge updates:
are {s2,s3}. Because the AG(¬(St∧EG¬Mv)) formula semantics and complexity issues”, Artificial
identifies the model don’t have the both states St and Intelligence, 164, 209-243.
¬Mv, is necessary an execution with states s2 and s3 Cacovean L., Popa E.M., Brumar C.I., 2007,
so it should apply the updated model. From Implementation of CTL Model Checker Update, in
Proc. 11th WSEAS Int. Conf., COMPUTERS, Greece
execution of UpdAG function, we shown the case in Clarke E.Jr., O. Grumberg, and D.A. Peled, 2000, “Model
which applying P3 on the state s2 and s3. The first Checking”, MIT Press, Cambridge
translate will be from ¬(St∧EG¬Mv) to Gomma H., 1993, “Software Design Methods for
¬St∧¬EG¬Mv, therefore s2 and s3 are updated with Concurrent and Real-Time Systems”, Addision-
Wesley Publishing Company, Reading Massachusetts
any ¬St or ¬EG¬Mv by the main function CTLUpd
Harris H. and M. Ryan, 2003, ”Theoretical foundations of
what is dealt with ∨ and with the Upd¬ function. In updating systems”, in Proc. 18th IEEE, 291-298.
other words, the new states of s2 and s3 shall be Huth M. and M. Ryan, 2000, ”Logic in Computer Science:
denoting with s2′ and s3′. The UpdAG(M,¬(St∧ Modelling and Reasoning about Systems”, Cambridge
EG¬Mv)) function calls the main function University Press.
McMillan K. and N. Amla, 2002, ”Automatic abstraction
CTLUpd(M,¬St) or CTLUpd(M, ¬EG¬Mv) for the without counterexamples”, in Cadence Berkeley Labs.
case f1 ∨ f2. We choose the ¬St because this is Wing J. and M. Vaziri-Farahani, 1995, ”A case study in
simplest than ¬EG¬Mv. In this case is necessary to model checking software”, in Proc. 3 ACM SIGSOFT.
update the St in states s2 and s3 of path π with ¬St Winslett M., 1990, ”Updating Logical Databases”,
Cambridge University Press, 1990.
instead, then no states on path π have the Ding Y., Yan Zhang, 2006, ”CTL Model Update:
specification EF(St∧EG¬Mv). M ′=(M′,s1)╞ ¬EF(St Semantics, Computations and Implementation”. ECAI,
∧ EG ¬Mv). The state s2′ is set {¬St,¬Cl, ¬Mv, Er} Italy.

80
 COMMON TEXTILE VOCABULARIES AND DOCUMENTS
A Conceptual Foundation of a Globally Interoperable Textile e-Marketplace

Jingzhi Guo and Zhuo Hu

Department of Computer and Information Science, University of Macau, Av. Padre Tomás, Pereira, S.J., Taipa, Macau
[email protected], [email protected]

Keywords: Textile electronic marketplace, business vocabulary, business document, product data integration, vocabu-
lary integration, document integration, semantic consistency maintenance, semantic interoperability, elec-
tronic commerce, electronic business.

Abstract: This paper has proposed a novel common textile vocabulary and document framework (TexVDF) in a col-
laborative network to enable cross-domain level business information sharing and business document ex-
change in a semantically consistent way. The approach to this framework is motivated through presenting
some real-world examples of business inquiries with product specifications. By these examples, two prob-
lems are detected on how to achieve semantic commonality between cross-domain level business vocabular-
ies for textile e-Marketplace mediators and how to allow specificity of cross-domain level common business
document templates for local textile e-Marketplace mediators yet still maintaining semantic consistency. To
solve these two problems, this paper has firstly reviewed CONEX technologies relevant to the newly devel-
oped TexVDF approach, which includes a TexVDF framework, a P2P collaborative textile concept mapping
model and a textile business document template model. These two models have been demonstrated by ex-
amples to see how they should work.

1 INTRODUCTION Making textile information understandable be-

tween participants on e-Marketplace is an important
In global textile market, many textile firms face information interoperability issue (Guo 2007). It
great challenges of increasing global competition to requires building a globally interoperable textile e-
cope with quicker responsiveness of competitors Marketplace by integrating heterogeneous textile
(Teng et al. 2006), better global partnership, and information systems of all participated firms. This at
advancing technology for manufacturing and dy- least involves two aspects: the integration of busi-
namic consumer demand (Ostic 1997). To meet ness vocabulary used by all textile firms and the
these challenges, textile industry needs to build integration of exchanged business documents such
global e-Marketplaces, where producer, supplier, as inquiries, offers, counteroffers and orders.
and retailers can efficiently communicate and ex- This paper aims to propose a novel common tex-
change information to strengthen their competitive- tile vocabulary and document framework (TexVDF)
to semantically integrate complex textile firms to
ness. However, textile industry is complex. It in-
enable them to participate in textile e-Marketplace,
volves raw materials such as cotton, silk and poly-
and also to facilitate the proposed approach as the
mer that produce both natural and synthetic fibres, foundation of the future design of globally interop-
which again are converted into many kinds of fab- erable textile e-Marketplace.
rics and finally to become products such as carpet The rest of the paper will be arranged as follows:
and apparel. Section 2 will provide a motivational scenario to
Besides, textile industry has thousands of large, raise our discussion issues. Section 3 will briefly
medium and small sized retailers and manufactures introduce the relevant technologies and propose a
(Teng et al. 2006). They vigorously trade with each new TexVDF approach to lay a solid conceptual
other. This indicates that a global e-Marketplace in foundation for future design of textile e-
design must be able to manage the flow of millions Marketplace. Section 4 will exemplify TexVDF
of textile products between a very large number of approach. Related work is discussed in Section 5.
firms. This requires the information exchanged by Finally, the conclusion with a contribution list of this
computers be understandable by all textile partici- paper is given, together with the required future
pants. work.

81
ICE-B 2008 - International Conference on e-Business

2 A MOTIVATIONAL SCENARIO This problem can be abstracted as a research is-

sue of cross-domain level business vocabulary
e-Marketplace has four properties: distribution, commonality, which states that all levels of textile
autonomy, interdependence and emergence (Guo industry shall be able to semantically communicate
2007). These also apply to textile e-Marketplaces. with each other via a common vocabulary model.
However, textile e-Marketplaces have a more spe-
Table 4: Dyed Fabric Inquiry from Fabric Industry.
cific property of levelled-domain interdependence,
i.e. one level of textile e-Marketplace vocabularies Concept Concept Value
and documents are tightly-coupled with another Name Fabric
level of textile e-Marketplace vocabularies and Type Combed
Colour Yarn dyed
documents. Since they belong to different domains, Technique woven
they have specific requirements, For example, they Style Jacquard
involve interdependence between levels of industries Usage Garment
of raw supplies (e.g. cotton, silk and polymer), fibre, Composition
Cotton 60%
yarn, thread, fabric, printing and dyeing, and gar- Polyester 40%
Construction 45 s × 45 s
ment.
Density 130 × 70
To motivate the problem to solve, Table 1 to 5 Width 57/58”
summarize the inquiries created by different levels
of textile industries to show their interdependence. Table 5: Men’s Coat Inquiry from Garment Industry.
Concept Concept Value
Table 1: Raw Cotton Inquiry from Material Industry.
Name Coat
Concept Concept Value Fabric
Product name Indian raw cotton Cotton 60%, ring, spun
Type Shankar-6 Polyester 60%
Staple length 29 mm OR 28.5 mm Weight 300 grams/square meter
Strength 29 gpt or 28 GPT minimum Specification
MIC 3.5-4.9 Zipper Front
Pocket Pouch
Cuff rib, lycra (cotton 60%, polyester 40%)
Table 2: Polyester Fibre Inquiry from Fibre Industry. Bottom rib, lycra (cotton 60%, polyester 40%)
Concept Concept Value Size S, M, L, LL, LLL, LLLL
Name Polyester Stable Fibre Drawcord Fabric
Colour Semi-dull raw white Use Men
Actual Denier 1.40 + / -0.05 (DE)
Dry Tenacity 6.50 + / -0.05 (G/D) This problem can be abstracted as a research is-
Dry Elongation 30. + / -1.5 sue of cross-domain level business vocabulary
Crimp number 13.00 + / -3.00 EA / Inch commonality, which states that all levels of textile
Degree of Crimp 13.00 + / -0.39 industry shall be able to semantically communicate
Shrinkage 7 + / -0.07 with each other via a common vocabulary model.
(2) Inquiry templates, as shown in Tables, from
Table 3: Blended Yarn Inquiry from Yarn Industry. different levels of industry are different in syntactic
Concept Concept Value forms and semantic use of terms, though they are all
Name Yarn called as inquiry sheets. This phenomenon implies
Colour white that a same type of business documents must be
Type Carded treated differently in specific domain-level industry.
Cotton 60%
Composition
Polyester 40%
We abstract this phenomenon as a research issue
Count 32 s of cross-domain level common business document
Quality high specificity, which states that all domain-levels of
Application Quality knitting and weaving fabrics textile industry shall be able to personalize docu-
An analysis to the above Tables reveals at least ment templates from common document templates.
two of the following problems:
(1) Term usages are not only industry domain-
specific. The higher-level industry has a high fre- 3 TEXVDF APPROACH
quency to use the vocabularies of lower-level indus-
try, but practically each level of industry has its own In this section, we will propose a novel common
explanations on their terms in vocabularies. Textile Vocabulary and Document Framework
(TexVDF) to solve the above two problems to lay a

82
 COMMON TEXTILE VOCABULARIES AND DOCUMENTS - A Conceptual Foundation of a Globally Interoperable
Textile e-Marketplace

solid foundation for future globally interoperable theory (Guo 2008). Its implementation is specified
textile e-Marketplace. We vision that this e- in CONEX Grammar (Guo 2008).
Marketplace will be a common textile information The information exchange by PM from one local
space that all kinds of services and collaborative firm (L) to another local firm (L) follows a concept
activities will be enabled, based on our designed supply chain (Guo 2008), such that:
TexVDF solution. Concept(L1) map onto Concept(M1) map onto
In the following, we will first discuss the rele- Concept(M2) map onto Concept(L2),
vant technologies and then propose the solution where concept(L1) of L1 finally arrive at L2 as
framework and its details. concept(L2).
Since concepts of L1, M1, M2 and L2 are all col-
3.1 Relevant Technologies laboratively created and mapped, there is no seman-
tic inconsistency between them in theory. Thus,
In CONEX research (Guo 2008), a generic e- CONEX e-Marketplace provides an approach of
Marketplace is designed like a multi-sons solar sys- accurate yet automatic information exchange in the
tem as a set of common collaborative service media- perspectives of all users (U) of L.
tors, each having a set of local collaborative service
designers for their own service users on CONEX 3.2 TexVDF Framework
Network (ConexNet). It can be shown in Figure 1.
The TexVDF framework follows the design of
P2P Collaboration
U L L U CONEX e-Marketplace but adds the new layer
Use M M Use
U L L U thought to the framework. To provide a smooth
Use Use discussion, we illustrate this framework in Figure 2,
U L L U where different levels of textile industry have been
Use M M Use
U L L U separated but integrated in a coherent collaboration
D2F Collaboration D2F Collaboration framework.
U: User L: Local Designer M: Mediator
In Figure 2, CONEX e-Marketplace has been ex-
tended to include six layers of textile e-
Figure 1: CONEX E-Marketplace. Marketplaces, which are:
• Raw material e-Marketplace, which focuses on
In this CONEX e-Marketplace, business vocabu- trading cotton, silk, wool, fur, feather, plant,
laries and documents are firstly collaboratively de- polyester, polypropylene, polyethylene, etc.
signed at mediators (M) on a P2P collaborative net- • Fibre e-Marketplace, which focuses on trading
work. Mediators include designer roles of common fibres made from cotton, silk, polymer (e.g.
vocabulary design, dictionary design and document polyester, Dacron, nylon, Micron), plant (e.g.
template design. These designers are knowledge corn, soybean), etc.
experts (or knowledge engineers) and can make sure • Yarn and thread e-Marketplace, which focuses
that things go right. Thus, they are called dominators on trading yarns of cotton, polyester, blended
(D) when their results are sold to local firms. The polyester, wool, plant, different threads, etc.
local firms subscribing or purchasing mediators’ • Fabric e-Marketplace, which focuses on trading
design services localizes these services into their woven and non-woven fabrics of cotton, polyes-
ter, blended polyester, metallic, etc.
own personalized forms, that is, local business vo-
cabularies and document templates. The local de- • Printing and Dyeing e-Marketplace, which
signers are knowledge workers but are not experts. focuses on trading dye, dyeing services of vari-
Thus, they are followers (F) of the dominators (D). ous yarns, threads, fabrics, printing, etc.
The collaboration between dominators and followers • Garment e-Marketplace, which focuses on fin-
is a dominator-to-follower (D2F) relationship on a ished textile products like jacket, pullover, shirt,
point-to-point communication network. The follower protective garment, gown, gloves, etc.
cannot modify the common design but follow to
generate their own. When local designers have de- By this layered classification of textile e-
signed local information about firm-based vocabu- Marketplaces, a new technology of P2P collabora-
lary and document templates, their users of their tive mapping between e-Marketplace mediators is
own firms can then automatically exchange business developed to solve the problems stated in Section 2.
information, such as inquiries and offers. This technology complements the previous CONEX
In CONEX e-Marketplace, vocabularies and technology to enable cross-domain level vocabulary
documents are designed following Product Map (PM) and document interoperability.

83
ICE-B 2008 - International Conference on e-Business

Garment Local use Local D2F Garment

E-Marketplace User Designer Mediators
P2P Colla-
Printing & Local use Local D2F Printing & borative
Dying User Designer Dying Mapping
Network
Fabric Local use Local D2F Fabric
Standard
E-Marketplace User Designer Mediators
Terminolo-
Vocabularies gy
Yarn & Thread Local use Local D2F Yarn & Thread Document Templates Servicing
E-Marketplace User Designer Mediators

Fibre Local use Local D2F Fibre P2P Collabo-

E-Marketplace User Designer Mediators ration Net-
work
Raw Material Local use Local D2F Raw Material
E-Marketplace User Designer Mediators

Figure 2: TexVDF Framework.

A standard terminology servicing centre is added mous with “handkerchief linen”; and “drab” is a
to enable the adoption of existing standard terms homonym with meanings of “colour of a moderate
such as standard measure, currency code, and de to grayish or light grayish yellowish brown or light
facto standard use of textile terms. This service will olive brown” and “cloth of a light dull brown or
reduce the collaborative work effort occurred in both grayish brown or unbleached natural colour, espe-
vocabulary and document templates for cross do- cially a heavy woollen or cotton fabric”.
main levelled textile e-Marketplaces. To bind the concept expressions of individual
Particularly, the TexVDF framework adopts a mediators with a common concept expression, the
collaborative mapping solution to achieve semantic mediators’ concept expressions must be framed in a
commonality of cross-domain level vocabularies and common structure, where they can freely express
a specificity solution to using common textile docu- their ideas of categorizing their individual textile
ment template. concepts. CCM provides this function in the follow-
In the next two subsections, we will elaborate the ing form, such that:
design of these two solutions.
Definition 1 (CMM). Collaborative Mapping
Model
3.3 P2P Collaborative Mapping
Given a set of individual concept expressions σ1,
A P2P collaborative mapping model (CMM) is a σ2, ..., σn and uniquely identified by iid1, iid2, ..., iidn
solution to enabling various mediators of levelled under different mediators’ contexts x1, x2, ..., xn
textile e-Marketplaces to collaboratively work to- about a piece of common formal concept FC, de-
gether to map their textile concepts in a common fined by AN, uniquely identified by IID, typed by
level but with possible expression variations for their CT, inherited from IF, concept or context of CO, and
own. This model is necessary because there are possibly referenced to (RT) a standard term of a
many synonyms and homonyms appeared in textile terminology, such that:
industry. Each textile mediator has its own concept Γ |- concept[IID, FC, AN, CT, IF, CO, RT] ⇒
mediation context and is not possible to define all map[concept[IID, iid1, σ1)@x1], ..., concept[IID,
synonyms and homonyms for each textile concept. iidn, σn]@xn],
The impossibility for each textile mediator to in-
clude all is due to the complexity that textile terms where, “Γ” is a P2P collaborative mapping en-
are developed from different contexts, where syno- gine that enables all mediators to work together in
nyms and homonyms come from various textile real-time, “|-”is a declaration that all mapping follow
types, usages of dated and obsolete terms, term deri- the mapping structure that is defined, and “⇒” refers
vations, languages of origin, and toponyms (Leech to that the semantics of each mediator’s concept
1999). strictly follows the semantics of common concept
For example, “baby combing wool” has variation for all.
of “French combing wool”; “linen lawn” is synony- The above CMM has following principles:

84
 COMMON TEXTILE VOCABULARIES AND DOCUMENTS - A Conceptual Foundation of a Globally Interoperable
Textile e-Marketplace

(1) Each mediator’s semantics about the same con- fabric”. The inherited terms of “ramie” and “cotton”
cept must be semantically consistent by P2P are useful to infer the concept meaning.
collaboration following CMM structure. Concept type CT. It defines in which term type
(2) Personalized concept expressions of individual the concept applies. For example, “bombazine”
mediators, rather than the formal concept FC, means “a twilled or corded dress-material, composed
are allowed to adapt to the local mediator’s e- of silk and worsted; sometimes also of cotton and
Marketplace environment but have to be worsted, or of worsted alone”. Thus, it belongs to
mapped onto common concept via IID. the domains of “silk” and “cotton” as a noun-form,
(3) Standard terms of terminology, from standard but it also belongs to the large domain of “textile” as
terminology servicing centre shown in Figure 1, an adjective form to refer to “worsted”.
are encouraged to be referenced during common
With above descriptions, the common concept
concept design.
concept[IID, FC, AN, CT, IF, CO, RT] is clear for
Particular to CMM, some details are important to
individual mediators to collaboratively map onto
mention.
their own local forms in map[concept[IID, iid1,
Definitional annotation AN. It is a full definition
of a common concept, and is not a single word or σ1)@x1], ..., concept[IID, iidn, σn]@xn]. In the local
phrase that may not fully capture the meaning of the form, practically, the context xi can be designed as
concept, or may lead to sense ambiguity. an individual mediator’s unique identifier (LID) to
Formal concept FC. It is a machine-readable refer to its specific context.
term about a common concept. It may capture full or
almost full meaning of the concept definition AN, 3.4 Document Template Model
but not guaranteed for accuracy. It is primarily used
for information retrieval for search services. FC is A textile document template model (Doclate) is a
typed as a set to include multiple words and phrases solution to enabling specifying business document
with exact or similar meanings to the defined con- templates to a domain-specific level but still being
cept such as abbreviation. able to utilize the common document templates
Internal unique concept identifier IID. It is prepared in advance and to consistently use common
unique identifier of AN with semantic causal order concepts designed by CMM.
To realize this model, this paper regards a busi-
relationship such as AN ⇒ IID. It implies that any
ness document template a set of hierarchical con-
IID cannot be created to use without the meaning of
cepts and improves the previous work of CODEX
AN conveyed in IID. If such happens, it will be
(Guo 2006) by removing P2P collaboration require-
prohibited to process.
ment and adding RT to CMM, such that:
Reference to a concept RT. It is a reference to a
semantically equivalent concept, often a well- Definition 2 (Doclate). Document Template Model
defined term in terminology or an already-defined Given a set of concepts defined in CMM, there is
vocabulary by CMM. It is typed as a namespace use. a common document template (com) with a set of
Context of a higher level concept CO. It defines concepts (called elemon) hierarchically identified as
the direct context of the current concept in a vocabu- IID, annotated by AN, in the context of CO, pre-
lary hierarchy. For example, given “domestic appli- sented as DP, referenced to RT of CMM, and have
ances (domestic refrigerators)”, the “domestic appli- occurrence OC; each “element” has a reification
ances” is the direct context of “domestic refrigera- structure “value” to reify the concept with represen-
tors”. It is similar to a broad term (BT) in relation to tation format PT, data type DT and possible function
a narrow term (NT) in controlled vocabulary (Fidel FN for managing reification. This “com” will further
1999). The CO is important for efficient and accu- be extended to (⇒) “loc” to provide specificity of
rate information exchange. It is also very useful for “com”, such that:
accurate machine translation by word sense disam-
Γ |- com: elemon[IID, AN, CO, DP, OC,
biguation (Vickrey et al. 2005) through CO context.
RT](value[PT, DT, FN]) ⇒ loc: elemon[IID, AN,
Inheritance from sources IF. It defines that
CO, DP, OC, RT](value[PT, DT, FN]),
where the concept is inherited or derived. It states
the origin of the concept. It is useful for improving where “Γ” is a business document template edit-
information retrieval and disambiguating the sense ing engine, “|-” declares that the editing follows the
of the concept when the concept is applied in ma- structure defined.
chine translation. For example, “ramie cotton The above model has the following principles:
blended fabric” is inherited from both “ramie” and (1) Each common document template “com” is only
“cotton” and under the context_of (CO) of neither a semi-finished template and could be further
“ramie fabric” nor “cotton fabric” but “blended

85
ICE-B 2008 - International Conference on e-Business

specified in any way to “loc”, but both must fol- 4 COAT EXAMPLE ON TEXVDF
low the Doclate structure model.
(2) Both “com” and “loc” can only create “elemon”
In this section, we demonstrate TexVDF approach
concepts through RT to CMM vocabularies.
through a garment inquiry specified in Table 5. The
Particularly, the Doclate structure elements have the following examples will adopt the written form of
following semantics: vector concept tree like “1.i...i” to represent the
AN, IID, CO and RT. They exactly have the same concept hierarchy of any vocabulary and document
interpretation as in CMM, but RT refer to CMM. template (Guo 2008).
Document concept display phrase DP. It is a
phase for visual display to represent concept. For 4.1 CMM for Coat Vocabularies
example, a concept of “an appliance, a cabinet, or a
room for storing food or other substances at a low To implement the garment inquiry example, we first
temperature” could be visually displayed as “refrig- design the common vocabulary, shown in Table 6,
erator” or “domestic refrigerators” as needed in the based on CMM of definition 1.
different designs of Doclate template. With the above collaboratively designed garment
Concept occurrence OC. It defines occurrence of vocabulary commonly for all levels of textile e-
concept in designed Doclate template. The concept Marketplace mediators, individual mediators can
occurrence may happen. For example, “product item” localize their personalized vocabularies applicable to
in a purchasing order may occur many times for their own e-Marketplaces. Table 7 shows the local-
different purchased items. ized mapping for one English mediator x1 and a
Value concept VALUE. This is a reification sym- Chinese mediator x2 for their own e-Marketplaces.
bol to introduce a reification of a concept to a par-
ticular concept, for example, “colour” → “red”. Table 7: Mapping of Local Vocabulary onto Common
Presentation style of reified concept PT. It de- Vocabulary.
fines how a reified concept should be displayed. For
map[iid=“1.1”, (iid1=“aa” an=“cotton”)x1, (iid2=“111” an=“棉花”)x2]
example, “1” could be displayed as “1”, “one”, or map[iid=“1.2”, (iid1=“ab” an=“polyester”)x1, (iid2=“112” an=“聚酯”)x2]
“USD1/piece”. It is a logic module and implemented map[iid=“2”, (iid1=“b” an=“fabric”)x1, (iid2=“22” an=“面料”)x2]
in a remote namespace. map[iid=“2.1”, (iid1=“ba” an=“ribbing”)x1, (iid2=“221” an=“针织布”)x2]
map[iid=“3”, (iid1=“c” an=“weight”)x1, (iid2=“33” an=“重量”)x2]
Data type of reified concept DT. It defines the map[iid=“4”, (iid1=“d” an=“coat”)x1, (iid2=“44” an=“上衣”)x2]
data type of the reified concept, for example, “string” map[iid=“4.1”, (iid1=“da” an=“zipper”)x1, (iid2=“441” an=“拉链”)x2]
or “decimal”. map[iid=“4.2”, (iid1=“db” an=“pocket”)x1, (iid2=“442” an=“口袋”)x2]
Operational function of reified concept FN. It de- map[iid=“4.3”, (iid1=“dc” an=“hood”)x1, (iid2=“443” an=“帽子”)x2]
map[iid=“4.3.1”, (iid1=“dca” an=“drawcord”)x1, (iid2=“4411” an=“拉绳”)x2]
fines how the reified concept value could be com- map[iid=“4.4”, (iid1=“dd” an=“cuff”)x1, (iid2=“444” an=“袖口”)x2]
puted, for example, automatically generated reified map[iid=“4.5”, (iid1=“de” an=“bottom”)x1, (iid2=“445” an=“下摆”)x2]
“date” value, reified computational group concept map[iid=“5”, (iid1=“e” an=“size”)x1, (iid2=“55” an=“尺寸”)x2]
map[iid=“6”, (iid1=“f” an=“used by”)x1, (iid2=“66” an=“适用”)x2]
“total”, or a result of a logic module. map[iid=“7”, (iid1=“g” an=“cuff”)x1, (iid2=“77” an=“规格”)x2]
With the above descriptions, common document
templates could be easily specified locally but accu- Given the above local mapping onto the common
rately maintain semantic consistency between using vocabulary shown in Table 6, the users of local
parties of different levels of textile e-Marketplace. textile e-Marketplace of x1 and x2 can then ex-
change business information without any semantic
Table 6: Textile Common Vocabulary for Garment. conflicts.
concept[iid=“1.1” fc=“cotton” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
concept[iid=“1.2” fc=“polyester” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
concept[iid=“2” fc=“fabric” an=“-” ct=“n” if=“-” co=“-” rt=“-”]
4.2 Doclate for Coat Inquiry
concept[iid=“2.1” fc=“rib” an=“knit ribbing” ct=“-” if=“-” co=“-” rt=“-”]
concept[iid=“3” fc=“weight” an=“-” ct=“-” if=“-” co=“-” rt=“-”] In this part, we exemplify the work of Doclate mod-
concept[iid=“4” fc=“coat” an=“-” ct=“-” if=“-” co=“-” rt=“-”] el in the example of Table 8 and Table 9.
concept[iid=“4.1” fc=“zipper” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
concept[iid=“4.2” fc=“pocket” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
concept[iid=“4.3” fc=“hood” an=“-” ct=“-” if=“-” co=“-” rt=“-”] Table 8: Common Document Template for Inquiry.
concept[iid=“4.3.1” fc=“drawcord” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
elemon[iid=“1” an=“” co=“” dp=“coat” oc=“” rt=“4”]
concept[iid=“4.4” fc=“cuff” an=“-” ct=“-” if=“-” co=“-” rt=“-”]
elemon[iid=“1.1” an=“-” co=“” dp=“fabric” oc=“” rt=“2”]
concept[iid=“4.5” fc=“bottom” an=“-” ct=“garment” if=“-” co=“-” rt=“-”]
elemon[iid=“1.2” an=“” co=“” dp=“size” oc=“” rt=“5”]
concept[iid=“5” fc=“size” an=“-” ct=“-” if=“-” co=“-” rt=“msr:123456”]
elemon[iid=“1.3” an=“” co=“” dp=“use for” oc=“” rt=“6”]
concept[iid=“6” fc=“use for” an=“-” ct=“-” if=“-” co=“-” rt=“hba:12345”]
elemon[iid=“1.4” an=“” co=“” dp=“specification” oc=“” rt=“7”]
concept[iid=“7” fc=“specification” an=“-” ct=“-” if=“-” co=“-” rt=“spc:1”]

86
 COMMON TEXTILE VOCABULARIES AND DOCUMENTS - A Conceptual Foundation of a Globally Interoperable
Textile e-Marketplace

In this simple document template, some content apparel, textile and fibre within a particular supply
of the garment inquiry sheet has been designed. It chain. While supply chain method for textile e-
then can be further defined by local e-Marketplace Marketplace integration is worth investigating, it has
designers of document templates. The function of some entry limitations for small and medium sized
common level document templates is to reduce the enterprises to participate in.
time of local e-Marketplace document designers and In Europe, MODA-M (MODA-ML; Leech 1999)
thus to lower the e-Marketplace cost. is a research on textile e-Marketplace. Its approach
When local e-Marketplace designers obtain the is based on the exchange of standardized XML
rough document templates, they personalize them as documents, where ebXML protocol has been
their own needs, as shown in the example Table 9. adopted to transfer XML messages. Moda-ML is
ontology-based in design. It generates a modular
Table 9: A Particular Garment Inquiry Sheet Template. ontology where each basic concept can be managed
elemon[iid=“1” an=“” co=“” dp=“” oc=“coat” rt=“4”] independently from the others and is identified by its
elemon[iid=“1.1” an=“-” co=“” dp=“fabric” oc=“” rt=“2”] own namespace (Gessa 2007). It has defined a
elemon[iid=“1.1.1” an=“” co=“” dp=“weight” oc=“” rt=“3”] common platform (Gessa et al. 2004), which at-
elemon[iid=“1.1.2” an=“” co=“” dp=“cotton” oc=“” rt=“1.1”]
elemon[iid=“1.1.3” an=“” co=“” dp=“polyester” oc=“” rt=“1.2”] tempts to be adopted by firms to improve their inter-
elemon[iid=“1.2” an=“” co=“” dp=“size” oc=“” rt=“5”] operability. Ontology-based ebXML document ex-
elemon[iid=“1.3” an=“” co=“” dp=“use for” oc=“” rt=“6”] change is an attractive approach and fits in most
elemon[iid=“1.4” an=“” co=“” dp=“specification” oc=“” rt=“7”]
elemon[iid=“1.4.1” an=“” co=“” dp=“zipper” oc=“” rt=“4.1”]
research prototypes in many other e-Marketplaces.
elemon[iid=“1.4.2” an=“” co=“” dp=“pocket” oc=“” rt=“4.2”] However, its success depends on the ontology inter-
elemon[iid=“1.4.3” an=“” co=“” dp=“hood” oc=“” rt=“4.3”] operability by its own in semantic level.
elemon[iid=“1.4.3.1” an=“” co=“” dp=“drawcord” oc=“” rt=“4.3.1”] It is necessary to make a clear distinction be-
elemon[iid=“1.4.4” an=“” co=“” dp=“cuff” oc=“” rt=“4.4”]
elemon[iid=“1.4.5” an=“” co=“” dp=“bottom” oc=“” rt=“4.5”]
tween the three general strategies of integrating
elemon[iid=“1.4.1” an=“” co=“” dp=“zipper” oc=“” rt=“4.1”] heterogeneous business information (Guo 2008).
They are mandatory standardization (a standard is
In the above Table, a more specified garment in- enforced for all participants, in which heterogeneous
quiry sheet template has been designed. With this information integration between standard systems
document template, users of local textile e- and the participants’ local systems is the task of the
Marketplace can reify the template by filling the local participants), automated mediation (an intelli-
“value” information to automate inquiry exchange. gent agent as a mediator to mediate heterogeneous
information between disparate participants’ systems
based on predefined rules, in which if no rules can
5 RELATED WORK be applied, mediation of heterogeneous concepts is
not accurate), and collaborative conceptualization
Textile e-Marketplace design that enables business (heterogeneous business concepts between disparate
information sharing and business document ex- local participants’ systems must be collaboratively
change is an important research field. In this design, mapped on a higher level common system before
semantic integration of textile vocabularies and they can be exchanged).
documents are the foundation for a globally interop- This paper adopts collaborative conceptualiza-
erable textile e-Marketplace. Currently, active re- tion strategy to avoid the weakness of mandatory
searches can be found in DAMA and Moda-ML. compliance of standards by users and inaccurate
In U.S., TEXNET is a textile industry data- concept mediation by intelligent mediator. Under
sharing network to address data sharing among busi- this strategy, mediator is designed as a collaborative
ness partners. It presents shared data in the screen or mediator to ensure the semantic consistency between
saved it in standard formats on a local platform heterogeneous concepts. This strategy absorbs some
(Lovejoy, a). Particular to the textile e-Marketplace, of the merits from both mandatory standardization
DAMA (Chapman et al. 2000) is a project of such and automated mediation strategies, but it creates no
type. DAMA applies a pipeline analysis method standards and mixes automated agents with human.
(Lovejoy, b). In DAMA’s research, supply chain
concept is adopted through Supply Chain Integration
Program (SCIP), where an inter-enterprise decision 6 CONCLUSIONS
support tool is developed to analyze supply chain
tradeoffs. DAMA is designed to use TEXNET for This paper has proposed a novel common textile
data sharing. It intends to support information shar- vocabulary and document framework (TexVDF) in a
ing and decision making between firms of retail, collaborative network to enable cross-domain level

87
ICE-B 2008 - International Conference on e-Business

business information sharing and business document ACKNOWLEDGEMENTS

exchange in a semantically consistent way. The
approach to this framework is motivated through The work reported in this paper has been supported
presenting some real-world examples of business by University of Macau Research Grand.
inquiries with product specifications. By these ex-
amples, two problems are detected on how to
achieve semantic commonality between cross-
domain level business vocabularies for textile e- REFERENCES
Marketplace mediators and how to allow specificity
of cross-domain level common business document Censoni, P., De Sabbata, P., Cucchiara, G., Vitali, F.,
Mainetti, L. and T. Imolesi, 2002. MODA-ML: a Ver-
templates for local textile e-Marketplace mediators
tical Framework for the Textile-Clothing Sector based
yet still maintaining semantic consistency. To solve on XML and SOAP. In Challenges and Achievements
these two problems, this paper has firstly reviewed in E-business and E-work. IOS Press.
CONEX technologies. Then, TexVDF approach is Chapman, L. and M. Petersen, 2000. Demand Activated
presented in a TexVDF framework, which is an Manufacturing Architecture (DAMA) Model for Sup-
improvement of previous CONEX model. To realize ply Chain Collaboration. In: International Conference
TexVDF framework, a P2P collaborative concept on Modeling and Analysis of Semiconductor Manu-
mapping model and a textile business document facturing.
template model have been developed. The former Fidel, R., 1999. Searchers' selection of search keys: II.
Controlled vocabulary or free-text searching. Journal
has resolved the problem of semantic commonality
of the American Society for Information Science 42(7)
of local mediators’ individual vocabularies, and the 501-514.
latter has solved the problem of designing common Gessa, N., 2007. An ontology-based approach to define
business document templates and their flexible and manage B2B interoperability. Technical Report
specificity to document templates adaptable to local UBLCS-2007-11. Department of Computer Science,
textile e-Marketplaces. These two solution models University of Bologna.
have been demonstrated in examples to see how they Gessa, N., Novelli, C., Busuoli, M. and F. Vitali, 2004.
could work. Use and extension of ebXML business profiles for
The TexVDF approach has been provided as a Textile/Clothing firms. LNCS 3182. 186-195.
Guo, J., 2007. A Term in Search of the Infrastructure of
conceptual foundation for future design of globally
Electronic Markets. IFIP Volume 255, Springer, Bos-
interoperable textile e-Marketplaces. It has advan- ton. 831-840.
tages compared existing solutions. (1) It does not Guo, J., 2008. Collaborative Concept Exchange. VDM
enforce standards on business vocabulary and Publishing, Germany.
document templates. This implies a flexible solution Guo, J., 2006. Inter-Enterprise Business Document Ex-
to semantic consistency maintenance between par- change. In Proceedings of ICEC’06. ACM Press. 427-
ticipated textile e-Marketplaces. (2) The new 437.
framework it provides allows cross-domain level Lovejoy, J.(a) Pipeline Analysis (PA):How Does your
semantic interoperability but still enable personaliza- Business Flow? [online] http://www.techexchange
.com/thelibrary/Dama/pa.html
tion. (3) Useful terminology standards are welcome
Lovejoy, J. (b) Textile Industry Data-Sharing Network: A
to be flexibly integrated into the new framework. It prototype for building bridges, using the Internet.
implies a standard integration but not rigid. (4) Vo- [online]http://www.techexchange.com/thelibrary/Dam
cabulary design between e-Marketplace mediators is a/texnet.html.
collaborative in real-time. This enables semantic MODA-ML. http://www.moda-ml.org
accuracy and avoids erroneous inference between Leech, P., 1999. Who Says Manchester Says Cotton.
individual e-Marketplaces for those taking the ap- inTRAlinea 2 [online] www.intralinea.it.
proach of independent vocabulary design. (5) Ostic, J., 1997. An Introduction to USITC Enterprise
Document template design and specificity adopt a Analysis. [online] http://www.techexchange.com/
thelibrary/Dama/An_Introduction.html.
simply hierarchical document structure where each
Teng, G. and J. Hector, 2006. Integrating the US textile
document element concept referenced to a well- and apparel supply chain with small companies in
defined collaborative concept vocabularies in CMM. South America. Supply Chain Management, 11(1) 44-55.
This enables simple creation and use of document Vickrey, D., Biewald, L., Teyssier, M. and D. Koller,
templates. 2005. Word-Sense Disambiguation for Machine
Currently, the improved structure specification of Translation. In HLT/EMNLP’05, Human Language
TexCVF framework in terms of XPM is in final Technology Conference and Conference on Empirical
release stage. The future work of this paper will be Methods in Natural Language Processing). The Asso-
the implementation of the textile e-Marketplace ciation for Computational Linguistics. 771–778.
based on this conceptual foundation.

88
 PROPOSAL OF AN ARCHITECTURE FOR DIGITAL CITIES
CREATION
Proposal of an Architecture P2P for Digital Cities Creation

André M. Panhan, Denys G. Santos and Leonardo S. Mendes

Communication Networks Laboratory, Communications Department, Eletrical and Computer Engineering Faculty
University of Campinas – UNICAMP - Cidade Universitária "Zeferino Vaz"- Barão Geraldo, Campinas, SP - Brazil
[email protected], [email protected], [email protected]

Keywords: Digital Cities, Middleware, Open Access Metropolitan Networks.

Abstract: The term digital cities do not have a precise definition. The number of digital cities, whose development
started from the 90s, is constantly growing. Motivated by the increase of these cities, this paper describes
architecture for digital cities environment as a proposal to create a middleware for the interconnection of the
various segments of society. This middleware is structured in a Platform for P2P networks, also known as
DHT, with emphasis in the JXTA modified protocol.

1 INTRODUCTION Digital Cities, as defined by Graham (Graham,

2007), can serve as a tool to improve communication
Also known as Virtual City, Digital City, Intelligent between citizens and local government, stimulating
City and other names, digital cities represent a kind many activities that promote new opportunities to
of projection of a city in the digital world and the citizens, as we identify below.
emerge as one of the forces that may contribute to "... virtual cities are electronic spaces, generally
space organization (Silvia, 2002). based on the World Wide Web, which were
The digital cities can have many definitions. In developed to connect, in an explicit way, the
(Silvia, 2002), it is defined that digital cities are development schedules of each city. These virtual
people, systems and institutions that remain cities are working as political tool for a variety of
connected through a digital communication urban plans and objectives: global urban marketing,
infrastructure. In digital cities, one uses real cities as stimulus to business and consumption, improving
reference for their digital counterpart, aiming at communications between citizens and local
developing social relations of different purposes. governments, increase the local enterprise
These include one or more of the items cited below. competitiveness, greater local economies integration
- Creation of political and cultural spaces for and the rebirth of civic and local culture." (Graham,
people and groups manifestation; 2007)

- Creation of communication channels between Motivated by the increased number of digital

individuals and groups; cities developed during the 90s, this article presents
an architecture for digital cities environments and a
- Creation of communication and negotiation proposal to create a middleware based on a DHT
channels between municipal government and structure to people, groups and services (public and
citizens; private) interconnections for this proposed
- Favoring a greater identification from the environment.
visitors with the reference city;
- Creation of a collection of information from
many different species and easily accessible
on the reference city.

89
ICE-B 2008 - International Conference on e-Business

2 DIGITAL CITIES 2.3 Access Media

ARCHITECTURE In the segment Access Media one explains issues
relating to devices standards, which are used to
In (Yamaoka, 2006) it is defined a minimal set of access the services offered by Digital Cities. We also
assumptions, policies and technical specifications define how devices must be treated. Some of these
that rules the use of Information and Communication
devices are computers, mobile phones, hand-helds
Technologies (ICT) by the government, establishing and digital television, for example.
the conditions for interaction. This middleware
emphasizes the JXTA modified protocol, extending
the interoperability traditional models in government
2.4 Organization and Information
for other society segments as citizens, companies Exchange
and the third sector. Our proposition is developed
based on these minimal premises. This segment deals with aspects related to the
processing and transfer of information among
services offered by Digital Cities. It includes
components for data exchange, data processing, data
definition and Standards Database Catalogue (SDC).
Information distribution for these situations requires
the construction of suitable metadata structures for
each of the cases above.
Clearly defined policies and specifications for
Figure 1: Interaction with digital cities and segments interoperability and information management are
society.
key issues for providing the enterprise connection
among citizen, enterprise, government and service
The proposed architecture establishes rules on providers. The Digital City is designed as a basic
the following areas responsible for the structure for the government strategy and business in
interoperability of various society segments, the several business scenarios (B2B, B2C, B2G,
Interconnection, Security, Access Media and G2G, G2C and C2C). This approach allows the
organization and Information exchange. rationalization of investments in ICT, through the
For each of these areas we specify components, sharing, reuse and exchange of technology
for which standards are established. resources.
Enterprises, government and services providers
2.1 Interconnection information resources are valuable economic assets.
By ensuring that information or service can be
The segment "Interconnection" establishes the rules quickly located and exchanged between the public
for enterprises (such as service or commercial and private sectors, keeping the privacy obligations
companies) and governments to interconnect and security, one encourages and stimulates local
themselves upon the local network, which can be a economy.
metropolitan network, a community network, or Nowadays, most of the countries of world
even the Internet. This segment also treats the recognize the needs for high investments in
interoperation rules of enterprises and governments developing universal access to multiservice
with the citizen. In this segment are defined networks for all citizens. Economic analyses show
Electronics Message Transport, Electronics Message that universal services cause a direct and almost
Safely Content, Intercommunication LAN/ WAN immediate increase in GNP. Its also understood that
and Transportation. universal access to services, particularly when we
talk about e-Gov services, can only be achieved
2.2 Security through the creation of virtual (and, sometimes, real)
local networks (or Digital Cities). In creating the
This segment deals with the proposed architecture Digital Cities one is primarily concerned with
ICT security aspects that should be considered for establishing policies development, procedures and
Digital Cities creation. We treat in this segment standards for ICT, mounting dedicated structures to
security IP patterns, encryption, System reach local interoperability and seeking better
Development and Network Services. quality provision services at reduced costs. The
existence of a middleware as a foundation for

90
PROPOSAL OF AN ARCHITECTURE FOR DIGITAL CITIES CREATION - Proposal of an Architecture P2P for Digital
Cities Creation

electronic services creation is a pre-requisite for another mechanism must be activated to return the
better services provision at a reduced cost for all the identifier of a peer in time n.
actors of the Digital City. Figure 3 represent the proposed architecture for
middleware model described in this paper. In this
architecture the Rx peers provide routing, the Ex
peers services and the Px peers access to the
interaction portal of Digital City.

Figure 2: Proposal of Architecture for Digital Cities. Figure 3: Proposal of Architecture for Middleware.

A modern knowledge society requires the Another relevant element to build a Digital City
development of integrated and interoperable is the creation of a protocol of multiple connections
intelligent systems, which must work in a safe and to the peers connected to the network. In order to use
consistent environment. These requirements cannot the multi-connection resources already available in
be attained unless a common interoperable the JXTA architecture and seeking to make an
standardized environment is established. The efficient use of the network and maintain its
middleware, whose architecture is shown in Figure workload low, it is proposed the creation of an
3, is also the solution for this exigency. This abstraction layer that keeps the peer connected to
architecture covers the information exchange only one Digital City. However, the protocol allows
between the services offered by all society segments. the peer to make connections to other peers and
other Digital Cities without loosing the main
connection. This resource allows that the peer uses
3 MIDDLEWARE services from other digital cities and, at the same
time, allows other digital cities to access its services.
To create the Middleware, we propose to add an To provide this resource, the abstraction layer
abstract slice on JXTA architecture. This slice will cannot treat the peer identifier like it is treated in the
be acting as an access control to digital cities JXTA architecture. In the proposed abstraction
network. layer, the peer identification must be done using a
pair of identifiers (GPID, PRID), where GPID is the
group identifier and PRID is the peer identifier.
3.1 Structure
The basic structure to the proposed middleware is 3.2 Security
based on the JXTA overlay network topology as can
In a corporative or government communication
be seen in Théodoloz (2004), or in Traversat (2003).
This model creates a network of Randezvous servers environment, data security is an extremely important
issue both in storage and transport of information.
to establish interconnection links among these, and
forming a complete graph, where each server is a To provide security in the operations of data
vertex and the connections are the edges. applications and transactions, one can use resources
offered by the JXTA API. This resource has a set of
In a JXTA network each server has a complete
and updated list of connected peers. Therefore, it is algorithms that provide a base to security control in
possible that the search for a service in the network the applications developed using this API.
be performed in a linear order when the network
servers are consistent. Otherwise, when an
inconsistency is found in the server’s network,

91
ICE-B 2008 - International Conference on e-Business

3.3 Digital City Directory The DHTs proposed in this work has as one of its
main characteristics the separation between
To maintain a consistent identifier services list in the identifier and locator. This feature provides superior
digital cities network, one can create agents mobility and improves security. Another important
associated to the digital city portals and systems feature of this DHT is the service doesn’t have
interfaces. These agents can perform periodic mapped in a physical address. The service can be
searches in the network to look for active services mapped in a simple name, allowing an improved
and to map these services with an appropriate transparency with respect to migration.
description. With this map, the creation, ordering To provide interoperability among different
and availability of data are responsibility of the digital cities without creating inconsistency or
respective digital city through their portals and service address replication, the creation of the
access interfaces. abstraction layer proposed in this paper is
indispensable.

4 USE CASE
ACKNOWLEDGEMENTS
The concept of Digital City can be easily explained
how a user makes use of the digital city environment The work presented here has been developed under
to plan and contract a travel to a congress in another the umbrella of the projects “Municipal Infovia – An
city. Open Access Network for Cities” and “SIGM – An
The citizen enters in the portal a make Integrated e-Gov Environment for Cities”. These
authentication. Authentication done, the citizen projects have been supported in part by the
makes them registration and makes a payment with a governments of the cities of São José do Rio Preto,
credit card. As the bank system also was developed Pedreira, Penápolis, and Campinas, São Paulo State,
within digital city architecture, the congress page Brazil.
makes a request to the bank system with totally
transparent to the user.
After registration, the citizen chooses a hotel in REFERENCES
the city of the congress, buy the airplanes tickets and
rent a car, paying all of this in the same bank Silva, Michéle T. Candido., 2002 A (Ciber) Geografia das
systems Cidades Digitais. In Universidade Federal Fluminense
Although these actions performed by the citizen - Brasil.
are familiar to the common user of the Internet, the Zancheti, Sílvio Mendes., 2001. Cidades Digitais bases
Internet users are not able to follow, and even para um Programa de ação. In Sociedade da
interfere in the process. In the Digital City, the user Informação - Brazil.
is able to perform all these activities by accessing GRAHAM, Stephen., 2007, Rumo a Cidade em tempo real
- Desenvolvimento Urbano numa Sociedade
only one page and yet follows all the process that
Globalizada e Telemediática. In
happening behind the scenes. http://www.wisetel.com.br/espaco_de_futuros/vcidade.
htm - Brazil.
Pressman R., 1995. Engenharia de Software. Makron
5 CONCLUSIONS Books, São Paulo – Brazil.
Théodoloz N., 2004. DHT-Bases Routing and Discovery
in JXTA. In School of computer and Comunication
The creation of digital cities provides Sciences – Computer Sciences departement – École
communication channels to facilitate and improve Polytechnique Fédérale de Lausanne - Switzerland.
the relations between citizens and government, Traversat B. at all, 2003. Project JXTA 2.0: Super-Peer
citizen and enterprises, and enterprises and Virtual Network, In Technical Report Sun
government. This communication environment is Microsystems - USA.
suitable to head these relationships because the Yamaoka J. at all. 2006. Padrões de Interoperabilidade de
sophisticated identification, authentication and Governo Eletrônico. In Documento de Referencia,
security offered by the digital city middleware. For Governo Federal - Brazil.
the authenticated user, the digital city can provide an
environment rich in information and services.

92
OPEN SOURCE SOFTWARE AND LEVERAGING OF BUSINESS
EFFECTIVENESS IN SMES
A Case Study

Steven Butler, Dotun Adebanjo

Management School, University of Liverpool, Chatham Street, Liverpool, U.K.
[email protected], [email protected]

Hossam Ismail
Management School, University of Liverpool, Chatham Street, Liverpool, U.K.
[email protected]

Keywords: e-Business, Open Source Software, Small-to-Medium Enterprise (SME).

Abstract: This research investigated the issues that impact on an SME adopting e-Business. It was found that many
barriers may exist for e-Business adoption in an SME. One of which is limited financial resources to
purchase internet technology required for e-Business adoption and development. However, open source
technology has grown in popularity for a number of years, with governments, business firms, military and
educational organisations incorporating open source software. This paper provides an overview on open
source technology which could be adopted for e-Business architecture for the development of e-Business
applications. Furthermore, it provides two action research case studies on SMEs adopting open source
software for the development of e-Business capabilities. It was found the projects were successfully
developed and implemented for each case study company. Although successful, the research team found
that more research is required in open source software for the development of e-Business applications.

1 INTRODUCTION provide an organisation with the opportunity to

extend business functionality by providing users,
In the United Kingdom (UK), Small-to-Medium such as employees, suppliers and customers to
Enterprise, (SMEs) provides a significant access business logic through a browser, or
contribution to the economy. Some authors report commonly referred to as ‘business though a
the contributions to Gross Domestic Product (GDP) browser’.
up to 40 percent (Taylor and Murphy, 2004). The E-Business applications and architecture uses a
world economy is changing, with growth in three-tier client server model, as illustrated below in
emerging markets and developing economies figure 1. A three-tier client server model is where
continuing to outstrip that in developed economies, the client is mainly used for display with application
(Jutla et al, 2002). This has resulted in more pressure logic and the business rules partitioned on the server,
on SMEs to discover ways in which to deliver its which is the second tier, and a database server is the
products and services more effectively, and improve third tier. (Chaffey, 2004).
operations activities to remain competitive.
The growth in the internet and e-Business
applications has provided organisations with new
opportunities to offer products and services, share
information with business partners and meet
customer demands. Therefore, e-Business Figure 1: Three-tier architecture.
applications and the internet have the potential to
increase the competitiveness and growth of small Although e-Business can provide so many
firms, (Paliwoda, 2004). E-Business applications opportunities, many barriers may exist, preventing

93
ICE-B 2008 - International Conference on e-Business

an organisation adopting e-Business. The effective Access and MySQL to consider suitability for e-
adoption and implementation of ICT may rely quite Business applications for the case study.
a lot on individual factors such as organisational Microsoft Access is a DBMS developed by
size, structure, mix of available human and financial Microsoft Corporation and belongs to the Microsoft
resources and capabilities (Pavic et al, 2007). Office family of products. Microsoft Access
This is further supported by resource issues includes database functions that allow users to
relating to finance, skills and personnel are expected perform database queries and has some very good
to impact on the decision to adopt e-Business, (Fillis reporting interface. Microsoft Access is a flexible
et al, 2004). Research has found that there are a package and is suitable for organisations that have
number of barriers and success factors associated small database requirements and require few
with e-Business adoption for an SME. (Taylor & database records. In a database environment with
Murphy, 2004) identified high initial set-up costs of multiple users, connections and queries, Microsoft
ICT and e-Business can act as a barrier for an SME. Access will probably become unreliable and slow.
Clearly, SMEs are presented with a number of However, Microsoft Access can be less expensive
challenges to realise the benefits e-Business that can than some other DBMS on the market.
provide to an organisation. However, open source Microsoft SQL Server is another Microsoft
technology may provide a solution by providing package, but this product is primarily marketed
access to technology which may not be cost towards large organisations with large database
prohibitive. requirements. SQL server is an enterprise-level
database system and is widely used in organisations
1.1 Open Source that require a database that can handle high use,
multiple transactions and offers high reliability.
Open source is a development method for software Microsoft SQL server is certainly a viable option for
that harnesses the power of distributed peer review large organisations and the cost highlights this,
and transparency of process. The promise of open while SQL server is an expensive option and
source is better quality, higher reliability, more requires a lot of training to be able to use the product
flexibility, lower cost, and an end to predatory effectively.
vendor lock-in (Open Source Imitative OSI, 2008). MySQL is an open source project which has
Open source software has grown in popularity in been publicly available since 1996, but has a
recent years, with governments, business firms, development history going back to 1979, (Welling &
military organisations, and educational entities Thomson, 2003). MySQL is a relational database
incorporating open source software into their management system (RDBMS), which enables data
enterprise functions to counter tightened budgets and storage, searching, sorting and retrieval. MySQL has
rising operational expenses (Hedgebeth, 2007). Any many attractive features which include speed, ease
reservations that exist regarding security may be of use, query language support, capability,
short-sighted, with financial services institutions connectivity, security and portability (DuBois,
such as Experian, embracing open source software 2003). MySQL has become a popular database
recently, (Ferguson, 2008). It appears open source solution as its popularity within the database
popularity will continue to grow, with predications community has increased and would probably suit
that by 2012, more than 90 percent of enterprises most database needs.
will use open source in direct or embedded form, It should be noted that there are several database
(Judge, 2008). options available and the views expressed by
Open source appears to be here to stay, but what authors, practitioners and application developers on
open source software is available which can be used available database systems varies significantly.
to develop e-Business applications, based on an e- Many developers and practitioners may have a bias
Business three tiered architecture (Figure 1). towards database systems with which they may have
had previous experience. It is clear that the choice of
database system will depend on a number of factors
2 TECHNOLOGY REVIEW such as price, speed, efficiency, etc. The database
comparison table illustrated below in Table 1
highlights the differences in the most suitable
The research team identified various Database
database systems available.
Management Systems (DBMS) which could be used
The research team analysed suitable scripting
in the data tier of the three-tier architecture. The
languages for the business logic layer of the three-
research team analysed SQL Server, Microsoft

94
 OPEN SOURCE SOFTWARE AND LEVERAGING OF BUSINESS EFFECTIVENESS IN SMES - A Case Study

tier architecture. The research team analysed ASP, Furthermore, it is well supported by several PHP
ASP.NET and PHP to consider suitability for e- discussion groups and communities. PHP has grown
Business applications. from a set of tools for a personal home page
development to the world’s most popular web
Table 1: Database Comparison. programming language, and it now powers many of
Access SQL MySQL
the Web’s most frequent destinations (Shiflett,
Server 2006).
Cost Inexpensive Expensive Free The choice of ASP, ASP.NET or PHP will
Speed Weak Strong Strong ultimately come down to the application
Efficiency Weak Strong Strong requirements and the hosting provider environment.
Users Few Many Many The choice will also include the developers’
Connections Few Many Many familiarity with programming language. If the
developer is familiar with C, Java, JavaScript, etc,
ASP is a server-side scripting language then PHP may be the best choice. If however, the
developed by Microsoft to run dynamically developer has little experience of programming
generated HTML pages. ASP was originally languages the choice may be the .NET concept of
introduced with Microsoft’s server platform and code behind the page. ASP.NET or PHP would be
utilises VBScript, a variation a Visual Basic more suitable for e-Business applications, as
Language, consequently both Microsoft products. software support and updates for ASP becomes
Although scripting languages have been changing redundant in the future. The technology comparison
rapidly, many competing technologies have table illustrated below in Table 2 highlights the
introduced new integrated features, while ASP has differences in the most suitable technologies
been lacking in this area. However, many business available.
environments have Microsoft servers. This is
probably a major driver for the continued use of Table 2: Scripting Language (Source: Oracle).
ASP and ASP supports Microsoft SQL server, which
is also widely used. ASP.NET PHP
ASP.NET is part of Microsoft's .NET platform Cost Free Free
and should not be confused with ASP. Although Platform Cost Expensive Free
both are developed by Microsoft, the only other link Speed Strong Weak
Efficiency Strong Weak
between both technologies is that they both use
Security Strong Strong
VBScript. ASP.NET can be used to build dynamic
Platform Strong Weak (IIS Only)
web sites, web applications and XML web services.
ASP.NET offers developers the ability to build A number of technology options exist for e-
pages using controls such as buttons or labels, with Business applications and only a handful have been
properties assigned to the controls. The controls analysed due to suitability and requirements for the
produce segments of code, e.g. HTML for forms. e-Business applications to be developed in this
This concept introduces the idea of all the code paper. This paper investigates the use of open source
behind the page. This may reduce development time, software for e-Business applications as access to
but it is fair to conclude that programmers and/or financial resources can be limited in an SME.
developers will always need a deep understanding of The research team found that MySQL would be
the technology, even with the .NET concept of code the most suitable database option for each case study
behind the page. organisation due to cost, stability and security.
PHP is an open source scripting language and MySQL is a true open source database management
was originally started as a hobby, this programming package. As a result, the research team decided that
language has grown in popularity and use in recent PHP would be the most suitable scripting language,
years. It can be imbedded into HTML pages and as it works seamlessly with MySQL and, once more,
allows communication with a database. A number of it is an open source product.
databases support PHP, which include MySQL, Therefore, the research team agreed to develop
dBASE, Microsoft SQL Server, Oracle and many the e-Business applications using MySQL and PHP
more. PHP supports most back-end databases, but it for the data layer and business logic layer,
is most commonly known for its seamless respectively. MySQL or PHP do not require a
integrations with MySQL. It can also run on a commercial licence and all are battle-proven on high
number of operating systems such as Windows, volume websites. Apache was selected to provide
Linux, MAC OS and most varieties of Unix. the web server environment. The apache web server

95
ICE-B 2008 - International Conference on e-Business

open source project was developed and as an open Organisation B case study was a medium-size
source HTTP server for operating systems such as car finance broker based in the UK. This
Unix and Windows. Apache has been the most organisation has been established for some time,
popular web server on the internet since 1996 with a growing finance portfolio. The primary role
(Apache, 2008). Apache is also the most popular of this organisation was to link car dealerships and
web server and has the largest market share, used on finance organisations able to provide funding for
over 50 percent of web servers, (Netcraft, 2008). vehicle leasing and purchasing, acting as an
The software selected for the development of e- intermediary. The organisation used a proprietary off
Business applications are Apache, MySQL, and PHP the shelf software package to support its operations.
are open source software and are known collectively The software did allow some customisation, but it
as AMP. was found this had its limitations.
Through management briefings, conducting
interviews and process reviews at each visit, it was
3 CASE STUDY possible to identify issues and problems with
business processes. The research team identified
Organisation A was a micro-size manufacturing issues with current business processes, data
organisation based in the UK. The research team integrity, processing time and legacy system. The
also found that the case study company was organisation rely on external parties, (car
experiencing sustainable growth, with a year-on- dealerships), to provide legible and complete
year sales change of more than forty percent. The information for finance proposals. It was found that
case study found that the organisation was suffering proposals are hand written by car dealerships and
from ‘growing pains’ in a number of areas of the then faxed to the contact centre for processing. Once
business. The research team carried out interviews, received, employees were required to transfer the
process review meeting and structured walk-through information on each proposal onto the legacy
to understand business operations and indentify system. However, if an application has not been
improvement opportunities. It was found that the hand written in a legible format employees were
organisation used office applications and had a expected to decipher information on the proposal,
legacy system for stock management, but many of presenting the organisation with data integrity
its key processes were paper-based and manual. It issues.
was found that the organisation was becoming Furthermore, it was found that because proposals
inefficient in a number of its key processes. The were faxed this affected the quality of information
research team and management reviewed all on a proposal. The organisation team were fully
improvement opportunities to identify ‘urgent aware of this problem and were keen to resolve this
improvements’. A key process in the dispatch issue through the use of technology. The research
department was causing a bottle neck and affecting team identified this as an area of improvement, as
other areas of the business. This was identified for improvements, or elimination of data integrity
improvement actions. The process was know as the problems would provide the organisation with clear
‘verification process’, was critical to the business, and accurate information and more efficient business
and directly affects the organisations ability to processes.
service customers. The organisation offers products which require a
This process was a manual process. It required decision of acceptance or decline within two hours.
personnel in the dispatch department to validate the It was found that sometimes this is exceeded due to
data on six documents against a master data sheet to incomplete or incorrect data collection, which
ensure all the data on the documentation matches necessitated follow up actions and telephone calls by
that on the master data sheet. Once complete, a employees resulting in extended processing time. It
manager would then be required to repeat this was found that if the processing time extended
process to verify all the documentation was correct, beyond two hours then a dealership would source
and data was not overlooked accidently. This finance from a competitor, resulting in lost
activity was required to be completed on every commission for the organisation.
product. It was a critical step, as the data and
information contained on the documentation
comprised of product information, safety data and
certified values.

96
 OPEN SOURCE SOFTWARE AND LEVERAGING OF BUSINESS EFFECTIVENESS IN SMES - A Case Study

4 RESEARCH AIMS & The research team carried out several face-to-
OBJECTIVES face interviews with stakeholders from case study A
and B. The interviews with case study A were
primarily carried out with key stakeholders, but this
The overall aim of this study was to determine ways
was then extended to include employees involved in
in which open source software may be used to
business processes to better understand each process.
develop e-Business applications to improve business
Once each process was fully understood, the
processes in each case study organisation. The
research team then began identifying areas for
research focuses on identifying suitable open source
improvement through the development of a web-
software to develop such applications. The research
based system.
then moved onto proposing solutions utilising
In respect of case study B, initial interviews were
internet technology, and then developing e-Business
held with senior management (directors), and the
applications. To achieve this aim, a number of key
Information Technology (IT) manager. The research
objectives would have to be met. These key
team began with understanding existing business
objectives include:
processes. This process enabled the research team to
• Identity open source software for e- understand finance proposal process from initial
Business applications. enquiry to release of funds.
• Determine ways in which solutions can be In case study A and B the research team was
developed using internet technology. involved in the system analysis and development
• Develop e-Business applications for each process for each web-based solution. In case study
case study organisation. A, the research team designed web-based
• Identify benefits e-Business provided each applications to improve existing process, data
organisation. handling and reduce resource requirements. In case
• Use findings in the paper for the study B the research team designed a new e-
development of an e-Business framework Business application to link car dealerships and the
for an SME adopting and implementing e- case study company to improve operational
Business. activities and enable information sharing. In both
cases the research team developed the e-Business
applications using open source software, which had
been designed for each case study organisation.
5 RESEARCH METHODOLOGY
The primary research for this paper is based on a
case study of two UK based SMEs, the first is in the 6 FINDINGS
manufacturing industry, and the second is involved
in the financial services industry. Two case study Organisation A
organisations were selected for this paper so that It was decided that a web-based verification
research could be completed on a manufacturing application would be developed by the research team
organisation and non-manufacturing organisation. to improve the ‘verification’ process for organisation
The case study is action research lead in both cases. A. The study had already identified that access to
In first case study the research team spent more than financial resources for technology development
18 months working with the case study company on would be a constraint. This resulted in the
an e-Business development programme, to develop stakeholders agreeing that the ‘quick win’ project
e-Business capabilities and implement operations should be developed through the utilisation of
management tools and techniques. An associate barcode and open source technology. This was the
from the research team was based at the case study perfect opportunity for both the research team and
company throughout the programme. the organisation to witness the viability and
The second case study company is a medium- suitability of open source technology, which was a
size enterprise based in the UK. The organisation new concept to the case study company.
investigated the use of internet technology in both The web-based application would be developed
front and back end activities to improve business so that barcode data on a product master data sheet
operations to minimise processing time and data could be scanned and stored on a database. The data
integrity issues to improve competitiveness of the required for validation would be stored in barcodes
organisation. on product documentation when printed. An

97
ICE-B 2008 - International Conference on e-Business

employee could then use the verification application positively witnessed and experienced by the entire
to complete data verification checks using a barcode organisation. Management and personnel involved
scanner. The application would check the barcode in the verification process have commented that
data on the documentation, against the master data without the adoption of the system, it would have
stored in the database, informing the employee of a been impossible to meet current customer demands
pass, or fail. This would result in eliminating using the existing approach.
management from this process, as the employee
Organisation B
would not be required to perform data verification,
thus eliminating human error from the process. The research team and employees from the
The application successfully developed with organisation reviewed several web-based
open source technology (APACHE, PHP and applications used by competitors and organisations
MySQL). It was successfully implemented and providing financial services. This provided the initial
replaced the existing manual process, thus concept of a web-based application to the
improving the process significantly. It was found organisation. The research team held focus group
that the new application provided several benefits, meetings to discuss, and identify desirable
such as more efficient business processes within the functionalities of the proposed web-based
dispatch department because the current bottle neck application. It was agreed that the application would
had been eliminated. This in turn helped speed up also include finance quotation tools for dealerships
the dispatch of goods from the department. to use whilst using the web-based proposal form. It
Furthermore, through better utilisation of human was thought this would enhance value proposition to
resource management, more time can now be spent dealerships, thus encouraging adoption. The web-
in other areas of the business, which could provide based proposal form would include validation
even more cost savings or help improve revenue functions to ensure that all required information was
streams through new product development. completed on the form before submission, thus
The total project cost for the application was ensuring data integrity and minimising errors, thus
calculated at £4006.35. This included hardware, reducing processing time.
software, training and development costs. The new It was agreed to develop and implement the
process provided a resource saving of fifty percent solution in phases. In the first phase the solution
and a noticeable improvement in throughput time. developed would store all the proposal information
This was calculated as a total cost saving of in a database. The proposal would then be encrypted
approximately £20K in the first year, providing an and emailed to the organisation for processing, this
approximate five hundred percent Return on would replace the fax. The next stage of
Investment (ROI) for the case study company. development would look at identifying solutions to
It was found that the verification system allow data sharing between different operating
provided a number of benefits which could not be systems and programming platforms.
fully quantified. Some of which included the ability The organisation successfully implemented a
of the case study to eliminate the bottle neck web-based e-Business application using open source
observed in the department. This in turn helped software (PHP & MySQL). The application
increase the throughput of goods in the dispatch improved information flows between the car
department. Furthermore, through better utilisation dealerships and the organisation, replacing the paper
of human resource management, time can be spent passed proposal application process. The application
in other areas of the business, which could provide was first released to field sales staff for review and
even more cost savings or help improve revenue testing, before being released to car dealerships.
streams through new product development. The However, it was identified that the next stage of the
process was de-skilled, resulting in any employee development may be more difficult, as the
being able to perform verification using the web- application would be required to share data directly
based system, after initial training. with legacy systems, posing a security risk.
In addition, the development of the web-based Furthermore, a dealership has requested information
application includes the storage of data, using the sharing between both organisations to simplify the
MySQL database. As a result, management can use proposal process. If the organisation can provide a
the data to monitor the process and retrieve solution, the dealership as agreed to complete all
historical data to help make business decisions. business transaction with the case study company.
Finally, the research team found that knowledge and This has resulted in the proposal for the application
understanding of e-Business application had been to be developed further, into a web service.

98
 OPEN SOURCE SOFTWARE AND LEVERAGING OF BUSINESS EFFECTIVENESS IN SMES - A Case Study

7 DISCUSSION each organisation were observed by the research

team, and these need to be considered by academics
Although both case study organisations had and practitioners involved in e-Business
implemented e-Business applications successfully development projects for SMEs. Although many
using open source software, the research team found barriers may exist, each organisation experienced e-
that a number of barriers existed, which included Business benefits, as illustrated below in Table 3 and
access to skills, key personal and business culture Table 4.
issues.
It was found that Case Study A had little Table 3: Case Study A - e-Business Benefits.
knowledge of e-Business, e-Business applications or Cost Saving Process Effects Culture Effects
the benefits and opportunities that could be realised £20K (year 1) Bottle neck e-Business
by an organisation adopting e-Business. The removed knowledge &
research team managed to involve as many understanding
employees as possible in the development of the e- Data integrity e-Business buy-in
Business application, to help deepen knowledge and Increased Positive
understanding and encourage buy-in at all levels of throughput experience of
the organisation. It was also found that the open source
software
organisation lacked the necessary skills and
De-skilled Open-source
knowledge to develop e-Business applications.
process buy-in
Moreover, the research team found that access to 50% human
personnel required to implement and deploy the resource saving
application was limited, due to personnel involved in
other business activities. Once the e-Business Table 4: Case Study B - e-Business Benefits.
application was successfully implemented, the
research team arranged a feedback meeting to Cost Saving Process Effects Culture Effects
ascertain personnel experience of e-Business. It was N/A Information e-Business buy-in
sharing
found that the approach to involve personnel had
Legible data Positive
been received positively, as personnel felt that e- experience of
Business knowledge and understanding was now open source
evident throughout the organisation. software
It was found that Case Study B Company did Reduced Open-source
have knowledge of e-Business applications, but did processing time buy-in
not fully understand how internet technology could
be deployed to improve operational activities. The
research team found that the organisation did have 8 CONCLUSIONS & FURTHER
an information technology department, which were
highly skilled in a number of areas, but did not WORK
possess the necessary skills to develop e-Business
applications. Moreover, it was found that access to The case study found that SMEs face a number of
personnel was limited, similar to the first case study challenges, and that e-Business can provide
organisation. Clearly, key stakeholders must opportunities for organisations to improve business
appreciate that access to personnel is required, operations, information sharing and remain
which may remove personnel from daily business competitive. However, they may face several
activities. The project feedback meeting found that barriers to e-Business adoption, one of which is
all involved had a positive experience and found access to technology. It is found that access to new
open source could provide significant benefits to the technology can be cost prohibitive, or existing
organisation. This organisation is now looking to legacy system costs may prevent further investment
develop the e-Business application into a web- in technology. This may be overcome by the use of
service for its dealership network. This will also be open source technology.
developed using open source software. The research team found that open source
This paper had already indicated that access to software can be used for the development of e-
software can be cost prohibitive, but this was Business applications, namely MySQL and PHP. E-
overcome through the utilisation of open source Business applications were developed and
software. However, a number of other barriers for implemented for two case study organisations, to

99
ICE-B 2008 - International Conference on e-Business

support business processes and enable information http://www.oracle.com/technology/pub/articles/hull_a

sharing. Supporting the view of (Bocij, 2003) e- sp.html [Accessed April 2006]
Business is aimed at using information and Paliwoda, S., (2004). E-Business, Qualitative market
communications technology to integrate and research – An international journal, Vol.7, No.1.
Pavic, S, Koh, S.C.L, Simpson, M. & Padmore, J., (2007)
enhance processes. In a value-chain context, the Could e-Business create competitive advantage in UK
efficiency of processes is enhanced, which should SMEs?, Benchmarking: An International Journal,
result in lower-cost, higher quality products so Vol.14, No3, pp. 320-351.
giving the customer better value. Shiflett, C., (2006). PHP Security, O’Reilly Media Inc.
Although open source technology may provide United States of America.
an SME with an opportunity to develop low cost e- Taylor, M & Murphy, A., (2004). SMEs and the take up of
Business applications, access to personnel with the e-Business, Urban Geography, Vol.25 No.5, pp. 315-
skills, experience and knowledge of such technology 331.
Welling, L & Thomson, L., (2003). PHP and MySQL Web
is still required. Therefore, more research into access
Development, Sams Publishing. United States of
to open source skills is required to ensure that open America, Second Edition.
source technology development can be supported in
the future.

REFERENCES
Apache, (2008). Apache: HTTP Server Project,
http://httpd.apache.org/. Date accessed 14/04/2008.
Butler, S., (2006). Web-based MRO System for
Commercial Aircraft Services, University of
Liverpool. Final Year Project
Chaffey, D., (2004). E-Business and E-commerce
Management Second Edition, FT Prentice Hall.
Harlow.
Dubois, P., (2003). MySQL, Sams Publishing. United
States of America, Second Edition.
Ferguson, T., (2008). Experian embraces open source,
http://www.silicon.com/financialservices/0,380001032
2,39186071,00.htm?r=1. Date accessed 14/04/2008.
Fillis, I, Johansson, U & Wagner, B., (2004). Factors
impacting on e-Business adoption and development in
the small firm, Journal of Entrepreneurial Behaviour &
Research, Vol.10, No.3, pp. 178-191.
Hedgebeth, D., (2007). Gaining competitive advantage in
a knowledge-based economy through the utilisation of
open source software, The journal of information and
knowledge management systems, Vol.37, No.3, pp.
284-294.
Judge, P., (2008). Gartner: Open source will quietly take
over, http://news.zdnet.co.uk/software/
0,1000000121,39379900,00.htm. Date accessed
14/04/2008.
Jutla D, Bodorik, P & Dhaliwal, J., 2002. Supporting the
e-Business readiness of small and medium-size
enterprises: approaches and metrics, Internet
research: electronic Networking Applications &
Policy, Vol.12, No.2, pp.139-164.
Netcraft, (2008). April 2008 Web Server Survey,
http://news.netcraft.com/. Date accessed 14/04/2008.
Open Source Initiative OSI, (2007). Home,
http://www.opensource.org/. Date accessed
14/04/2008.
Oracle Technology Network. PHP vs. ASP.NET Go head-
to-head. [Internet] Available from:

100
AN ONTOLOGY-BASED ARCHITECTURE FOR MULTI-AGENT
SYSTEM ENVIRONMENT

Roberto Paiano, Anna Lisa Guido and Enrico Pulimeno

Dipartimento Ingegneria dell’Innovazione, SalentoUniversity, Via per Arnesano, 73100 Lecce, Italy
Tel: +39 0832 297229, Fax: +39 0832 297279
{roberto.paiano, annalisa.guido, enrico.pulimeno}@unile.it

Keywords: Multi agent systems, ontology reasoning, domain independent, domain dependent.

Abstract: The increasing interest towards e-business systems, and thus the need of the companies to communicate
with other companies in an efficient and flexible way, brings to a new way of thinking about information
systems that open itself towards distributed systems where the exchange of information may happen. The
problem is in the way that this exchange can be made both from the technology point of view and from the
conceptual point of view. If from a side a recent technology for information exchange (DDS- Data
Distribution Service) can help us from the other one it seem interesting the use of ontology as representation
tool of application domain that must open itself to the communication. In this paper we present an
architecture that is oriented to a multi-agent communication that uses ontologies and DDS as information
exchange protocol. The architecture here presented must be repeat without difficulties on an e-business
system assuming that each agent of the proposed architecture is a company in an e-business system.

1 INTRODUCTION data requested from the business processes in order

to go on with the execution are not immediately
An e-business system is a system where involved available. It is necessary a mechanism that allows to
companies communicate among them exchanging ask for the data and wait until the data is made
data in as much as possible efficient and flexible available by some other information system that can
way with the goal to satisfy the final user. Two are provide it.
the main problems that rise in this area: A base architecture useful to think about an e-
business system based on a multi-agent has been
ƒ Conceptual problem related to the definition of
presented in a research project lead by SSI (Space
a semantics of interchange of the data in order
Software Italia) and the Department of Engineering
to assure the communication of the
Innovation of the University of Salento. The project
companies in heterogeneous environments;
was founded by Apulia region. Starting from a
ƒ Technological problem related to the use of a
multi-agent system already made up from SSI in a
suitable data interchange system useful to
demining system where each agent is represented by
assure the communication between
a robot, the goal of the project has been to enable
heterogeneous systems.
the communication between agents through a level
Both the problems are difficult to be faced and of intelligence made up by semantic web
surely can be of great help the formal ontologies to technologies, in order to define a run-time all the
solve the first problem (conceptual problem) and, parameters to publish/subscribe depending on the
relatively to the technological problem, it can come operational context where agents work.
in help the DDS technology (Data Distribution Naturally, when in the multi-agent system the
Service) (DDS, 2001) standard OMG and surely parameters to publish/subscribe has been identify , it
useful not only for synchronous comminication is necessary to avoid that they are manually
among companies but also for the asyncronous one. published/subscribed in the DDS; it would be,
Very often, in fact, the companies don’t need therefore, useful that the classes (the parameters are
only information in real time: we can think, for published/subscribed in the DDS through classes)
example, to a business process that needs to have a that materially define these parameters inside the
data coming from another information system. The DDS, are automatically produced.

101
ICE-B 2008 - International Conference on e-Business

The "intelligent" level, that it is important to add, ƒ interactions among agents: information
it has to operate so that to allow "to understand", interchange among agents in distributed
depending on the context, the parameters to environment;
publish/subscribe. To add this level, the formal ƒ Evolution toward component more and more
ontologies are particularly useful. Not only the endowed with autonomy: agents that in full
ontology but also the technologies of reasoning that autonomy reach their own goals.
allow from a side to define the context where the The multi-agent systems need a study of the
system operates and from the other to individualize nature of the interactions. It results therefore
the rules of action that each agent can take in order important the notions of collaboration and
to answer to a well defined event. cooperation.
The comparison with this kind of system and the Interesting, in this paper, to underline that for the
e-business systems is immediate: every agent realization of a multi-agent architecture it is
represents every information system involved in the necessary to have a layer of communication and a
information interchange. The "intelligent" level layer of conceptual modelling.
allows subsequently the different information ƒ Communication Layer: currently the most
systems to publish/subscribe in the DDS the data of qualified is standard seems to be FIPA-ACL
interest to an appropriate system of reasoning that (FIPA, 2002) (Agent Communication
allows to identify the correct information. Language) created from the Foundation for
In this paper we present a high-level architecture Intelligent Physical Agents (FIPA). This
designed in the research project: the architecture is standard is founded on the linguistic action
useful in order to allow the communication among theory, elaborated by John Searle (Searle,
the involved agents. As it will be clearer 1969). An important implementation of the
subsequently, the base idea will be that to make the FIPA standard is the framework JADE
application domain independent in comparison to (http://jade.cselt.it) an open source platform
the definition of the key elements that make possible for peer-to-peer agent based communication
the communication (DDS). Besides particularly developed by Telecom Italia Lab.
interesting is the sharing of an only one knowledge ƒ Conceptual Model Layer: it is very important
base among several agents of the system. to identify the domain where agents operate
After having introduced in the section 2 the state and it is important the dynamics of
of the art to the multi-agent systems, in the section 3 interactions between agents.
will be introduced three alternatives considered as it
Currently, particularly interesting within the
regards the positioning of the knowledge base on the
modelling of the multi agent systems it results:
various agents that constitute the system. In the
section 4 we present the architecture of the multi- ƒ The BDI Model (Beliefs-Desires-Intentions)
agent system, finally, in the section 5, we present the (Chang-Hyun, Guobin et. Al, 1969): it
conclusions of the paper. considers the agent environment belief, which
is the result of its knowledge and perceptions,
and a set of Desires. Intersecting these two
sets, we obtain a new set of intentions, which
2 BACKGROUND can become actions.
ƒ Tropos (Bresciani, Perini et. Al, 2004) a
The e-business systems have not been thought and software development methodology founded
developed, until now, according to a multi-agent on concepts used to model early requirements.
logic. It is useful, however, to present the state of the In particular, the proposal adopts Eric Yu's
art related to the multi-agent systems in order to modelling framework, which offers the
understand the problem list related to the notions of actor, goal and dependency, and
development of a system of this type. uses these as a foundation to model early and
In a multi-agent system we may speak about 4 late requirements, architectural and detailed
main aspects: design. The language used in Tropos for the
ƒ decisional aspects related to the agent: what conceptual modelling is formalized in a meta-
actions an agent undertakes depending on the model described with a set of UML class
external environment; diagram.
ƒ outsourcing of the execution: possibility that In literature there are several papers that examine
more agents collaborate together, performing the software engineering paradigm applied to the
elementary actions to complete a complex multi-agents system, among them an article of Pratik
task; K. Biswas (Patrik, Biswas, 2007) describes

102
 AN ONTOLOGY-BASED ARCHITECTURE FOR MULTI-AGENT SYSTEM ENVIRONMENT

extensively the multi-agent system elements and ƒ presence of a supervision node from which it is
their correlations, confronting them with UML possible to access all the information and to
paradigm that it reuses for their modelling. provide precise commands to the other agents;
Very interesting in a multi-agent system is the The disadvantages are:
ontological approach very useful to provide an ƒ The supervisor node results a critical Point of
explicit and formal representation of the domain. Failure, in fact if the supervisor node had to
This representation is simple to realize and it is come less for some reason the knowledge base
simple to exchange between agents thanks the would result unreachable from the other nodes
ontological languages such OWL (W3C, 2004). and, therefore, the overall system go down.
ƒ Agents have little decision autonomy.

3 USE OF KNOWLEDGE BASE IN 3.2 Knowledge Base Total Distribute

MULTI-AGENT SYSTEM In a structure that introduces an high degree of
distribution, all the agents work in independent way
An e-business system thought in the multi-agent and the exchange of data it has the goal to make
terms it will have, for the complexity of the possible the coordination among the agents. All the
scenarios in which it will operate, an elevated degree agents are in communication exploiting a system of
of hardiness and modularity. To reach this goal it is connection reliable (the DDS) that limit the Point of
important to make more independent possible the failure. At ontology level, for this scenario two
various agents (nodes) of the system, that they need different solutions can be identified, one that
a continuous interchange of data in order to reach foresees to repeat the whole knowledge base of
their own goal. domain on every agent and the other is to foresees
The first problem to face is that to add to a the distribution of the domain ontology so that to put
multi-agent system a layer of "intelligence" that is on every agent only the part of knowledge base that
able to provide a good level of autonomy and facility interests the specific agent for the carrying out of
of updating the system, besides it provides the his/her own role in the system.
possibility to describe the several events that are
verified inside the system and to which the agents 3.2.1 Knowledge Base Replied on the Agents
must answer. This level of intelligence is constituted
by an ontology (opportunely supported by a system In this structure, in which knowledge base is repeat
of rules) that it results particularly useful to on every agent every node of the system, knows the
guarantee a complete understanding of the domain to whole domain.
all the interested agents, and accordingly one swifter The advantages of this approach are:
change of his in case of changes or the insertion and ƒ Non-existence of a Point of Failure;
management of unexpected events. ƒ The autonomy degree of the agent enhance
In the use of ontology in the e-business ƒ All the agents have peer decisional ability
systems based in multi-agents it is had to analyze exploiting the knowledge base;
with attention as it must be defined and above all if The disadvantages are:
this ontology has to be positioned only on a node
inside the system or must be distributes on different ƒ Every change on the knowledge base has to be
nodes. We describe 3 hypotheses of work repeat on every agent;
individualized defining among them the most ƒ The system needs a good structure of
suitable to the context in which we work. coordination;
ƒ The complexity of each agent enhances.
3.1 Centralized Knowledge Base 3.2.2 Knowledge Base Distribute on the
Agents
The centralized system foresees that all the agents
that cooperate make reference to a knowledge base Another solution foresees the distribution on the
centralized on only one agent. This system involves several agents of the description of the domain that
that, all the information push through the central must be decomposed in modules depending on the
node which stores them in the knowledge base and specific criterions of competence. In other words, it
elaborates them. deals with decomposing in several functionalities the
The advantages of this approach are: ontology and to implement them on the several
ƒ the system is simple to manage; agents based on the demands of these.

103
ICE-B 2008 - International Conference on e-Business

The advantages of this approach are: 4.2 Goal in the Realization of the
ƒ Least redundancies of the data, every agent Architecture
has only the information of which it has need;
ƒ Absence of a single Point of Failure, and The idea that is at the base of the conceived
therefore greater independence of the robots; architecture is born from two fundamental requisite:
ƒ Possible changes to be brought to the ƒ Decouple the decisional aspect related to the
ontological structure of the data base would be publishing/subscribing of well defined
alone on the single agent and not on information, from the technological aspect
everybody. tied up to the necessity to produce on the fly
The disadvantages are: the useful classes to publishing/subscribing
ƒ Planning is complex because each agent is the parameters in the DDS
complex. ƒ To provide an high flexibility level to the
ƒ Limited decisional ability for the agents. system in this way to allow an adaptation of
the same to the different operational scenarios
3.3 The Selected Solution where the architecture could operate, also
without denying the consequential
The selected solution is to have a distributed system potentialities from the existence of a
with a knowledge base repeat on every agent of the middleware able to guarantee the
operational context. In this way the several nodes communication among the various agents.
that operate in the system will be fully autonomous Having decided to use a semantic base replied on
and able to develop his/her own task without every agent of the environment, we decided to
depending in some way on the other participants to decouple the component of the knowledge base of
the mission, also communicating and exchanging domain from that related to the management and
data, in continuous way, with the other agents. representation of the concepts that describe the
middleware of communication (DDS).
The proposed decoupling makes the system
4 LOGICAL ARCHITECTURE independent from the particular operational context,
OVERVIEW providing a flexible and easily adaptable structure.
The proposed architecture is in fig. 1.
Before introducing the logical architecture
conceived it is fundamental to clarify shortly the
operation of the middleware of communication
selected.

4.1 Use of the DDS as Tool for the

Communication among Agents
The DDS is a useful tool proposed by the OMG that
enable the communication within data-centric
systems. The system of communication is
asynchronous and it is based on the
publishing/subscribing protocol: when agent that
operates in the system has the necessity of a data, it
makes a subscribing of it pointing out the data of
which it has need; when some other agent of the
system makes the data available, it effects a
publishing of it. The agent that has previously
effected the subscribing is able, to this point, to get
the data of interest. The data publishing/subscribing
is encapsulated in a class and send to the DDS. Since Figure 1: Logical architecture.
information may change depending on the particular
operational context, also the relative First of all we observe the presence of two layers
publishing/subscribing class can vary depending on one called "domain independent" and one called
the particular data to send to the DDS. "domain dependent". Within these two layers,

104
 AN ONTOLOGY-BASED ARCHITECTURE FOR MULTI-AGENT SYSTEM ENVIRONMENT

interconnected through a system of rules as will be specific necessities, from the information obtained
detailed after, the presence of two knowledge bases following inference, from the KB domain
is observed: dependent. For example, if it is important, for the
ƒ Knowledge Base Domain Dependent: this information system to publish/subscribe a well
module represents the KB specific of a special specific data obtained by the sensor, for example the
operational context. temperature data, the information useful to make a
ƒ Knowledge Base Domain Independent: this topic for the temperature will be moved from the
module represents several elements that knowledge base domain dependent to the knowledge
constitute the DDS. base domain dependent.
We can observes, within the layer "domain
dependent" the presence of an "agent". It represents 4.5 Jena Reasoner Module
a generic agent that will contain both the knowledge
base domain dependent and the knowledge base The Jena reasoner module have the task to realize,
domain independent and that will make operation of through the application of ad-hoc SWRL rules
publishing/subscribing of the information on the written by the business expert, the deductions on the
DDS. knowledge base that allow to intercept events that
must happen and that they require, eventually, of a
4.3 Ontology Domain Dependent publishing/subscribing of information on the DDS
with the goal to complete one determined activity.
Module With an opportune analogy, every information
system within an e-business system will deduce
The knowledge of the context, in which the agents
from the domain ontology what information to
work, and of its dynamics, it allows to be able to
publish/subscribe in order to communicate with the
manage in the best way the events that can be
other systems.
verified in it. The module has the goal to represent
the application domain in which the agents operate
and the possible semantics relationships existing 4.6 Rule Engine Module
among the information that the various agents could
The rule engine module will provide, through the
exchange. In this module it will be present,
therefore, this domain ontology that must have opportune rules defined in the SPARQL language, to
realized from the expert of domain that, better of extract from the KB domain dependent the
information to publish/subscribe and will add
everybody, it is able to define every aspect of it. In
the knowledge base it is possible to find the specific individuals to the ontological classes that represent,
information about the domain and their semantic instead, the DDS (KB domain independent).
relationships as an example if the domain is an
information system that manage data about 4.7 Java Class Generator Module
environment, the concept in the knowledge base will
be the sensor used to obtain data or the area where Since the system of publish/subscribe of the DDS
data will be obtained and these two concepts will be founds him, of fact, on a system of publish/subscribe
related each other in order to provide the semantic of classes, the presence of this module is
link. For each sensor it will be possible to define the fundamental because it able to produce, beginning
period of time when the data will be obtained and from the individuals of the KB domain independent,
the threshold values defined in the specific context. the useful classes to make the operation of
These information will constitute a small part of the publish/subscribe. The class draws the information
overall domain ontology that describe the specific that the domain expert associates to every topics
information system. (“topics” are the elements that able the
communication in the DDS) this information,
through the rule engine module, is repeat in the KB
4.4 Ontology Domain Independent
domain dependent and from here used for the
Module creation of the relative classes.
In this module there is the ontology that represents
the specifications of the substratum of
4.8 Description of the Information
communication chosen for this architecture. This Flow
knowledge base reproduces, of fact, the OMG
specification for the DDS: this knowledge base will The information flow in the architecture can be so
be completed with the individuals, depending on the defined:

105
ICE-B 2008 - International Conference on e-Business

ƒ The agent captures the information stored in every informative system (which those present in the
the knowledge base domain dependent; database) in format compatible with those described
ƒ The Jena reasoner module makes the in the ontology domain dependent? And what mean,
appropriate deduction through the SWRL rule practically, to describe a domain?
engine; Parallels to these problems, the problem of
ƒ The RULE ENGINE module defines the reasoning is still open: few has been done in
mapping rules between the two knowledge international scientific community in this sense so
base (domain dependent and domain much that doesn't exist, until now, a standard
independent). language universally recognized for realizing the
ƒ The java classes generator generates the .java reasoning. In this architecture SWRL is used but this
file and the corresponding .class file; doesn't exclude, in a next future, the use of a
ƒ The .class file will be used to send information language more efficient.
through the DDS. In every case, is interesting the idea at the base
It is important to observe that the separation of the present paper that consists of using in an e-
between knowledge base domain dependent and business system the formal ontologies and the
knowledge base domain dependent increases the mechanisms of reasoning.
flexibility level of the system. Changing the context
of business in which the system operates mean to
define the knowledge base domain dependent and ACKNOWLEDGEMENTS
the system of reasoning (module rule engine and
Jena reasoner) leaving the dynamics of We would thank SSI (Space Software Italia)
communication made up through the DDS. company and Antonella Falzone for the tangible
support.

5 CONCLUSIONS AND FUTURE

WORKS REFERENCES
In this paper we propose an architecture conceived Bresciani, P., Perini, A., Giorgini P., Giunchiglia, F.,
for the communication of a data centric multi-agent Mylopoulos, J. 2004 Tropos: An Agent-Oriented
systems that they use as middleware of Software Development Methodology. In Autonomous
Agents and Multi-Agent Sytems, 8, 203–236, 2004
communication the DDS recently proposed by
Chang-Hyun, J., Guobin, C., James, C. (2004) A New
OMG. Approach to the BDI Agent-Based Modelling In 2004
Particularly interesting is the analogy among the ACM Symposium on Applied Computing
data-centric systems and the multi-agent systems Data Distribution Service for Real-time Systems Version
applied to an e-business context. The analogy can be 1.2 OMG Available Specification formal/07-01-01
done thinking about every information system FIPA ACL Message Structure Specification 1996-2002
involved in the e-business environment as a single Pratik K. Biswas, Towards Agent-Oriented
agent of the architecture here presented. In this way Conceptualization and Implementation. 2007, IGI
it will be possible to bring on a e-business system Global,
Searle, J. 1969 Speech Acts: An Essay in the Philosophy
the whole efficiency and the flexibility that the
of Language, Cambridge University Press, 1969 ISBN
proposed architecture introduces: the great 052109626X
advantage obtainable is the separation between the W3C (February, 10 2004) “OWL Web Ontology language
KB domain independent and the knowledge base Reference” W3C
domain dependent, another advantage is the
possibility to produce in automatic the useful classes
to make the data publishing/subscribing of the data.
Naturally, repeating the architecture proposed on
an e-business system there are many aspects to
clarify that brings to several research ideas.
The first problem is surely tied to the semantic
interoperability: it is necessary to build a knowledge
base that describes the whole domain in which every
information system participates but as it is possible
to effect the transfer of the information proper of

106
 SECURE IT/TELCO ENVIRONMENT PLANNING MADE EASY
A Concept of a Tool for Planning Secure IT/Telco Infrastructure and Applications

Wolfgang Haidegger
SECUDE Global Consulting GmbH, Lassallestrasse 7b, 1021 Wien, Austria
[email protected]

Keywords: e-Business, ISMS, ISO/IEC 27001, ISO/IEC 17799, ISO/IEC 15408, Risk Analysis, Information Security
Policies, Security Guidelines and Procedures, Evaluation Assurance Level.

Abstract: This paper first motivates the necessity of a planning tool for IT infrastructure and applications, which
allows the inclusion of security measures in an automated way. Then the author summarizes the
requirements coming from legal and technical standards, which serve as a framework to assure the
compliance of the results of planning activities with the respective applicable regulations. Next, a rough
concept for realizing the planning tool is presented and finally conclusions are presented.

1 INTRODUCTION management protocols and application service

philosophies.
The IT/Telco industry supports e-commerce with a The last two paragraphs show that security issues
selection of electronic business applications aimed at pertaining to information stored of transferred within
commercial transactions. The following list networks for the purpose of doing electronic
enumerates some of the better known business business can get arbitrarily complex both on a
applications, but is by no means exclusive: technical and a procedural level. At the same time
the need for security grows as more and more
ƒ electronic funds transfer, personal (and sometimes very private) data is
ƒ supply chain management, involved in e-commerce transactions. This is also
ƒ e-marketing, reflected by the fact that compliance requirements to
ƒ online marketing, national and international regulations concerning
ƒ online transaction processing, confidentiality, availability and integrity of
ƒ electronic data interchange (EDI), information become stricter.
ƒ automated inventory management systems, and Taking the complex surroundings just described into
ƒ automated data collection systems. account, then in order to carry out a proper analysis
From a technical point of view this paper of security requirements and planing of the
distinguishes between the network services (also according security measures a comprehensive tool
called “infrastructure”) and the application services, (or tool chain) needs to be developed, which
which are necessary to implement business guarantees
applications as listed above. This distinction is ƒ technical correctness,
important, as it mandates different technical and ƒ compliance to all relevant regulations and
procedural security measures. ƒ proper tailoring to the business needs
In addition to distinguishing the type of service
necessary (infrastructure or application) to put for the solution found.
together a business application, one is very often A final remark: The author of this paper does not
faced with a converged IT/Telco landscape distinguish between Greenfield analysis and analysis
concerning both types of services. This means one of existing infrastructure and applications
has to accommodate for the security needs of both concerning the capabilities of the planning tool, as
circuit switched and packed oriented networks with this does not seem to be relevant for a preliminary
their different transport and control planes, treatment of the topic.

107
ICE-B 2008 - International Conference on e-Business

2 THE NORMATIVE The following subsections give an idea about the

legal and technical recommendations the tool (or
FRAMEWORK tool-set) will have to be able to abide to. They
represent additional, project external constraints.
The following paragraphs show the steps necessary
for risk analysis and mitigation. These have been 2.1 International Legal, Procedural
described in different level of detail but with the
same general intention in ISO/IEC FDIS 27001,
and Technical Standards
ISO/IEC 15408 and ASIS.
There are many international legal standards, which
Figure 1 gives an overview over the procedure
have to be considered as important when treating
the steps are embedded in.
IT/Telco security. Some examples are:
ƒ Treaties of the European Union,
ƒ European Convention for the Protection of
Human Rights and Fundamental Freedoms,
ƒ European Directives
To show one important European standard in the
area of Identity Management:
The European Directive 95/46/CE: deals with
data protection, is aimed at giving to the data subject
Figure 1: Process, which the tool has to support. (owner of data) the most control possible on its own
identity and personal data, posing a series of
Step 1: Risk Analysis requirements on recipients, controllers, processors
ƒ Assets, their value and their owners are and even third parties. Art. 2, letter a), giving a
identified. definition of "personal data", says: "identifiable
ƒ Vulnerabilities of the assets as well as threat person is one who can be identified, directly or
agents, which might exploit the indirectly, in particular by reference to an
vulnerabilities, are identified. identification number or to one or more factors
ƒ For the resulting risks possible countermeasures specific to his physical, physiological, mental,
are identified and the appropriate ones are economic, cultural or social identity".
selected. There are also quite some procedural (non-
technical) international standards, which will play a
Step 2: Definition of Mitigation Measures role when planning security measures. Two
ƒ Development of security policies examples are:
ƒ Analysis of relevant international, national and
corporate legal and technical standards ƒ ISO/IEC FDIS 27001: This International
relevant for the situation. Standard specifies the requirements for
ƒ Development of procedures and guidelines, establishing, implementing, operating,
designing the way the security policies shall monitoring, reviewing, maintaining and
actually be realized. improving an IT security management plan
focussing on the overall business risks of the
Step 3: Integration into Service/Infrastructure organization the plan is made for.
architecture (not shown in figure 1) ƒ COBIT: The Control Objectives for
ƒ Design of infrastructure with selected security Information and related Technology (COBIT)
measures provides a set of generally accepted measures,
ƒ Design of applications according to selected indicators, processes and best practices to
security measures assist
o in maximizing the benefits
Step 4: Verification (not shown in figure 1)
derived through the use of
ƒ Selection of appropriate Assurance Level information technology and
ƒ Development of Evaluation Assurance Level o developing appropriate IT
Criteria corresponding to the Assurance Level governance and control in a
selected company

108
 SECURE IT/TELCO ENVIRONMENT PLANNING MADE EASY - A Concept of a Tool for Planning Secure IT/Telco
Infrastructure and Applications

International technical standards will come from 3 CONCEPT FOR THE

ISO, IETF, ITU-T, OMA, TMF, W3C and other
international standardization bodies or industry
REALIZATION OF THE
forums. PLANNING TOOL
2.2 National Legal Standards The threat analysis will be based on a database that
holds information about known threats to the
Every nation will have a set of national legal specific system components, their architecture,
standards, which have to be considered as important protocols. This database will also be filled with the
in addition when treating IT/Telco security. This results of the tasks dealing with risk assessment.
time the examples are selected from the USA, again There are tools today, which do this on a very
with the focus on relevance to Identity Management: general level (e.g. COBRA). The main challenge
will be to drill down to implementation level if
ƒ Privacy Act of 1974: all government agencies - possible and/or necessary.
federal, state and local - which request social Furthermore a network/application planning tool
security numbers are required to provide a will be developed based on the threat database
disclosure statement on the form; above, which is capable of identifying threats to a
ƒ Family Educational Rights and Privacy Act planned network, suggesting methods to mitigate the
(FERPA, also known as the "Buckley threats according to a specific Evaluation Assurance
Amendment," enacted in 1974, 20 USC Level and including the results into the network
1232g): social security numbers fall within the plan, a threat model document and a test
scope of personally identifiable information specification. Here two issues will be of main
that is restricted from disclosure by schools concern:
that receive federal funding under the Family
Educational Rights and Privacy Act; ƒ Developing solution variants, which really fit
ƒ Children's Online Privacy Protection Act the topic and providing guidance for the
(COPPA) - 15 U.S. Code 6501 et seq.: The selection of the “right” solution.
act's goal is to place parents in control over ƒ Using the selected Evaluation Assurance Level
what information is collected from their as the driving parameter for the automated
children online; solution development
ƒ Financial Services Modernization Act, Gramm-
The tool will be capable of tracking the changes
Leach-Bliley (GLB), Privacy Rule - 15 USC to the planned network, the according changes in the
6801-6827: The 1999 federal law permits the threat model, and the changes in the security
consolidation of financial services companies
measures to be taken and finally, track the changes
and requires financial institutions to issue in the documentation (network plan, test
privacy notices to their customers, giving specification).
them the opportunity to opt-out of some
sharing of personally identifiable financial
information with outside companies;

2.3 Corporate Standards

Finally, organizations whose security needs to be
planned or evaluated and updated will probably have
additional regulations, which need to be taken into
account. For example it might be necessary for every
employee to wear a badge with her or his photo id
on it to guarantee an additional possibility to identify
her or him. Figure 2: General architecture of the tool (tool chain).

Figure 2 give a first impression of how the tool

chain shall be designed. The main part will be an
intelligent central data base, which will serve all
planning applications likewise.

109
ICE-B 2008 - International Conference on e-Business

4 CONCLUSIONS
The paper shows the necessity of a general IT/Telco
security planning tool (or tool chain) as a
consequence of heightened complexity concerning
the infrastructure and the applications as well as
tightened security and compliance regulations.
The regulatory framework, which such a tool
chain has to accommodate, is dealt with
conclusively concerning the types of regulations, but
only exemplary concerning the actual
recommendations.
Finally a first idea of a possible realization is
given together with the crucial points to solve.
Next steps will be the collection of the technical
ingredients of the database (protocols, HW
architectures, SW architectures, SW frameworks, ...)
and a first design of the database itself. This will
have to happen in the light of the applications, which
will use the database.

REFERENCES
ISO/IEC FDIS 27001: “Information technology —
Security techniques —Information security
management systems — Requirements” Final Draft
2007
ISO/IEC 17799: “Information technology — Security
techniques — Code of practice for information
security management” 2005
ISO/IEC 15408: Common Criteria for Information
Technology Security Evaluation Version 3.1, Revision
1, September 2006 (Part 1 – Part 3)
ASIS: The General Risk Assessment Guidline, ASIS
International, November 13-th 2002, ASIS GLCO 01
012003, Sean Ahrens et al

110
WEB AND MOBILE BUSINESS
SYSTEMS AND SERVICES
FULL PAPERS
 MOBILE TOURISM SERVICES
Experiences from Three Services on Trial

Niklas Eriksson and Peter Strandvik

Institute for Advanced Management Systems Research, ÅboAkademi University, Jouhkahainenkatu 3-5A, Turku, Finland
[email protected], [email protected]

Keywords: Mobile tourism services, technology adoption.

Abstract: For this study a field trial was conducted to identify the determinants for tourists’ intentions to use three trial
services targeting tourists on tour, in this case on the Åland Islands in Finland. We identified that the major
barrier for the non usage of the trial services was linked to the type of travel that the trial group participated
in. Also price transparency and ease of use especially ease to take new mobile services into use should be
highlighted in mobile tourism service development. Moreover, we came across some basic reminders to
take into account when commercializing mobile services, such as carefully define a customer target group,
estimate potential usage volume and plan marketing / sales tactics. These aspects are not necessarily
realized enough in technology development.

1 INTRODUCTION government of the Åland islands, is a development

programme of increasing knowledge, production and
The use of the Internet for doing commerce or use of new interactive media on the Åland Islands1
interacting with customers has been growing rapidly in Finland. Within the project several mobile
in the world wide tourism industry. Mobile applications have been developed for the travel and
commerce, or e-commerce over mobile devices, on tourism sector on the islands. Three of these services
the other hand has had many conflicting predictions will be presented more in detail in this paper:
on its future popularity. Most predictions have been MobiPortal, TraveLog and MobiTour. A field trial
overly optimistic. However, the benefits that arise of these services with real incoming tourists to the
from mobile technology have not yet been fully Åland Islands using their own mobile phones has
delivered, which to some extent is explained by the also been conducted. Findings and experiences from
fact that mobile applications, due to complexity or this trial will be reported. Possible determinants for
lack of relevance, fail to meet customers’ consumers’ intentions to use mobile tourism services
expectations (Carlsson et al. 2006). Travel and will be discussed as well.
tourism is an industry in which several different
projects have been conducted where mobile
applications have been developed, tested and 2 SERVICE DESCRIPTIONS
implemented, some even with moderate success (e.g.
Ardissono et al 2003, Kramer et al 2005, Schmidt- The services have been planned with a common
Belz et al 2003, Repo et al 2006). Some of these logic namely the Braudel rule: freedom becomes
pilot projects (e.g. Kramer et al 2005, Schmidt-Belz value by expanding the limits of the possible in the
et al 2003) have been focusing on GPS which the structures of everyday life (as presented by Keen &
average tourist doesn’t yet have in his/her handheld Mackintosh 2001). The rule is then translated into a
mobile device. Therefore it seems relevant to build tourism setting which means that tourists’ real or
and test services that actually can be used by the perceived need has to be met by the services and
average tourists. Nevertheless previous pilots have
given us valuable information on the potential of 1
Åland is an autonomous and unilingual Swedish region in
mobile technology.
Finland with its own flag and approximately 26.700 inhabitants.
The New Interactive Media (NIM) project, with Åland is situated between Finland and Sweden and consists of 6
funding from the European Union and the regional 500 islands. (www.visitaland.com)

115
ICE-B 2008 - International Conference on e-Business

moreover, the services need to profoundly change Perceived ease of use is defined as “the degree to
the way a tourist does or experience something – and which a person believes that using a particular
to the better (Harkke 2007). system would be free of effort”. The two TAM
MobiPortal is a mobile version of an information determinants are proposed to identify the intended
portal www.visitaland.com which is the official usage behavior of a system and are widely used as a
tourist site of the Åland Islands. The portal includes backbone for research in adoption of technology.
search for events, restaurants etc., a map service and However, the first TAM variable perceived
facts on the Åland Islands. usefulness is foremost designed to research work
TraveLog is a mobile community for incoming performance improvements in organizational
tourists to share experiences from the Åland Islands contexts. In consumer markets consumer behavior is
with each other. The virtual meeting place includes also influenced by other factors. It is typical that
stories, pictures, tips and interactions. non-efficiency factors impact consumer adoption of
MobiTour is a guide for attractions such as the technology, e.g. good tourist technologies are not
Bomarsund fortress which is downloadable / only those that make tourists more efficient, but that
streamable to the visitors’ own devices. The guide also make tourism more enjoyable. Thus tourism can
includes voice and/or video guidance. be characterized as wandering, where tourists
All these three services ought to expand the attempt to enjoy the city environment and chance
limits of a tourist to the Åland Islands according to upon things of interest, rather than optimizing
the Braudel rule by enabling 1) instant access to (Brown & Chalmers 2003). As the mobility (on the
local information, 2) enhanced communications with move) capability is generally seen as the key value
other people with the same interests and 3) driver in m-commerce (Anckar & Eriksson 2003),
experience enhancement for certain unmanned mobile technology clearly has the potential to
attractions. Especially experience enhancement support the wandering aspect of tourism. A word
features are generally seen as key drivers for like flexibility has commonly been used to describe
successful customer satisfaction in tourism (Pine & the independence of time and space that is provided
Gilmore 1999). The determinants for consumer by mobile technology. According to Kim et al.
usage of mobile tourism services are, however, a (2005) the hedonic motivation or the enjoyment
complex issue which will be discussed next. aspect of tourism has, however, not been clearly
defined in mobile technology acceptance models.
The perceived type and degree of perceived value of
3 POSSIBLE DETERMINANTS a mobile service depend on the other hand on the
situation or context of usage (Mallat et al 2006, Lee
Several models of technology adoption have been & Jun, 2005). Anckar & Dincau (2002) introduced
developed. One of the most used models is the an analytical framework that identifies the potential
technology acceptance model (TAM) by Davis value creating features of mobile commerce. Mobile
(1989) which is based on the theory of reason action value elements in the framework for consumers on
(TRA) by Fishbein et al. (1975). Other often used the move are: Time-critical arrangements,
models in technology adoption research are the Spontaneous needs, Entertainment needs, Efficiency
diffusion of innovations theories (DIT) by Rogers ambitions and Mobile situations. Time-critical
(1995) and the unified theory for the acceptance and arrangements refer to applications for situations
use of technology (UTAUT) by Venkatech et al. where immediacy is desirable (arise from external
(2003) which combines TAM with other acceptance events), e.g. receive alerts of a changed transport
model e.g. DIT. Here different components of these schedule while on tour. Spontaneous needs are
models will be discussed, together with relevant internally awakened and not a result of external
research theories for adoption of electronic and events, e.g. find a suitable restaurant while
mobile services, to identify possible determinants for wandering around. Entertainment needs, killing
consumer intentions to use mobile tourism services. time/having fun, especially in situations when not
The TAM model proposes two determinants, being able to access wired entertainment appliances,
perceived usefulness and perceived ease of use, e.g. kill or fill time in transportation. Efficiency
which impact the acceptance of technology and ambitions aim at productivity, e.g. use dead spots
adoption behavior as a result (Davis 1989). during a travel to optimize time usage. Mobile
Perceived usefulness is defined as “the degree to situations refer to applications that in essence are of
which a person believes that using a particular value only through a mobile medium (e.g.
system would enhance his or her performance”. localization services), which ought to be the core of

116
 MOBILE TOURISM SERVICES - Experiences from Three Services on Trial

mobile commerce. Consequently perceived mobile intended adoption behaviour (Featherman & Pavlou
value represent the degree to which a person 2003). Trust, as trust in the service vendor to
perceives value arising from the mobility of the minimize the risks, has also been added to the TAM
mobile medium. model (e.g. Cho et al 2007, Kaasinen 2005) and
Nevertheless not only the medium creates value pointed out as a strong influencer on the intended
for the consumer but the essence of the services as use of mobile services due to that mobile commerce
well. We refer to such value as perceived service is still at its initial stage (Cho et al. 2007). We refer
value. For example for a tourist in a planning or to trust as the perceived risk defined by Featherman
booking situation the key to successful satisfaction & Pavlou 2003. They divide the perceived risk for
would be timely and accurate information relevant to electronic services into the following elements;
the consumer’s needs (Buhalis 2003). Equally performance risk, financial risk, time risk,
important for a tourist visiting a historical attraction psychological risk, social risk and privacy risk.
may be the satisfaction of educational and Performance risk refers to the possibility of a service
entertainment (edutainment) needs (HyunJeong & to malfunction and not performing as it was
Schlisser 2007). Similarly a person with a mission to designed and advertised. The financial risk refers to
share experiences with others may find satisfaction the potential monetary outlay associated with the
when a community responds (Arguello et al 2006). initial purchase price as well as the subsequent
The three examples refer to the essence of the three maintenance cost of the product and the possibility
services on trial. of fraud. Time risk refers to that the consumer may
The second TAM determinant perceived ease of lose time when making a bad purchasing decision
use has been widely discussed in mobile commerce. e.g. by learning how to use a product or service only
Limitations of mobile devices (e.g. screen size) to have to replace it if it does not perform to
cause consumers to hesitate whether to adopt mobile expectations. Psychological risk refers to the
commerce or not. According to Cho et al (2007) potential loss of self-esteem (ego loss) from the
device limitations suggest that focusing on easy to frustration of not achieving a buying goal. Social
use mobile applications could enhance the consumer risk refers to potential loss of status in one’s social
acceptance of mobile commerce. Kaasinen (2005) group as a result of adopting a product or service,
points out that mobile services need to be easy to looking foolish or untrendy. Privacy risk refers to
take into use as well as mobile services are typically the potential loss of control over personal
used occasionally and some services may be information, such as when information about you is
available only locally in certain usage environments. used without your knowledge or permission. At least
As a consequence, information on available services security and privacy issues have been highlighted as
should be easy to get and the services should be easy barriers to mobile commerce (O’Donnell et al.
to install and to start using. The ease of taking a 2007). Also financial risks in form of high costs,
service into use may in fact have a direct impact on including operating costs and initial costs, have been
the adoption behaviour of a mobile service highly ranked by consumers as hindrances for m-
(Kaasinen 2005). On the other hand when problems commerce in its early stages (Anckar et al. 2003).
arise, users in the consumer market are often In UTAUT social influence among other
expected to solve the problems on their own (Repo constructs is added to the two TAM components and
et. al 2006). Consequently the use may rely on defined as the degree to which an individual
proper instructions or on a helping hand from perceives that important others believe he should use
someone. Proper support conditions also in a the new system (Venkatech et al., 2003). Social
consumer market may therefore be important influence is also known as subjective norm in the
especially for advanced mobile services. theory of reason action (Fishbein et al 1975) and in
Nevertheless consumers many times expect to take a its extension theory of planned behavior (Arjzen
new product or service into use without instructions 1991). In consumer markets image and social status
or help. have been proposed to impact consumers’ adoption
According to Rogers (1995),”The innovation- of mobile services (Teo & Pok 2003). Also the
decision is made through a cost benefit analysis number of users may influence, especially for
where the major obstacle is uncertainty”. Perceived community services which usefulness heavily
risk is commonly thought of as felt uncertainty depend on activity of different participants
regarding possible negative consequences of using a (Pedersen et al 2006). Furthermore other external
product or service and has been added to the two sources such as media reports and expert opinions
TAM determinants as a negative influencer on

117
ICE-B 2008 - International Conference on e-Business

may influence consumers’ perception of electronic tourism services are: perceived mobile value and
services (Bhattacherjee 2000). service value, perceived ease of use, social
Demographic variables such as gender and age influence, perceived risk and tourist characteristics.
are commonly used in consumer research. For Mobile value and service value replace perceived
example gender and age might through other usefulness as presented in the TAM model. Ease of
constructs influence the intended adoption behavior use is defined as in the TAM model where also ease
of mobile services (Nysveed et al. 2005). According of taking a service into use is included. Social value
to the theory of planned behavior (Ajzen 1991) is defined as in UTAUT and perceived risk as
control beliefs constitute individuals’ belief that they presented by Featherman & Pavlou (2003). Tourist
have the necessary resources and knowledge to use characteristics constitute key characteristics of an
an innovation. For example skills or earlier individual on-tour. The defined determinants are
experience of using mobile services may influence summarized in table 1.
the adoption intentions of new mobile services.
When discussing consumer behavior in tourism and Table 1: Possible determinants for consumer intentions to
the impact of information and communication use mobile tourism services.
technologies (ICTs) a clear distinction should also Mobile value: the degree to which a person perceives
be made between experienced and inexperienced value arising from the mobility of the mobile medium
travelers (travel experience). The first group mainly Service value: the degree to which a person perceives
feels more comfortable organizing their holidays and value arising from the essence of the service.
thereby taking advantage of ICT tools available to Ease of use: the degree to which a person believes that
them more easily (Buhalis 2003). Moreover using a particular service would be free of effort
inexperienced destination travelers usually need a lot
Risk: the degree to which a person feels uncertainty
more local information. Innovations also need to
regarding possible negative consequences of using a
comply with the existing values and needs of the service.
individual in an everyday life setting (Moore & Social influence: the degree to which an individual
Benbasat 1991), in this case while on tour. For perceives that important others believe he should use
example the values of the individual may differ the service
depending on the type of travel they are on: leisure Tourist characteristics: Demographics, Experience of
or business, where the former ought to call for mobile services, Travel experience, Destination
services with enjoyability rather than efficiency. In experience, Type of travel, Personal Innovativeness,
consumer markets mobile services also compete Device readiness
against existing and constantly developed
alternatives. Thus consumer habits are usually quite
slow to change from known alternatives (Dahlberg 4 THE FIELD TRIAL SETUP
& Öörni 2007). People are on average risk-averse.
But that is not true for everyone as we have According to Repo et al. (2006) TAM theories and
individuals who are earlier to adopt new ideas than similar approaches have little relevance in the real
others (Rogers 1995). Such personal characteristics product development process. Product developers
make diffusion of innovations possible. Personal need first hand user feedback in form of personal
innovativeness is the willingness of an individual to interaction rather than by reading research reports.
try out and embrace new technology based services. The arguments are based on experiences from
Individuals’ limited mobile device readiness has as piloting a mobile blog service for tourists, where the
well been seen as a great negative influencer of the user gave direct feedback to the developers orally
usage of more advanced mobile services (Carlsson et and through survey forms. Involving the consumer
al. 2004). We refer demographic variables, in the development process of products or services
experience of mobile services, travel experience, can be very rewarding indeed (von Hippel 2005).
destination experience, type of travel, personal With the theoretical foundation (Table 1) in mind
innovativeness and user device readiness as and with the idea of directly interacting with the
discussed here to tourist characteristics as they consumers to receive direct and spontaneous
illustrate key characteristics of an individual that feedback to the product developers we designed a
may influence the intended use of mobile tourism field trial which included oral, observed and survey
services. data collection.
Based on the literature discussion possible The trial was conducted during a conference in
determinants for consumer intentions to use mobile the capital of the Åland Islands Mariehamn 21 –

118
 MOBILE TOURISM SERVICES - Experiences from Three Services on Trial

22.9.2007 at the legislative assembly where the main Excellent [2] Good [3] Satisfactory [4] Not at
activities of the conference were held. The all.
conference was arranged by the local Junior ƒ Type of travel: if they consider
Chamber of Commerce organization and it was WestCongress2007 to be: [1] a leisure trip [2] a
called WestCongress2007. Members of similar business trip.
organizations in the western regions of Finland were ƒ Personal Innovativeness: Three statements were
invited to attend the conference. A total of 191 proposed on a five point scale: [5] definitely
participants had registered in advance for the agree - [1] definitely disagree: I want to get
conference. The trial was coordinated in cooperation local information through my mobile phone
with the conference director who offered assistance when… 1. I plan my program e.g. in the hotel 2.
with e.g. stand preparations and informing the I’m on my way to a local place with e.g. bus 3. I
participants in advance of the mobile services in get acquainted with a local place on foot. The
conference guides, online and during registration. statements were developed based on the kind of
Our stand was set up at the main entrance of the mobility situations tourists may experience.
building where the main activities were held. The Kristoffersen & Ljungberg (2000) distinguish
main entrance was the place that we anticipated between three types of mobility: visiting,
would be the busiest during the first parts of the traveling and wandering. Visiting, an actor
conference when we were invited to promote and performs activities at different locations (e.g. a
demonstrate our services. The stand was equipped hotel). Traveling, an actor performs activities
with a video projector showing animated picks of while moving between different locations
the services and also flyers, tables and chairs for usually inside a vehicle (e.g. bus). Wandering,
comfortable discussions with the conference an actor performs activities while moving
attendants. between different locations where the locations
At our stand the conference participants were are locally defined within a building or local
informed more in detail of the services. The services area (e.g. on foot).
were also demonstrated, which gave us a chance to
For the follow up a semi-open web questionnaire
observe peoples first time reactions. The stand also
was used to receive feedback on the participant’s
provided for us a good place to freely discuss
actual use of the three services. The web
different issues regarding the services with the
questionnaire was sent to the participants by e-mail
participants. Participants filled out voluntarily a
two days after the conference finished ensuring that
questionnaire which also was an agreement to
their service experience would be fresh in their
contact them by e-mail after the conference to
minds. A reminder was sent a week later. The
follow up on their own independent use of the
participants were asked to state for each of the three
mobile services during their stay on the Åland
services whether they had used it or not. Their
Islands. Each phone and operator connection (device
answer was followed up with an open question on
readiness) was checked by the stand representatives
their primary motivation for using or not using the
to ensure that the participants actually were able to
service. In the analysis the answers were interpreted
use their own phones for the services.
according to the theoretical foundation on
In the questionnaire the participants were asked
determinants for the intended use of mobile tourism
to fill out questions according to the constructs
services. Additionally the participants were asked to
defined for tourist characteristics:
state what kinds of problems they had run into if
ƒ Demographics: Gender and age problems occurred. The participants were also to
ƒ Experiences of mobile services: Commonly state on a five point likert scale ([5] Yes, definitely -
used services were listed with the alternatives: [1] Definitely not) for each service what their
[1] continuously using [2] have tried [3] have intentions are to use similar services in the future
never tried. while visiting a destination. Finally the participants
ƒ Travel experience: How often they travel for were free to comment on the service.
more than one day: [1] several times a month
[2] ~ once a month [3] 3 – 9 times a year [4] <
three times a year. 5 THE PARTICIPANTS
ƒ Destination experience: If they have visited the
Åland Islands before: [1] Yes, > 5 times [2] Members signed up in advance for the conference
Yes, 2 – 5 times [3] Yes, once [4] Never and were 191 in total. However, about thirty persons
their knowledge of the Åland Islands [1]

119
ICE-B 2008 - International Conference on e-Business

didn’t register. We estimated that about 50 persons services” were given. Especially MobiPortal
visited our stand. Out of these 50 persons voluntarily awakened concrete interest as it was bookmarked by
and without a prize draw 23 filled out the a couple of stand visitors. A few persons also
questionnaire and allowed us to contact them after praised the visual design of the MobiTour guide.
the conference for the follow up. 20 out of 23 However, some people were spontaneously skeptic
persons had a mobile phone and an operator about the long download times for MobiTour. Nor
connection (device readiness) that allowed them to did anyone ask for transactions over Bluetooth
use the services. Thereby it was relevant to send the although it would have been possible at the stand.
follow up by e-mail to these 20 persons. Two mail Several persons instantly also asked for the price of
addresses did not respond. Out of the 18 persons that the services. The services were not charged for and
the follow up went to 9 answered it. it seemed like the transaction costs were obvious to
Of the 23 that filled out the questionnaire 12 most visitors and not a hindrance to use, except for
were men and 11 women. The average age was 35. the large files of MobiTour. Connection problems
The majority (66%) stated that they travel about occurred with at least one network operator which
once a month for more than one day. Most of them interestingly led to that a few thought there was
(66%) had visited the Åland Islands before at least something wrong with the trial services.
two times or more. However, a majority (66%) None of the nine respondents to the follow up
answered that they know the Åland Islands had on their own used any of the trial services. All
satisfactory or not at all. Almost all (96%) felt the reported that their primary motivation for the no use
WestCongress2007 to be a leisure trip. Of the 23 was that they didn’t experience a need to use the
participants all had at least at some point tried to use services during their stay at the conference on the
a mobile service and a clear majority (66%) used at Åland Islands.
least one mobile service continuously. A total of The future intended use of similar services as the
74% (17) of the participants answered that they want ones on trial were reported as shown in table 3.
to get local information with their mobile phone for Services similar as MobiPortal received the highest
at least one of the statements in table 2. score.

Table 2: Local information with mobile phone. Table 3: Intended use of similar services.
I want to get local information through my N %* When visiting a destination in the N Mean*
mobile phone when... Tot 23 future I intend to use …
I plan my program e.g. in the hotel 10 43 Similar services as MobiPortal 9 3,33
I’m on my way to a local place with e.g. bus, 14 61 Similar services as TraveLog 9 2,89
car Similar services as MobiTour 9 2,89
I get acquainted with a local place on foot 11 48
*[5] Yes, definitely – [1] Definitely not
* 5 = definitely agree and 4 = partly agree

7 DISCUSSION
6 EXPERIENCES FROM THE
TRIAL A customer target group needs to be defined for each
mobile service developed (Hoegg & Stanoevska-
To draw peoples’ attention to our stand we really Slabeva 2005). The primary target group for the
needed to sell the services. As people were moving three mobile services on trial is visitors to the Åland
for different things in the building and to other Islands. The trial targeted participants of
locations in the surrounding area a major job was to WestCongress2007 who visited the Åland Islands.
get them to stop by the stand. Very few participants When analyzing the trial group it can be said that it
stopped without a few sales lines from the stand was both right and wrong. It ought to be the right
representatives, although they were informed in group based on the fact that most participants who
advance of the services and the stand was filled out the questionnaire had a device readiness
strategically placed at the main entrance. (87%) that allowed the services to be used on their
Most people who visited the stand expressed a own phone. The group already continuously used
positive response by the first sight of the services. mobile services to a great extent (66%) and thereby
Comments like “that seems practical” and “I already the barrier to take on new services ought to be lower.
use mobile news services so why not use these Their knowledge of the Åland islands was only

120
 MOBILE TOURISM SERVICES - Experiences from Three Services on Trial

satisfactory or none (66%) which ought to create a services were free of charge and the transaction
need for local information. Also their willingness to costs didn’t seem to be a barrier. Nevertheless our
get local information in different situations (74%) experience from this trial is that the service price and
with their mobile phone was positive. Moreover the potential transaction costs must be transparent to the
group was an experienced group of tourists (66%) consumers to minimize uncertainty of the monetary
which generally is found to be positive regarding layout. The monetary aspect may be even more
usage of information and communication important for foreign visitors as transaction cost may
technology. On the other hand the group had a ready rise noticeably.
made program during the weekend and we observed Even though many participants expressed a
that they also asked their hosts for tips and general interest in the services it is also a fact that no
directions. The need for local information and one reported that they actually used the services on
guidance may therefore have been satisfied. trial. Thereby questions are raised from a business
Moreover they had their conference group who they point of view on the potential usage volumes of the
met with continuously to share their experiences services on trial at this time and place. We certainly
with. Consequently the service value of the three need to be very realistic when we launch mobile
services on trial was already met by other means of services on the potential volume of usage, especially
interaction. when setting the business logic (Collan et al. 2006).
The analyses of the trial group indicate that the Moreover product developers need to remember to
same people but with another mission to visit the look at things from a consumer perspective. For
Åland Islands could be a potential user group of the example in this trial the consumers thought the trial
services on trial. The mobile value of using mobile services didn’t work because of an operator
services is, as discussed in the theoretical problem. In the eyes of the consumer this means a
foundation, very much situation based. Moreover, malfunctioning product which is useless. Similarly
the proposed value needs to comply with the user’s long download times to access a service for a
existing on-tour values. In this case self arrangement temporary use may cause the consumer to view the
values by using a mobile phone necessarily didn’t service as too time consuming to take into use.
exist due to the packaged set up of the conference. Neither can we expect consumers to install services
Consequently the type of travel, as packaged or non- in advance as according to Kaasinen (2005), “users
packaged, is therefore to be taken into account as an are not willing to spend their time on something that
influencer of the intended use of mobile tourism they do not get immediate benefit from.”
services. A non-packaged tour ought to comply Consequently the ease of use aspect must be
better with an individual’s values of self highlighted by product developers as mobile tourism
arrangement / service. Nevertheless customized services may be only temporarily used during a visit
mobile services aimed at specific needs of packaged to a destination or a local place.
groups such as conference attendants may indeed
generate value.
The trial also shows that we cannot forget that 8 CONCLUSIONS
new technology innovations very seldom sell
themselves. Much of our efforts at our stand were This paper presented possible determinants for
sales related. Launching new mobile services consumer intentions to use mobile tourism services.
certainly need to be pushed by creating awareness A major driver according to the six identified
among the potential consumers as for any other new determinants; mobile value, service value, ease of
product. Similar pointers have been presented by use, risk and social influence and tourist
Collan et al. (2006): “Hot technology doesn’t sell characteristics couldn’t be determined as no one
itself, it has to be marketed to the consumer in the used the trial services on their own. The major
shape of value adding services that are easy to use”. barrier for the non usage of the three services among
Therefore marketing / sales tactics influence needs the trial group seemed to be linked to the value
to be set as a determinant for consumer intentions to aspect of the packaged tour (type of travel). Based
use mobile tourism services. on the experience from this trial we propose that
Questions on the prices of the trial services were researchers and practitioners especially take the
the most frequent ones asked during the trial. following into account:
Therefore it seems that the financial risk is carefully
accounted for by the consumers in their intentions to ƒ The type of travel is a key aspect in designing
use a mobile tourism service. In this trial the mobile tourism services

121
ICE-B 2008 - International Conference on e-Business

ƒ Marketing / sales tactics influence should be Special Issue on Artificial Intelligence for Cultural
highlighted as a determinant for consumer Heritage and Digital Libraries, vol. 17, no. 8-9, pp.
intentions to use mobile tourism services 687-714.
ƒ Price transparency is an important aspect to Arguello, J., Butler, B., Joyce, E., Kraut, R., Ling, K.,
Rosé, C., Wang, X., 2006: Talk to me: foundations for
minimize consumers’ perceived risk of mobile successful individual-group interactions online
tourism services communities, Proceedings of the SIGCHI conference
ƒ Ease of use aspects should be highlighted even on Human Factors in computing systems, Montréal
more for mobile tourism services as they may Canada
be only temporarily used Bhattacherjee, I., 2000. Acceptance of e-commerce
services: the case of electronic brokerages. IEEE
These pointers can also be seen as reminders in Transactions on Systems, Man and Cybernetics – Part
technology development where basic A: Systems and Humans, vol 30, no 4, pp. 411 - 420.
commercialization routines sometimes aren’t Buhalis, D., 2003. eTourism: information technology for
realized enough. As for any other product defining a strategic tourism management, Prentice Hall, Harlow.
customer target group, estimate potential usage Brown, B., Chalmers, M., 2003. Tourism and mobile
volumes and plan marketing are vital steps in technology. Proceedings of the Eight European
launching mobile tourism services. Conference on Computer Supported Cooperative
Work (Kuutti, K. and Karsten, E. H. Eds.), Helsinki,
It needs; however, to be kept in mind that the
Finland, 14-18 Sep. 2003, Kluwer Academic Press.
experience is based on only one field trial and Carlsson C., Carlsson J., Puhakainen J., Walden P. 2006.
therefore further research in evaluating mobile Nice mobile services do not fly. Observations of
tourism services and similar mobile services is mobile services and the Finnish consumers.
needed. The recruitment of trial users could as well Proceedings in the 19th Bled eCommerce conference.
be done differently. According to Kaasinen (2005) Carlsson, C. Hyvönen, K., Repo, P., Waldén, P., 2004. It’s
ideally users should be allowed to use the trial All About My Phone! Use Of Mobile Services In Two
services freely but it may lead to, as in this trial, to a Finnish Consumer Samples. Proceedings in the 17th
no usage. Therefore some rules on minimum trial Bled eCommerce conference 2004.
Cho, D., Kwon, H., Lee, H-Y, 2007. Analysis of trust in
times should be set up, where additional usage to the
Internet and mobile commerce adoption. Proceedings
minimum can be considered as real usage. Logs can of the 40th Hawaii International Conference on System
also be helpful in data collection to receive prompt Sciences (HICCS)
service usage data in addition to follow up data from Collan, M., Sell, A., Anckar, B., Harkke V. 2006. Using
the respondent. Moreover, phone interviews may E- and M-Business components in Business:
give more extensive answers and better response Approaches, Cases, and Rules of Thumb. Chapter IX
rates in a follow up data collection of the same in Entrepreneurship and Innovations in E-Business:
character as in this trial. An Interactive Perspective. Fang Zhao. Idea group.
Dahlberg, T., Öörni, A. 2007. Understanding changes in
consumer payment habits – Do mobile payments and
electronic invoices attract consumers? Proceedings of
REFERENCES the 40th Hawaii International Conference on System
Science (HICSS) 2007.
Anckar, B., D’Incau, D., 2002. Value creation in mobile Davis, F. D., 1989. Perceived Usefulness, Perceived Ease
commerce: Findings from a consumer survey. Journal of Use and User Acceptance of Information
of Information Technology Theory and Application, Technology, MIS Quarterly, vol 13, no 3, pp. 319-340.
vol. 4, pp. 43-65. Featherman, M., Pavlou, P. 2003. .Predicting e-services
Anckar, B., Eriksson, N. 2003. Mobility: The Basis for adoption: a perceived risk facets perspective. Journal
Value Creation in Mobile Commerce? Proceedings of of Human Computer studies, no. 59, pp. 451–474.
the International Conference SSGRR’03. Telecom Fishbein, M., Ajzen, I., 1975. Belief, attitude, intention
Italia learning Services L’Aquila, Italy and behavior: an introduction to theory and research.
Anckar, B., Carlsson C., Walden, P., 2003. Factors Reading, MA Addison-Wesley
affecting consumer adoption decisions and intents in Harkke, V., 2007. NIM Åland: the quest for useful mobile
mobile commerce:empirical insights. Proceedings in services. Proceedings of 20th Bled eConference.
the 16th Bled eCommerce conference Hoegg, R., Stanoevska-Slabeva, K., 2005. Towards
Ajzen, I., 1991. The Theory of Planned Behavior, Guidelines for the Design of Mobile Services. 13th
Organizational Behavior and Human Decision European Conference on Information Systems (ECIS)
Processes, vol. 50, pp. 179-211. HyunJeong, K., Schliesser, J., 2007. Adaptation of
Ardissono, L. et al. 2003. INTRIGUE: Personalized Storytelling to Mobile Information Service for a Site-
recommendation of tourist attractions for desktop and Specific Cultural and Historical Tour. Journal of
handset devices. Applied Artificial Intelligence:

122
 MOBILE TOURISM SERVICES - Experiences from Three Services on Trial

Information Technology & Tourism. vol. 9, no. 3-4 Technologies in Tourism 2003, Frew, A. J., Hitz, M. &
Cognizant, USA. O’Connor, P.(Editors.) Springer, Wien
Kaasinen, E., 2005. User acceptance of mobile services – Teo, TSH, Pok, SH., 2003. Adoption of WAP-Enabled
value, ease of use, trust and ease of adoption. Doctoral Mobile Phones among Internet Users. Omega, vol 31
Dissertation 566, VTT, Espoo, Finland pp. 483–498
Keen, P., Macintosh, R., 2001. The Freedom Economy – Tussyadiah, I., Fesenmaier, D., 2007. Interpreting tourist
Gaining the M-commerce Edge in the Era of the experiences from firstperson stories: a foundation for
Wireless Internet. McGraw-Hill, Berkeley. mobile guides. 15th European Conference on
Kim, D., Park, J., Morrison, A., 2005. A model of Tourist Information Systems (ECIS) 2007.
acceptance for mobile technology. Submitted to the Venkatesh, V., Morris, M. G., Davis, G. B., Davis, F. D.,
11th annual conference of Asia Pacific Tourism Assoc. 2003. User acceptance of information technology:
Kramer, R., Modsching, M., ten Hagen, K., 2005. A toward a unified view. MIS Quarterly, vol. 27, no. 3,
Location Aware Mobile Tourist Guide Selecting and pp. 425-478.
Interpreting Sights and Services by Context Matching. Von Hippel, E., 2005. Demoratizing Innovation, The MIT
Proceedings of the Second Annual International Press, London England
Conference on Mobile and Ubiquitous Systems:
Networking and Services.
Kristoffersen, S., Ljungberg, F., 2000. Mobility: From
Stationary to Mobile Work, Planet Internet, Chapter 6,
K. Braa, C. Sorensen, B. Dahlbom (eds).
Studentlitteratur, Lund, Sweden.
Lee, T., Jun, J., 2005. Contextual perceived usefulness?
Toward an understanding of mobile commerce
acceptance. Proceedings of the International
Conference on Mobile Business (ICMB).
Mallat, N., Rossi, M., Tuunainen, V., Öörni, A., 2006. The
impact of use situation and mobility on the acceptance
of mobile ticketing services. Proceedings of the 39th
Hawaii International Conference on System Science
(HICSS) 2006.
Moore, G., Benbasat, I, 1991. Development of an
Instrument to Measure the Perceptions of Adopting an
Information Technology Innovation. Information
Systems Research, vol 2, no 3, pp. 192-222.
Nysveen, H., Pedersen, P.E., Thorbjørnsen, H. 2005.
Explaining intention to use mobile chat services:
moderating effects of gender, Journal of Consumer
Marketing, vol. 22, no. 5, pp. 247 - 256
O'Donnell, J., Jackson, M., Shelly, M., Ligertwood, J.,
2007. Australian Case Studies in Mobile Commerce.
Journal of Theoretical and Applied Electronic
Commerce Research vol. 2, no. 2, August 2007 pp. 1 –
18.
Pedersen, P E., Munkvold, B E., Akselsen, S., Ytterstad,
P. 2006. The Business Model Concept - Relevance for
Mobile Tourism Services Innovation and Provision.
Fornebu, Telenor Research & Innovation. Report R&I
R 09/2006.
Pine, J., Gilmore J., 1999. The Experience Economy. Pine
and Gilmore Press.
Repo, P., Hyvönen, K., Saastamoinen, M., 2006.
Traveling from B2B to B2C: Piloting a Moblog
Service for Tourists. Proceedings of the International
Conference on Mobile Business (ICMB) 2006.
Rogers, E., M., 1995. Diffusion of Innovations. Free Press,
New York. 4th edition.
Schmidt-Belz, B., Zipf, A., Poslad, S. & Laamanen, H.,
2003. Location-based mobiletourist services – first
user experiences, in Information and Communication

123
 ADOPTION VERSUS USE DIFFUSION
Predicting User Acceptance of Mobile TV in Flanders

Tom Evens, Lieven De Marez

Research Group for Media & ICT (MICT-IBBT), Ghent University, Korte Meer 7-9-11, Gent, Belgium
[email protected], [email protected]

Dimitri Schuurman
Research Group for Media & ICT (MICT-IBBT), Ghent University, Korte Meer 7-9-11, Gent, Belgium
[email protected]

Keywords: User research, adoption diffusion, use diffusion, mutual shaping, mobile TV.

Abstract: In the contemporary changing ICT environment, an increasing number of services and devices are being
developed and brought to end-user market. Unfortunately, this environment is also characterized by an
increasing number of failing innovations; confronting scholars, policy makers as well as industry with an
explicit need for more accurate user research. Such research must result in more accurate predictions and
forecasts of an innovation’s potential, as a basis for more efficient business planning and strategy
implementation. However, the success of a new technology is not only depending on the adoption decision
and the number of people actually buying it, but relies at least as much on its actual usage. Hence, the focus
of truly user-oriented acceptance or potential prediction should focus on predicting both adoption diffusion
and use diffusion. Within this paper, we illustrate the added value of such an interactionist approach for the
study of future adoption and usage of mobile TV by the assessment of both a large-scale intention survey
and qualitative techniques such as diary studies, focus group interviews, observational and ethnographic
methods.

1 INTRODUCTION adoption, but at least as much on its usage. Hence,

the focus of truly user-oriented acceptance or
In the contemporary changing ICT environment, an potential prediction should not only be focussed on
increasing number of services and devices are being predicting adoption diffusion, but also on predicting
developed and brought to end-user market. use diffusion and potential usage. Evidently, the first
Unfortunately, this environment is also characterized research question to answer remains up to which
by an increasing number of failing innovations; degree the innovation has the potential to be
confronting scholars, policy makers as well as adopted. This should always be accompanied with
industry with an explicit need for more accurate user an answer to the question up to which degree the
research. Such research must result in more accurate innovation also has the potential to acquire a place in
predictions and forecasts of an innovation’s people’s and household’s daily lives (in terms of
potential, as a basis for more efficient business time and habits).
planning and strategy implementation. In terms of theoretical frameworks, the first
In most cases however, this need for more ‘adoption diffusion’ question relies on the diffusion
accurate user insight only gets translated in a cross- paradigm, while the second ‘use diffusion’ question
sectional investigation of the innovation’s adoption relies on the ‘social shaping’ and ‘domestication’
potential. However, the success of a new technology paradigm. Too often however, the Social Shaping of
or service is not only depending on the adoption Technologies (SST) and Domestication perspective
decision and the number of people actually buying is considered as the alternative to set off the lack of
it. For example, many people may have bought or attention for the user and his/her social usage
adopted a mobile phone with GPRS, UMTS or context in the diffusion theory. Traditionally, both
MMS without using the feature. The success of an perspectives (and the research based on them) have
innovation is thus not only depending on its too much been considered as opposites; while they

124
 ADOPTION VERSUS USE DIFFUSION - Predicting User Acceptance of Mobile TV in Flanders

are perfectly complementary to each other. The place within the microsocial context of that
purpose of this paper is to illustrate this household’ (Frissen, 2000: 67; Jankowski & Van
complementariness and the enrichment of combining Selm, 2001: 37). Domestication thus refers to
the more quantitative generalizing research approach integration of new technologies in the daily patterns,
of diffusionism with the more qualitative in-depth structures and values of users, relying on a more
SST research approach. Based on user research social determinism (Bouwman, Van Dijk, Van den
conducted on mobile TV, we illustrate how this Hooff & van den Wijngaert, 2002).
combination of approaches and methods resulted in Methodologically, the SST and domestication
a prediction of potential as well as usage of this new paradigm relies more on a qualitative tradition of
technology. This way, we intend to illustrate the methods such as in-depth interviews, ethnographic
observation and diary studies.
theoretical, methodological, managerial as well as
In the past, these two major paradigms have
policy relevance of this plea for a more mutual
mostly been regarded as opposite and competing,
shaping or interactionist approach on predicting user with convinced advocates from the two sides
acceptance (see Boczkowski, 2004: 255). engaging in vicious debates. However, with
diffusionism as the more quantitative tradition with
the focus on acceptance and adoption decisions and
2 TWO COMPLEMENTARY the domestication tradition as more qualitative with
a focus on the use and appropriation of technologies,
FRAMEWORKS both paradigms are clearly complementary (Punie,
2000). Or, as Boczkowski (2004: 255) states, ‘two
The oldest of the two theoretical frameworks is the sides of the same innovation coin’. To date a
‘diffusion framework’, of which Everett M. Rogers dialectical approach, which considers the
(1962) is assumed to be the founding father. development and diffusion of ICT innovations as
According to this framework, the diffusion of ‘joint processes of technological construction and
innovations in a social system always follows a bell- societal adoption’ (Boczkowski, 2004: 257), gains
shaped normal distribution, in which there can be ground. Instead of thinking in terms of diffusionism
successively distinguished between Innovators or social shaping, the mutual shaping or
(2.5%), Early Adopters (13.5%), Early Majority interactionism approach (Boczkowski, 2004;
(34%), Late Majority (34%) and Laggards (16%). A Lievrouw & Livingstone, 2006; Trott, 2003)
person’s innovativeness is assumed to be determined appeared in the late 90’s as a dynamic middle path
by the perception of the following set of innovation between the two previous linear deterministic
characteristics: relative advantage, complexity, predecessors. By integrating both quantitative and
compatibility, trialability and observability (Rogers, qualitative research outcomes within this paper, we
2003). Since the early 60’s the theory’s assumptions aim to illustrate the enrichment of such an
on segment sizes, diffusion pattern and determinants interactionist approach for the development and roll-
have been a basis for different types of (mostly) out of mobile TV in Flanders, the northern and
quantitative research such as econometric diffusion Dutch-speaking part of Belgium.
modelling or innovation scales (Goldsmith & Relying on the difference between ‘adoption
Hofacker, 1991; Meade & Islam, 2006; Moore & diffusion’ and ‘use diffusion’ (Shih & Venkatesh,
Benbasat, 1991; Parasuraman & Colby, 2001;
2004), we believe that the prediction of ‘adoption
Venkatsh, Morris, Davis & Davis, 2003).
Since the mid 80’s however, questions about its diffusion’ should rely on (1) a quantitative diffusion
technological determinism and lack of attention to approach by means of (intention) surveys and
the user and usage of the innovation have induced modelling to gain insight in the innovation’s
Rogers to adjust his approach to the adoption potential in terms of percentage of the target market,
decision process, but have also led to the rise of new penetration pattern and profiles of the different
paradigms such as domestication focussing on the adopter segments; and (2) the prediction of ‘use
‘way the use in households is being socially diffusion’, based on more qualitative techniques
negotiated and becomes meaningful, within the such as diary studies, focus group interviews,
social context of class, gender, culture or lifestyle’ observational and ethnographic techniques (if
(Van Den Broeck, Pierson, Pauwels, 2004: 103; possible in a field trial or living lab setting).
Haddon, 2007; Silverstone & Haddon, 1996) or ‘the
process of taming and house training ‘wild’
technological objects, by adapting them to the
routines and rituals of the household and thus giving
them a more or less natural and taken-for-granted

125
ICE-B 2008 - International Conference on e-Business

3 RESEARCH DESIGN
The empirical findings are based on the two-year
MADUF project which studied the possibilities of
mobile TV using DVB-H in Flanders. In first
instance, a large-scale user survey (n: 575) was set
up in order to forecast the market potential, or to
predict the ‘adoption diffusion’ potential for mobile
TV in Flanders. By applying the Product Specific Figure 1: Interactionist research design.
Adoption Potential (PSAP) scale, we were able to
map the size and nature of the future mobile TV
market in Flanders. The PSAP scale is an intention
based survey method in which respondents are
4 RESULTS: PREDICTING
allocated to Innovator, Early Adopter, Majority and ADOPTION DIFFUSION
Laggard segments based on the stated intentions on
a general intention question and on respondent- By applying the PSAP scale to 575 rich cases, we
specific formulated questions gauging for their obtained a reliable view on the size and nature of the
intention for ‘optimal’ and ‘suboptimal’ product various adoption segments for mobile TV in
offerings (De Marez & Verleye, 2004; Verleye & De Flanders in the following segmentation forecast.
Marez, 2005). The scale was compared on its While traditional fixed segment sized methods are
reliability with five other adoption models and has reflected by the black line (in this case Rogers’
been applied to and validated for a diversity of ICT Diffusions of Innovations), the red line represents
innovations such as digital TV, 3G, mobile TV and the adoption potential for mobile TV. The latter is
mobile news (De Marez, 2006; De Marez, Vyncke, contrasted to the potential of 3G (De Marez, 2006),
Berte, Schuurman & De Moor, 2008). which allows TV programmes to be received over a
In second instance, a representative panel of test unicast architecture network. Figure 2 clearly shows
users was randomly selected from the 575 survey that there is little demand for mobile TV over DVB-
respondents to experiment with mobile television H compared to Rogers’ full market approach and
devices in a ‘living lab’ setting during two weeks. even compared to the take-up of 3G services. Due to
Due to practical reasons (the DVB-H network was the lack of substantial innovative segments
operational in the city of Ghent only, so the panel (Innovators and Early Adopters), we would
contained people exclusively living but not recommend a partial market approach or even a
especially working in Ghent) and because of the niche strategy for the introduction of mobile TV in
rather explorative nature of this field trial, the Flanders. This implies a specific introduction
amount of test users was limited to 30. With this strategy for a limited market potential to serve the
field trial, we aimed at achieving a first realistic chosen segments in an optimal manner (about a 20%
view of how future users will integrate mobile TV in market penetration). Since the Late Majority and
their everyday practices. Users were asked to Laggard segment are clearly not willing to pay for
document their experiences in diaries while logging this mobile service, we will define the maximal
their activities, noting their comments and taking target group as Innovators, Early Adopters and Early
pictures of their usage situations. Majority promising a 16,7% segmentation forecast.
Next to these data, we also gained insight in their
personal evaluation of the trial phase by means of a
post-measurement. Comparing these results with the
findings of the market forecast before testing the
device allowed us to see whether user attitudes
towards mobile television had changed as a result of
the trial. In this manner, we aimed to measure the
effect of trialability, the degree to which an
innovation may be experimented with on a limited
basis (Rogers, 2003: 266). Explanations for possible
shifts between the pre- and the post-measurements Figure 2: Segmentation forecast mobile TV.
can be found in the usage diaries and two organised
focus groups. Figure 1 illustrates this interactionist In general, we witnessed a rather dual profile
approach combining both quantitative user attitude within the innovative segments with on the one hand
research and qualitative ethnographic techniques. well-earning, older executives (little time, potential

126
 ADOPTION VERSUS USE DIFFUSION - Predicting User Acceptance of Mobile TV in Flanders

for snacking) and on the other hand low educated test environment felt obliged to experiment more
young couples without children (much time, with the devices than they would do within a more
complementary to heavy TV viewing behaviour). natural context. Although we cannot ignore this trial
Although especially executives are facing a shortage effect, it plays a less important role within this
of time, most of them seem to be heavy television research set-up because we aim to generate
viewers, watching both entertainment and explorative rather than statistically representative
information programs. Especially Innovators and findings. In terms of this usage frequency pattern,
Early Adopters (joint for statistical reasons) possess we can distinguish three kinds of viewers: light
advanced mobile phones (with camera, MMS, WAP, viewers watching less than 10 times (n: 15), medium
MP3, FM radio…), which they use in an innovative viewers watching between 10 and 20 times (n: 13)
manner (e.g. sending e-mails on mobile phone, see
and heavy viewers watching more than 20 times (n:
Figure 3). Generally, these people show the highest
2). These two heavy viewers were identified as
willingness to pay for mobile TV while most of
them consider a mobile TV device (with integrated Innovator and Early Adopter within our large-scale
mobile phone) as a substitute for their current sample.
mobile phone. Within our user panel, we only found two heavy
viewers while the rest of the panel was about equally
divided among medium and light viewers. One
important finding during our test period is that the
different types of viewers used the mobile TV
device in a different way. Figure 4 represents all
watching moments and divides them amongst the
periods people watched mobile TV. In terms of the
Figure 3: Sending e-mails on mobile phone moments people watched mobile TV, we identified
six different time slots: night (0-6h), morning (6-
Clearly, such quantitative research may provide 12h), noon (12-14h), afternoon (14-18h), evening
reliable estimations of the adoption potential and (18-22h) and late evening (22-24h). When analysing
diffusion (in this case of mobile TV in Flanders), but the figure, we see that, except for the light viewers,
does not provide us with in-depth information trial participants are not inclined to watch mobile
regarding the domestication and potential use TV while having breakfast. This is probably due to
diffusion of mobile TV. What place will it take in the strong position in the morning of the medium
the lives of the consumers, how and when will it be radio, which is ‘together with the water and the
used? stove, the first thing that is turned on in the morning’
(Winocur, 2005: 325). Light viewers are also more
likely to watch mobile television at noon while
5 RESULTS: PREDICTING USE having dinner.
DIFFUSION
To answer the latter questions, one needs a more
qualitative ‘use diffusion’ and domestication
oriented research framework. In the case of mobile
television a combination of diaries, focus group
discussions, pre-post test comparisons and photo
elicitation within the boundaries of a living lab
setting was used to get further insight in people’s
usage patterns of mobile TV. Although we are aware
these results are not statistically representative due
to the very limited sample of 30 test users, they Figure 4: Usage patterns (per time slot).
nevertheless allow us to identify some explorative
usage patterns for mobile TV amongst our field trial Heavy viewers are most likely to watch mobile
participants. during the afternoon, while most of the other types
On average, people watched approximately of viewers only switch their device on in the evening
eleven times via their mobile television device after coming home from work or school (see Figure
during the two-week test period. However, it is 4). While light and medium viewers are watching
possible that people being part of a panel within a mobile TV in the evening, we notice a remarkable

127
ICE-B 2008 - International Conference on e-Business

decline in viewing of the heavy viewer-segment Especially heavy viewers made use of the
during this time slot (see red line). Nevertheless, we complementary function of mobile TV and
see that this segment starts watching again in the late considered it as an extra supply next to their regular
evening, the moment where the other segments television. This is illustrated by the fact that heavy
switch their device off. This results in peaked viewers watched significantly more in public space
watching patterns that differ quite much between the and on the move. We found that watching in the car
three user segments. While light and medium users is a rather popular activity to kill time, sometimes as
show one viewing peak during the evening, heavy fellow passenger but also as driver. These people
viewers have two peaks: one in the afternoon and driving to their work and back, spend a lot of time in
one in the late evening. The latter two-peaked their car and have to suffer traffic jams. It is hardly
surprising that in such cases mobile television is
pattern is rather complementary with traditional TV,
seen as a simple time killer although the radio can
as its peak time comes right in between the mobile
serve this purpose as well. Other persons preferred
peak times. We can conclude that heavy viewers watching mobile TV while waiting for or travelling
used mobile TV complementary to their regular with public transport services (bus, metro and train).
television and therefore watched the device in a Taking into account the massive success of the iPod,
manner it was meant to be watched: on the move. In mobile TV devices can be the next big thing to
contrast, light and medium viewers watched mobile spend time while commuting.
TV at home as a substitute for regular television. After the trial period, we asked our 30 test users
The previous findings are supported by the usage to fill out the same questionnaire they had
locations indicated in the diaries. Light and medium previously taken. Based on the combined results of
viewers especially watched mobile TV at home. both pre- and post-trial measurement, we were able
Undoubtedly, the most popular place was the living to compare the findings and see whether user
room where people are used to watch regular expectations and attitudes had changed during the
television while relaxing in their sofa. This also mobile TV field trial. The findings from the
seemed the case for mobile TV: most people qualitative part of this research project (i.e. focus
watched television in their natural habitat. Instead of groups and ethnographic methods such as usage
watching the large screen, our test users watched diaries) enabled us to explain possible shifts.
mobile TV, albeit for a rather short period. After General interest for mobile TV slightly increased
having tested the mobile device, they switched to the during the field trial. However, persons who
large screen again to enjoy their favourite programs. originally intended to purchase a mobile TV device
Here, we witnessed a substitution of the classical soon, now preferred to wait a bit longer. On the
screen at traditional peak times with mobile TV was other hand, the amount of people certainly not
willing to purchase a mobile TV device declined as
considered a second TV (see also Schuurman et al.,
well. A slightly increased average score (from 3,70
2008). This was especially the case for the light and
to 3,80) suggests that overall attitude towards mobile
medium users in our sample. This does explain the TV became a little bit more positive. Also the
similarities between peak times for mobile and average price people are willing to pay increased
regular TV for these groups. from €233 to €294. But it is striking that we witness
Another popular location for watching mobile a converging shift towards a non-decisive average.
TV was the kitchen. People seem to enjoy watching Convinced believers start to doubt while disbelievers
mobile TV while eating in the kitchen, where most might have seen some possibilities after all due to
of the time no TV set is at hand. We also witnessed the trial.
that a lot of people used the mobile device while In other words, less people are showing an
working at their desk or sitting behind the computer. innovative attitude towards mobile TV, but many
These people used mobile TV rather as a others shifted from ‘never’ to ‘maybe’. It thus seems
background medium or as tertiary activity (see that the field trial has raised awareness of mobile TV
Jacobs, Lievens, Vangenck, Vanhengel & Pierson, and that a lot of people do not consider the medium
2008). When they heard something interesting, they as a luxurious product any longer, making it less
switched attention from their work to the mobile appealing to the more innovative but more likely to
device. Although they watched mobile television, consider for the less innovative. Although these
these people considered the mobile television device people are not likely to purchase mobile TV soon,
often as a radio, which is in most cases also used as they are not longer against mobile TV since they
a background medium. Here mobile TV was clearly have experienced it as a handy medium to catch up
used in combination with other activities such as television content quickly. Innovators and early
doing the dishes or working (multitasking). adopters on the other hand were somehow
disappointed by the lack of interactive and

128
 ADOPTION VERSUS USE DIFFUSION - Predicting User Acceptance of Mobile TV in Flanders

interesting content, resulting in their downgrade. theoretical as well as industry and policy relevance
Despite the shift towards a more positive attitude, for the prediction of ICT user acceptance. For
the potential for mobile television remains instance, in the current debate of digital dividend
dramatically low, as the sample does not contain any such predictions could help policymakers to get
Innovators or Early Adopters anymore and that the insight in the feasibility of new communication
least innovative segments (Late Majority and technologies and for which new technologies they
Laggards) remain largely overrepresented. should preserve space in the future radio spectrum.
These estimations also allow marketing managers to
decide in which market segments they should invest
6 CONCLUSIONS and with what offer these segments should be
targeted. Finally, for researchers we hope this paper
With this paper, we intended to reconcile two gives some food for thought about the added value
opposing traditions: adoption diffusion and use of an interactionist approach and inspires them to
diffusion. Within the MADUF-project, we combined work out more creative innovations research designs
research techniques from both traditions in an in the future.
interactionist way in order to get a more holistic
view on the possible success of mobile TV in
Flanders. By means of a PSAP-estimation, it became ACKNOWLEDGEMENTS
clear that mobile TV is not ready yet for total market
acceptance so that a partial market or even niche The MADUF-project (Maximize DVB Usage in
strategy was suggested. By means of a diary study, Flanders) was supported by grants from the research
combined with a pre-test and post-test survey during centre IBBT (Interdisciplinary Institute for
a mobile TV-trial in a living lab environment, we Broadband Technology) and a consortium of both
were able to get a better understanding of the broadcasters and network solution companies. The
possible use diffusion of mobile TV. We found that MADUF project aimed to maximize the social and
for most test persons traditional television remains economic valorisation of DVB-H for the Flemish
the reference point for evaluating mobile TV. citizen, government and industry (broadcasters,
Television undoubtedly is one of the most operators and constructors) through the development
domesticated technologies within the home and of a technological and regulatory consensus model
became so dominant that people often schedule their (pax mobilis).
behaviour in function of the TV-set. We found that
light and medium mobile viewers used the device at
home as a second TV with watching behaviour in
line with traditional TV. Heavy users on the contrary REFERENCES
watched mobile TV in a truly mobile and much
more complementary way with traditional television. Boczkowski, P.J. (2004). The Mutual Shaping of
Technology and Society in Videotex Newspapers:
This resulted in mobile peak times coinciding with
Beyond the Diffusion and Social Shaping
regular TV for the former two groups, while for the Perspectives. The Information Society, 20(4): 255-267.
latter mobile TV allowed to extend the regular TV Bouwman, H., Van Dijk, J., Van den Hooff, B. & van de
viewing peak with two mobile peaks: one before and Wijngaert, L. (2002). ICT in organisaties. Adoptie,
after the regular peak. Finally, we witnessed the implementatie, gebruik en effecten. Amsterdam:
(modest) overall positive effect of trialability Boom.
through a slightly increased general attitude towards De Marez, L. (2006). Diffusie van ICT-innovaties:
mobile TV during the field trial. accurater gebruikersinzicht voor betere introductie-
By combining these two paradigms, we were strategieën. Gent: Universiteit Gent.
De Marez, L., Verleye G, (2004). Innovation Diffusion:
able to draw a clearer picture of the potential success
the need for more accurate consumer insight.
of mobile TV and the different factors influencing Illustration of the PSAP-scale as a segmentation
this success. While a quantitative potential instrument. The Journal of Targeting, Measurement
estimation can identify adoption segments and and Analysis for Marketing, 13(1): 32-49.
describe them for targeting purposes, the qualitative De Marez, L., Vyncke, P., Berte, K.; Schuurman, D. & De
usage diffusion-research provides input for the fine- Moor, K. (2008). Adopter segments, adoption
tuning of the technology in terms of usage patterns, determinants and mobile marketing. Journal of
features and content. We believe this methodological Targeting, Measurement & Analysis for Marketing,
plea for more interactionist research designs has 16(1): 78-95.

129
ICE-B 2008 - International Conference on e-Business

Frissen V. (2000). ICTs in the rush hour of life. Conference on Interactive Television. ‘Enhancing the
Information Society, 16(1): 65-75. experience’, 31/03/2004-02/04/2004, University of
Goldsmith, R.E. & Hofacker, C. (1991). Measuring Brighton, UK: 101-112.
consumer innovativeness. Journal of the Academy of Venkatesh, V.; Morris, M.G., Davis, G.B. & Davis, F.D.
Marketing Science 19(3): 209-222. (2003). User acceptance of information technologies:
Jacobs, A.; Lievens, B.; Vangenck, M.; Vanhengel, E; & toward a unified view. MIS Quarterly, 27(3): 425-478.
Pierson, J. (2008). Mobile television or television on Verleye G., De Marez, L. (2005). Diffusion of
mobile? In: A. Urban; B. Sapio & T. Turk (Eds.) innovations: successful adoption needs more effective
Digital Television Revisited. Linking Users, Markets soft-DSS driven targeting. The Journal of Targeting,
and Policies, 68-75. Measurement and Analysis for Marketing, 13(2): 140-
Jankowski, N. & Van Selm, M. (2001). ICT en 155.
samenleving. Vier terreinen voor empirisch
onderzoek. In: Bouwman, H. (Ed.), Communicatie in
de informatiesamenleving (p. 217-249). Utrecht:
Lemma.
Lievrouw, L. & Livingstone, S. (Eds) (2002). The
Handbook of New Media. London, Thousand Oaks,
New Delhi: Sage Publications.
Meade, N. & Islam, T. (2006). Modelling and forecasting
the diffusion of innovation. A 25-year review.
International Journal of Forecasting, 22: 519-545.
Moore, G.C. Benbasat, I. (1991). Development of an
instrument to measure the perceptions of adopting an
information technology innovation. Information
Systems Research 2(3): 192-222.
Parasuraman, A. & Colby, C.L. (2001). Techno-reading
marketing: how and why your customers adopt
technology. New York: Free Press.
Punie, Y. (2000). Domesticatie van informatie- en
communicatietechnologie. Adoptie, gebruik en
betekenis van media in het dagelijkse leven: continue
beperking of discontinue bevrijding?. Vrije
Universiteit Brussel: Vakgroep
Communicatiewetenschappen.
Rogers, E.M. (1962). The diffusion of innovations. New
York: The Free Press.
Rogers, E.M. (2003), Diffusion of innovations (5th ed.),
New York: The Free Press.
Schuurman, D.; De Marez, L. & Evens, T. (2008). Mobile
TV: content is king, but who’s to be crowned?
Proceedings of International Technology, Education
and Development Conference. March 3-5, Valencia,
Spain.
Shih, C.-F. & Venkatesh, A. (2004). Beyond Adoption:
Development and Application of a Use-Diffusion
Model. Journal of Marketing, 68(January): 59-72.
Silverstone,R. & Haddon, L. (1996). Design and
domestication of information and communication
technologies: technical change and everyday life. In:
Silverstone, R. & Mansell, R (Eds.), Communication
by design. The politics of information and
communication technologies (p. 44-74). Oxford:
Oxford University Press.
Trott, P. (2003). Innovation and Market Research. In:
Shavina, L.V. (Ed.), The International Handbook on
Innovation (p. 835-844). Oxford, UK: Pergamon,
Elsevier.
Van Den Broeck, W.; Pierson, J. & Pauwels, C. (2004).
Does interactive television imply new uses? A Flemish
case study, Conference Proceedings, Second European

130
 ADOPTION OF MOBILE SERVICES IN FINLAND
Conceptual Model and Application-based Case Study

Mari Ervasti and Heli Helaakoski

VTT Technical Research Centre of Finland, Kaitoväylä 1, P.O. Box 1100, FI-90571 Oulu, Finland
[email protected], [email protected]

Keywords: Adoption, Mobile service, TAM (Technology Acceptance Model), TPB (Theory of Planned Behaviour),
J2ME, Finland.

Abstract: This paper offers both theoretical and empirical contributions relating to the consumers’ motives for and
barriers to adopting mobile services. A case study of a mobile service is introduced to validate the factors
that facilitate and trigger, and respectively hinder the acceptance of mobile services in Finland. The factors
affecting the service adoption were researched based on a conceptual model that was a modified version of
the Technology Acceptance Model (TAM) and of the Theory of Planned Behaviour (TPB). Based on the
findings from a survey conducted with a sample of 52 Finnish campus students, the factors that affect
consumers’ attitudes towards mobile services were discovered. Usefulness was revealed to have the
strongest impact on attitude towards a mobile service, with context information and attitude acting as
significant determinants of the intention to use the service. This study attempts to help mobile service
providers choose right development and marketing strategies for mobile services and enhance the adoption
and diffusion of mobile services.

1 INTRODUCTION perspectives of physical, cognitive, security and

economic barriers, and from the perspectives of the
perceived entertainment value and perceived
For a long time Finland has been a European pioneer flexibility benefits of mobile services. Finnish
for testing new mobile services. But even in Finland, consumers’ usage of mobile services has been
the adoption of mobile services has been modest and studied in a report for National Consumer Research
the most popular services are still based on SMS Centre by Hyvönen and Repo (2005). The report
messaging. Current mobile services do not fulfil the focuses on mobile services that Finnish people use,
requirements set by the users since the services have the motives for and obstacles to using the services,
been developed based on technological development and the readiness to pay for mobile services.
instead of users’ needs. According to Moore and The success of mobile services is not only
McKenna (1999) the mobile services are at the depending on technological maturity of mobile
chasm, the pioneering and early adopters are using technology and the penetration of mobile handsets;
the services, but the masses have not yet followed in the end, eventually the users decide on the success
their lead. Many researches (Hyvönen and Repo, of the services. The predictions are hard to make, but
2005; Bouwman et al., 2007) have studied the there are certain factors that need to be considered
factors that affect the adoption of mobile services in when developing mobile services. Muller-Veerse
the Finnish context but there is still a need to (2000) has identified ubiquity, reachability, security,
evaluate the adoption factors in real-life case studies. convenience and personalisation as the key drivers
Kaasinen (2005) has studied the user acceptance of for the mobile market today. Rogers (1995) has
mobile services by using the Technology developed ‘Rogers’ basic five’ factors, later
Acceptance Model (Davis, 1989) and built user increased with two more factors (Moore and
acceptance on four factors: perceived value of the Benbasat, 1991) that influence the adoption and
service, perceived ease of use, trust, and perceived diffusion of new technologies throughout a social
ease of adoption. Bouwman et al. (2007) have system: relative advantage, compatibility,
complexity, trialability, observability, image and
researched mobile service adoption from the

131
ICE-B 2008 - International Conference on e-Business

trust. According to Rogers mobile services are likely consumer survey conducted in Finland (Anckar,
to be adopted if mobile solution has some clear 2002) indicated that m-commerce adoption mainly
advantages over existing products or services. appears to be driven by a need for solutions that add
Mobile services should also be compatible with convenience and flexibility to daily routines rather
existing practices, values, and skills of potential than excitement and entertainment. Anckar (2002)
adopters, and easy to understand and use. If it is not also found out that consumers perceive the ability to
possible to experiment with the service before satisfy spontaneous and timecritical needs as the
making an adoption or rejection decision, and if the most important driver of m-commerce adoption.
benefits offered by the service are not visible, then This paper researches and validates factors that
adopters are likely to perceive uncertainty and risk facilitate and trigger, and respectively hinder, the
in adopting the service. Today the mobile phone is adoption of mobile services in Finland. Finland has
more like a reflection of one’s identity, so if the been selected as a case country since it is considered
adoption and use of mobile service is perceived to as a pioneer of mobile technologies. In August 2007,
communicate one’s personality and enhance the 97% of Finns had at least one mobile phone
image and status, it has more potential to be adopted. (Statistics Finland, 2007). The findings from the
It is also important that the mobile service adopter survey conducted in autumn 2007 (Invest in Finland,
perceives the service provider to be trustworthy. 2008), based on information on four million mobile
There are several factors inhibiting the use of phones from three telecom operators, showed that
services, although the technology maturity is no Finns change their mobile phones every 2.7 years.
longer an obstacle and even so the adoption and More than 70% of the phones in use had a colour
diffusion of mobile services has not yet proceeded as screen and half included a camera. The 3G phones
fast and wide as have been assumed. Development gained an 18% share of the market, which was more
of technology is commonly seen as an important than double the figure in 2006. In the study in which
prerequisite, or even as a reason, for people to adopt participated over 1500 Finns aged 15-64 in year
new mobile services. According to Gillick and 2006, one quarter of those surveyed reported having
Vanderhoof (2000), the new mobility experience, browsed web pages with their mobile phones (It-
i.e. the anytime and anyplace access to desired viikko, 2007).
products and services, will be the greatest benefit for Furthermore the paper studies what kinds of
the consumer. However, Figge (2004) claims that services are likely to be adopted; and how
the ubiquity alone is not enough to ensure a user businesses can enhance the adoption and diffusion of
acceptance and the success of a mobile application: mobile services. In order to validate the adoption
as the user’s problems vary with the context in factors in practice this paper presents a real life case
which he or she accesses service, the service based study of launching the mobile service. The case
on information of the specific context is likely to mobile service was based on downloadable Java
become the user’s favourite. Gilbert and Han (2005) application, not yet eagerly adopted at least in
identified one key barrier of attracting a critical mass Finland. It was tested in a delimited user group of
of adopters as the lack of compelling content, and students and personnel in Finnish campus area in the
another as the differences among adoption patterns field trial. Mobile service offered information about
in individual market segments. Aarnio et al. (2002) internal news, events, announcements and schedules
proposed two major reasons above all inhibiting the of the study community, as well as a mobile flea
use of mobile services; the high pricing and the market for buying and selling own things through
restricted quality of services. mobile channel. The aim of provided services was to
Hyvönen and Repo (2005) have stated that enable easy access for essential information despite
mobile services are not adopted linearly according to the time and place. Especially students require
the assumptions of diffusion theory. Their studies multi-channel services due to their mobile lifestyle.
state that the socio-demographic factors do not This paper is organized as follows: chapter 2
essentially affect the adoption of mobile services; represents the conceptual model of mobile service
instead, the best way to explain the use of mobile adoption and chapter 3 describes the empirical
services is to research the general attitudes towards evaluation of case mobile service. In chapter 4 the
new technology, mobile services and devices. In a feedback questionnaire of the case study is presented
survey by Cheong and Park (2005) conducted in and the results of the survey are analysed. The paper
Korea, perceived playfulness was found to be concludes with the limitations of the survey and the
influential in predicting the behavioural intention to suggestions for future work.
use mobile Internet. Whereas findings from a

132
 ADOPTION OF MOBILE SERVICES IN FINLAND - Conceptual Model and Application-based Case Study

2 CONCEPTUAL MODEL OF antecedents of attitude (ATT) toward mobile

MOBILE SERVICE ADOPTION services. Barriers to use are also conceptualised with
context (CON) as an antecedent of intention (INT)
to use mobile services. Perceived behavioural
In our study, we developed a modified version of the control (PBC), reference group influence (REF) and
Technology Acceptance Model (TAM) (Davis, attitude are antecedents of intention, which, in turn,
1989) and the Theory of Planned Behaviour (TPB) affects actual use (USE). (Fishbein and Ajzen, 1975;
(Ajzen, 1991) for researching the adoption of mobile Davis, 1989; Ajzen, 1991; Moore and Benbasat,
services. According to TAM, a consumer’s 1991; Thompson et al. 1991; Davis et al., 1992;
behavioural intention to use a “system” is Taylor and Todd, 1995; Venkatesh and Davis,
determined primarily by its usefulness and ease of 2000).
use. In TPB, the intention to use a “system” is Based on the conceptual model in Figure 1, the
explained by attitudes toward a certain behaviour, following hypotheses for mobile services were
subjective norm and perceived behavioural control. created:
TAM and TPB have both been widely used among H1: Perceived usefulness of mobile services
researchers and found to be very useful in explaining positively affects attitude toward mobile services.
consumers’ attitudes and intentions toward a given H2: Perceived ease of use of mobile services
behaviour. TPB is a general theory of human positively affects attitude.
behaviour while TAM is specific to information H3a: Perceived barriers to use mobile services
systems. negatively affect attitude.
Studies on acceptance of new technology H3b: Perceived barriers to use mobile services
indicate that traditional adoption models need to be negatively affect intention to use mobile services.
extended and modified to better explain the adoption H4: Utilisation of context-specific information
of the innovations. This study modifies TAM by on mobile services positively affects intention.
proposing additional acceptance factors associated H5: Perceived behavioural control of mobile
with mobile services, Rogers (1995) and Figge services positively affects intention.
(2004) having especially influence on the creation of H6: Reference group influence positively affects
construct items stated in Figure 1. Attitudinal intention.
construct was adopted from TPB (e.g. Karjaluoto H7: Positive attitude toward mobile services
and Alatalo, 2007). Mobile service adoption model positively affects intention to use mobile services.
was decided to be based on utilitarian-oriented H8: Intention to use mobile services is positively
aspects. Thus, the pure entertainment factors, such associated with actual use of mobile services.
as perceived enjoyment and playfulness were not In order to further evaluate and develop the
included as construct items. Figure 1 shows the conceptual model with related hypotheses, the use
created conceptual model for case mobile service and adoption were tested in a real-life case study of
with causal links between the construct items mobile service.
(Ervasti, 2007).

3 THE EVALUATION OF CASE

MOBILE SERVICE

The practical evaluation of the mobile service

adoption was made in the case study, named Mora.
The case mobile service focused on additional value
of mobility, which means the possibility to access,
produce and send information, products and services
at anytime and from anywhere, regardless of the
user’s location or the time of day. The major aim of
Figure 1: Conceptual model for mobile services.
the mobile service was to provide essential
information for the end-users as the administration
Perceived usefulness (PU), perceived ease of use
of the campus area defined. Mora service was
(PEOU) and barriers to use (BU) are posited as

133
ICE-B 2008 - International Conference on e-Business

evaluated in a delimited user group of students and database is used as the basic information source for
personnel in Finnish campus area. Mora was user-related information (name, email address, study
launched at the campus, focused on technology and group), besides the basic personal data the database
economics education, in Finland in September 2007, could be contain more detailed information on
in order to research the adoption of mobile services customer’s special preferences. The mobile end of
among campus’ students and personnel. the concept utilised mobile Java technologies. The
The technical implementation of the case mobile concept approach of using mobile application has
service is based on a client-server approach to several benefits both for the end user and the service
providing a mobile access to intranet of campus provider. Java application provides versatile content
area. The contents of Mora were internal news, delivery and interaction channel, a service provider
events, announcements and schedules of the study can create a fully branded mobile service concept
community, as well as a mobile flea market. Mobile and the application is unobtrusive and convenient to
flea market enabled users to sell and buy things such use, avoiding the spam effect. When compared to
as studying material easily and quickly. In later browsing, the application-based approach offers
development the mobile service could also provide better possibilities for improving usability in mobile
pleasure, fun, and enjoyment. The mobile client services. In addition, the optimisation of data traffic
Mora works as an interaction channel between between the terminal and servers is easier to
customer and service provider by providing implement in mobile applications. The small size of
personalised information to the user (Ervasti, 2007). the displays also favours application-type services.
The server side application carries most of the The case mobile service started with an
functionality. The client side of Mora is Java mobile information and marketing campaign advertising
application (J2ME) that works in most common Mora web pages and giving presentations about the
mobile phones. field trial to various student groups and personnel in
The main technical building blocks of a the campus area. The purpose was to make the test-
personalised mobile service concept are the service users feel they were privileged to have the mobile
provider’s existing information systems, mobile service like Mora exclusively on their campus, and
service, customer database and mobile application. have the opportunity to participate in the service
The main building blocks (adapted from Alahuhta et testing and development. These initial Mora users
al., 2005) of the Mora service concept are introduced were encouraged to compete for the best
in Figure 2: development ideas. In the adoption of application-
based mobile services, the key barrier to overcome is
to get the users to do the first download (Nokia,
2003). Therefore, Mora instructions presented
simple steps for starting the use of Mora as follows:
1) Registration via web page; 2) Downloading the
Mora mobile application via SMS or web page; 3)
Setting up the required connection settings; 4)
Setting up the mobile client preferences.
The test-users were able to post questions for us
during the whole field trial and separate events for
getting help in downloading and installing the
application were arranged to further ease the
adoption process. Users had also an opportunity to
test the Mora service with separate test phones in
case a user’s own mobile phone didn’t support the
Figure 2: Overview of the Mora mobile service. application. The nature of the marketing campaign
was selected due to the time and costs limitations. In
The major source of information for most of the future research the diffusion of mobile services by
data and communication processed in the Mora viral marketing should also be considered and
service is the existing information system, the researched.
campus’ web intranet. This required adaptation of
the content from web-based content to mobile
content. The user database is utilised for providing
personalised services for customers. The existing

134
 ADOPTION OF MOBILE SERVICES IN FINLAND - Conceptual Model and Application-based Case Study

4 FEEDBACK QUESTIONNAIRE Hungary. Most of the survey respondents had a

AND ANALYSIS OF THE MORA monthly income of less than 1,000 euros and a
relatively new mobile phone in use, the handset’s
MOBILE SERVICE age being under two years for the majority. The
sample’s monthly mobile phone expenditure was
The feedback from the Mora mobile service was mainly under 30 euros, and most used their phone
collected after two months’ field trial through online more for private than business purposes.
and paper questionnaires. The feedback The survey respondents’ attitudinal valuations of
the Mora mobile service are presented in Table 1
questionnaire was created based on the hypotheses
with mean values and standard deviations. In
in conceptual model (Figure 1). All of the
perceived usefulness, the respondents mostly agreed
questionnaire’s construct items consist of multiple
with the statement “Using Mora gives me topical
variables, and a five-point Likert scale ranging from information”. In perceived ease of use, most agreed
1 (strongly disagree) to 5 (strongly agree), with 3 with the statement “I quickly learned to use Mora”,
(cannot determine) as a neutral anchor point was and they were also aware of the kind of mobile
used to measure the variables. In Table 1 is services they can use with their mobile phones. Most
presented the feedback questionnaire for the case users agreed with the statement that Mora
study (adapted from Venkatesh et al., 2003). downloading and installation was easy. Users
During the field trial, the Mora mobile service disagreed with the statements concerning perceived
gathered 67 registered users and the survey resulted barriers to use. Fear of technical features, required
in 52 completed feedback forms. The results were effort in downloading and installing the application,
treated and analysed anonymously (Ervasti, 2007). or even the data transmission billing weren’t seen as
The analysis of the research data was done by using obstacles in adopting mobile services. The utilisation
SPSS 14.0 software. A small number of of time and date-related contextual information in
questionnaire forms were incompletely filled in and Mora services was seen as more useful than the
the answers were ambiguous in some places. The utilisation of location-specific information. The
ambiguity was resolved by interpreting the answer in respondents weren’t especially eager to spend time
the least favourable way for the question presented. providing their personal details to make the content
The rather small amount of sample data restricted in Mora more relevant to their needs.
the number of feasible analysis methods. The When examining the perceived behavioural
analysis was conducted using the basic frequency control, the respondents found it important that they
and descriptive calculations. The regression analysis could control the use of Mora by deciding when and
was utilised by first examining the effect of the where to use the service. The reference group
independent variables on the dependent variables influence was perceived as rather irrelevant in this
according to the set hypotheses, and then testing the survey. The attitudinal construct showed that the
combined effect of the explanatory variables as respondents had a more preferable attitude toward
stated in the conceptual model. downloadable applications than toward SMS-based
Of the respondents, 11.5% were female and mobile services. Most agreed with the statement
88.5% male and the majority were less than 25 years “Using the mobile phone as a channel to access the
old. By nationality, 69.2% of the sample were content of school intra is a good idea”. The
Finnish and 30.8% foreign; the foreign respondents’ respondents reported that they were willing to use
nationalities mostly representing Nigeria, Russia and

Table 1: Feedback questionnaire items and attitudinal dimensions of test-user responses to Mora mobile service.
Construct item Variable Mean s.d.
Perceived usefulness
PU1 Using Mora gives me topical information. 4.29 0.610
PU2 Using Mora saves me time / helps me accomplish things more
3.77 1.022
quickly.
PU3 Using Mora gives me personal information. 3.39 1.150
PU4 Using Mora gives me information that is of interest to me. 3.81 0.817
PU5 I find Mora useful in my daily life 3.90 1.053
Perceived ease of use
PEOU1 I know what kind of mobile services I can use with my phone. 4.14 1.167

135
ICE-B 2008 - International Conference on e-Business

Table 1: Feedback questionnaire items and attitudinal dimensions of test-user responses to Mora mobile service (cont.).
PEOU2 It was easy for me to download and install Mora in my mobile
3.88 1.211
phone.
PEOU3 I quickly learned to use Mora. 4.62 0.718
PEOU4 I find Mora easy to use. 4.33 0.964
PEOU5 My interaction with Mora is clear and understandable. 4.00 1.085
PEOU6 I find it easy to get Mora to do what I want to do. 3.92 1.100
Barriers to use
BU1 Fear of technical features is an obstacle to me adopting mobile
1.63 1.085
services.
BU2 Difficulty and effort caused by the downloading and installation of
the application in a mobile phone is an obstacle to me adopting 1.76 1.205
mobile services.
BU3 Knowledge of data transmission billing is an obstacle to me
2.51 1.528
adopting mobile services.
Context
CON1 I would view the content of Mora related to a specific time or date
4,39 0.750
(e.g. exam day, weekend) as useful.
CON2 I would view the content of Mora related to me being in a specific
4.10 0.953
location (e.g. campus restaurant, library) as useful.
CON3 I would be prepared to spend time providing my personal details (a
user profile) to make the content of Mora more relevant to my 3.78 1.112
needs.
Perceived behavioural control
PBC1 I find it important that I can choose between different styles in
3.17 1.354
Mora.
PBC2 I find it important that I can easily control the use of Mora (that I
4.50 0.828
can decide when and where I use Mora).
PBC3 I find it important that I can easily stop using Mora. 4.44 0.978
PBC4 It would be important for me to be in control in terms of the ability
4.19 0.793
to filter the content of Mora related to my interests/needs.
Reference group influence
REF1 I started to use Mora because it was recommended to me by
2.71 1.576
someone I know.
REF2 I started to use Mora because my friends are using it. 2.12 1.199
REF3 I would have been more likely to start to use Mora if the message
containing a link for downloading the application was forwarded to 2.83 1.424
me by someone I know.
REF4 I recommended Mora to people I know. 3.49 1.271
Attitude
ATT1 Using mobile services (downloadable applications) is pleasant. 3.90 1.159
ATT2 Using Mora is pleasant. 4.08 0.710
ATT3 Using mobile services (SMS-based) is pleasant. 3.38 1.223
ATT4 Using mobile technology to access a variety of services is
4.31 0.707
interesting.
ATT5 Using the mobile phone as a channel to access the content of the
4.73 0.528
university intranet is a good idea.
Intention
INT1 I feel positively about Mora. 4.54 0.503
INT2 I am willing to use Mora in the future, if possible. 4.48 0.641
INT3 I would use Mora regularly in the future. 4.02 1.038
Actual use
USE1 I am using Mora every day. 2.58 1.226
USE2 I am using Mora every week. 3.71 1.226
USE3 I regularly check what’s new with Mora. 3.40 1.302

136
 ADOPTION OF MOBILE SERVICES IN FINLAND - Conceptual Model and Application-based Case Study

Mora in the future. However, the actual use of Mora those services. The relationship was found to be
wasn’t that regular since it happened more on a positive and statistically significant. The final
weekly than daily basis. hypothesis, H8, argues that INT is directly
71.2% of respondents reported having used other associated with USE, and the coefficient is strong
mobile services in addition to Mora and 67.3% of and positive. Thus both H7 and H8 were supported.
them had used downloadable applications, whereas Figure 3 illustrates the standardised coefficients
only 38.5% SMS-based services. 57.7% of for the conceptual model.
respondents reported having used entertainment
services, and 55.8% had used both the information
and news services and email; 38.5% had used instant
messaging, and 23% banking and financial services.
17.3% of respondents had experience of VoIP,
11.5% of ticket reservation, 7.7% of shopping, and
5.8% of travel booking.
Table 2 shows the standardised coefficients and
t-values of the tested hypotheses regarding mobile
services (t-statistics indicating the significance of
individual variables when the t-value is > 2
(Schwager, 1995).
Figure 3: Conceptual model results.
Table 2: Hypotheses examination.
Hypothesis Standardized t- Sig. During the field trial it was noticed that mobility
Coefficient β value does not provide enough added value for the users
H1: PU to ATT .419 3.164 .003 when the Mora service is in its current form.
H2: PEOU to ATT .303 2.182 .034 Consumers were not attracted to the service because
H3a: BU to ATT -.173 -1.205 .234 they already had an easy and familiar way to access
H3b: BU to INT -.050 -0.345 .731 almost the same content via the web, so they felt that
H4: CON to INT .539 4.430 .000 the Mora is just a mobile substitute for the campus’
H5: PBC to INT .250 1.828 .073 web intranet. Even though the challenges with
H6: REF to INT .237 1.707 .094 marketing the Mora service and establishing test-
H7: ATT to INT .572 4.882 .000 users’ interest were recognised and actions were
H8: INT to USE .381 2.913 .005 taken to overcome them, one reason for the
The strong coefficient of H1 shows that there is a somewhat modest use of the Mora service was that
positive and direct relationship between PU and the benefits of the service weren’t enough visible to
ATT. Thus, the perceived usefulness of mobile potential adopters. For Mora service users, the
services is a strong predictor of attitude and H1 was easiness and flexibility of the service should have
supported. It can also be seen that the PEOU of been stressed even more, i.e. that service can be used
mobile services was directly and positively at any time and in any place, freeing the users from
associated with ATT, providing support for H2. H3a the traditional time and place constraints even more
and H3b address the role of BU, the coefficients efficiently than Internet services. Another significant
were negative, but very low, which indicates that obstacle was the effort required in downloading and
barriers to use have no significant influence on installing the application. To diminish this problem
either attitude or intention, thus H3a and H3b were several separate events were arranged where
not supported. Next, the strong coefficient of H4 guidance and support in service adoption was
indicates that context-related information acts as a offered to users.
strong and positive predictor of intention to use
mobile services, so H4 was supported. H5 and H6
hypothesise that PBC and REF have a positive and 5 CONCLUSIONS
direct relationship with INT. Both coefficients are
similar with low magnitudes, therefore neither H5
nor H6 were supported by the data. H7 states that the This paper introduces a conceptual model for mobile
more favourable attitude a person forms toward service adoption that is a modified version of
mobile services, the higher his or her intention to use established adoption models. The developed
conceptual model was evaluated and validated in the

137
ICE-B 2008 - International Conference on e-Business

real-life case study of mobile service, through which further validate the conceptual model. In an
we have gained practical insights into application- international comparison, Finland is an advanced
based mobile service development and promotion, as information society especially when it comes to
well as having identified the factors affecting the mobile communication. Thus, the study findings
adoption of mobile services in Finland. might vary in other countries with far lower mobile
The findings from the survey suggest that phone penetration rates and different mobile
perceived usefulness of mobile services is the communication culture. However, when compared
strongest predictor of attitude toward mobile to Japan, Finland is far behind in the versatility of
services. That is also supported by previous studies mobile communication: in Japan, a nation of 127
(e.g. Rogers, 1995) on consumer acceptance of new million the number of mobile internet subscribers
technologies. People will adopt a technology when recently passed 100 million (The Guardian, 2007).
the added value offered by technology fulfils their The conceptual model should also be extended
needs. Hence, compatibility with users’ needs (ibid) further to include more factors that reflect the unique
is a critical variable for predicting technology characteristics of the usage intentions of mobile
adoption. Ease of use wasn’t perceived as having a services. Further attention should be devoted to the
strong relationship with acceptance of mobile comparison of rival models in explaining consumer
services, contrary to (ibid), so it can be reasoned that attitudes and intentions.
usefulness is a stronger determinant of attitude
toward mobile services than perceived ease of use.
Utilisation of contextual information was found to REFERENCES
have a strong impact on intention to use mobile
services and proves that consumers desire relevant Aarnio, A., Enkenberg, A., Heikkilä, J., Hirvola, S., 2002.
and personalised services. Adoption and Use of Mobile Services: Empirical
The respondents did not perceive the technical Evidence from a Finnish Survey. In Proceedings of the
features of mobile services, the downloading and 35th Hawaii International Conference on System
installation process or data transmission billing as Sciences, 3(35), pp. 1454-1463. IEEE Computer
significant barriers to adopting mobile services as Society.
prior researches suggest (Aarnio et al., 2002; Nokia, Ajzen, I., 1991. The theory of planned behaviour.
2003). When examining the impact of control and Organisational Behavior and Human Decision
Processes, 50(2), pp. 179-211. Elsevier.
social norms, the results of the survey were not Alahuhta, P., Helaakoski, H., Smirnov, A., 2005.
congruent with the theoretical background (Ajzen, Adoption of Mobile Services in mCommerce –
1991; Rogers, 1995) either, since perceived Technologies and Success Factors. In IEEE ICEBE
behavioural control and reference group influence 2005, pp. 531-534.
weren’t stated as important determinants of service Anckar, B., 2002. Motivators for Adoption of Mobile
adoption. A preferable attitude toward mobile Commerce: Findings from a National Consumer
services was however proved to have a strong Survey. In Proceedings of the International
impact on intention to adopt services, whereas the Conference on Decision Making and Decision Support
intention in the end didn’t act as a good predictor of in the Internet Age (DSIage 2002), pp. 749-762.
Bouwman, H., Carlsson, C., Molina-Castillo, F.J.,
actual use. Walden, P., 2007. Barriers and drivers in the adoption
Based on the case study findings, it is suggested of current and future mobile services in Finland.
that the extended and modified conceptual model Telematics and Informatics, 24(2), pp. 145–160.
can be used to guide industry players’ evaluation of Elsevier.
the adoption potential of new mobile services. Case Cheong, J.H., Park, M., 2005. Mobile internet acceptance
study results can be used to help mobile service in Korea. Internet Research, 15(2), pp. 125-140.
providers choose right development and marketing Emerald Group Publishing Limited.
strategies for mobile services and accelerate the Davis, F.D., 1989. Perceived Usefulness, Perceived Ease
diffusion of services. However, further and more of Use, and User Acceptance of Information
Technology. MIS Quarterly, 13(3), pp. 319-340.
extensive studies should be conducted in order to Management Information Systems Research Center,
attain a more profound understanding of the issues University of Minnesota.
concerning consumers’ drivers for adopting (or Davis, F.D., Bagozzi, R.P., Warshaw, P.R., 1992.
rejecting) mobile services. Similar research need to Extrinsic and Intrinsic Motivation to Use Computers
be done in other countries as well in order to obtain in the Workplace. Journal of Applied Social
benchmarks for comparing the differences in Psychology, 22(14), pp. 1111-1132. Blackwell
adoption characteristics in different countries, and to Publishing.

138
 ADOPTION OF MOBILE SERVICES IN FINLAND - Conceptual Model and Application-based Case Study

Ervasti, M.A., 2007. The Development of Mobile Service telecommunications in Finland seen through statistics.
Concepts – Technologies, Adoption and Use, Accessed 03.06.2008. URL:
University of Oulu, Department of Electrical and http://www.stat.fi/tup/suomi90/syyskuu_en.html
Information Engineering. Master’s Thesis, Finland. Taylor, S., Todd, P.A., 1995. Assessing IT Usage: The
Figge, S., 2004. Situation-dependent services – a Role of Prior Experience. MIS Quarterly, 19(2), pp.
challenge for mobile network operators. Journal of 561-570. Management Information Systems Research
Business Research, 57(12), pp. 1416-1422. Elsevier. Center, University of Minnesota.
Fishbein, M., Ajzen, I., 1975. Belief, Attitude, Intention Thompson, R.L., Higgins, C.A., Howell, J.M., 1991.
and Behavior: An Introduction to Theory and Personal Computing: Toward a Conceptual Model of
Research, Addison-Wesley. Reading, MA. Utilization. MIS Quarterly, 15(1), pp. 124-143.
Gilbert, A.L., Han, H., 2005. Understanding mobile data Management Information Systems Research Center,
services adoption: Demography, attitudes or needs? University of Minnesota.
Technological Forecasting & Social Change, 72(3), Venkatesh, V., Davis, F.D., 2000. A Theoretical Extension
pp. 327-337. Elsevier. of the Technology Acceptance Model: Four
Gillick, K., Vanderhoof, R., 2000. Mobile E-commerce: Longitudinal Field Studies. Management Science,
Marketplace Enablers and Inhibitors. In Smart Card 46(2), pp. 186-204. INFORMS.
Forum Annual Meeting, 23 p. Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.,
The Guardian, 2007. Why mobile Japan leads the world. 2003. User Acceptance of Information Technology:
Accessed 03.06.2008. URL: Toward a Unified View. MIS Quarterly, 27(3), pp.
http://www.guardian.co.uk/technology/2007/sep/27/gu 425-478. Management Information Systems Research
ardianweeklytechnologysection.mobilephones Center, University of Minnesota
Hyvönen, K., Repo, P., 2005. Mobiilipalvelut
suomalaisten arjessa (Mobile services in the every day
life of Finns). In Vox consumptoris – Kuluttajan ääni,
pp. 41-59. National Consumer Research Centre,
Helsinki. (In Finnish)
Invest in Finland, 2008. Top 56 mobile phones in Finland
were all made by Nokia. Accessed 03.06.2008. URL:
http://www.investinfinland.fi/news/2008/en_GB/nokia
/
It-viikko, 2007. Every fourth Finn surfs on their mobile
phone (In Finnish). Accessed 03.06.2008. URL:
http://www.itviikko.fi/talous/2007/01/23/neljannes-
suomalaisista-surffaa-kannykalla/20071827/7
Kaasinen, E., 2005. User acceptance of mobile services –
value, ease of use, trust and ease of adoption, VTT
Publications 2005. Espoo.
Karjaluoto, H., Alatalo, T., 2007. Consumers’ attitudes
towards and intention to participate in mobile
marketing. International Journal of Services
Technology and Management, 8(2/3), pp. 155-173.
Inderscience.
Moore, G.A., McKenna, R., 1999. Crossing the Chasm:
Marketing and Selling High-Tech Products to
Mainstream Customers, Harper Business. New York.
Moore, G.C., Benbasat, I., 1991. Development of an
Instrument to Measure the Perceptions of Adopting an
Information Technology Innovation. Information
Systems Research, 2(3), pp. 192-222. INFORMS.
Muller-Veerse, F., 2000. Mobile Commerce Report,
Durlacher Research Ltd. London.
Nokia, 2003. Java™ technology enables exciting
downloading services for mobile users. In White
paper. Nokia Corporation, Finland.
Rogers, E.M., 1995. Diffusion of innovations, Free Press.
New York, 4th edition.
Schwager, J.D., 1995. Fundamental Analysis, John Wiley
and Sons. United States.
Statistics Finland, 2007. From hand-cranked connections
to hands-free mobile calls – history of

139
SHORT PAPERS
 THE ROLE OF AUDIO-VISUAL METAPHORS IN AIDING THE
COMMUNICATION OF CUSTOMER KNOWLEDGE
User Satisfaction Prespective

Dimitrios I. Rigas and Mutlaq B. Alotaibi

Department of computing, University of Bradford, Bradford, West Yorkshire, BD7 1DP, U.K.
[email protected], [email protected]

Keywords: Customer, Knowledge, Multimodal, Interaction, Satisfaction, Sound, Speech, Earcons, Auditory Icons,
Audio-visual.

Abstract: The purpose of this paper was to examine the implications of employing multimodal user interaction in
Electronic Customer Knowledge Management Systems (E-CKMS). There are various challenges to E-
CKMS mentioned in current literature and reported (trust and knowledge hoarding, structured transfer of
knowledge and content and relevancy issues). As empirical CKM studies that examine the role of
multimodal interaction in this field are generally lacking, this paper reports research conducted to evaluate
this role and measure user satisfaction. This evaluation was undertaken empirically by developing two E-
CKMS experimental platforms (text with graphics only and multimodal). The major findings indicated that
user satisfaction was significantly improved by using multimodal metaphors. Further investigation is needed
to determine the usefulness of two multimodal versions on E-CKMS instead of comparing text with
graphics only with multimodal.

1 INTRODUCTION form multiple communication channels (Bueren,

2005). This led to integrating both Knowledge
The new means of IT, such as internet, extranets and Management (KM) and Customer Relationship
intranets, leads to dramatic shift from information to Management (CRM) in E-Business contexts. More
knowledge societies (Thierauf, 1999, Goh, 2005). details on the determination and characteristics of
Harnessing invisible assets is becoming one of the KM, CRM and E-Business can be found in (Becerra-
primary sources of creating value and competitive Fernandez et al., 2004, Tiwana, 2001, Skyrme, 2001,
advantage in the current age of knowledge (Goh, Alotaibi and Alzahrani, 2004).
2005). Knowledge as a concept covers vast area of E-CKMS is derived from the integration between
various views (Becerra-Fernandez et al., 2004), KM and CRM in E-Business (Tiwana, 2001), and its
levels (Davenport and Prusak, 1998, Rowley, 2002), typical example is represented in Amazon case study
principles (Such et al., 2001), taxonomies (Gebert et al., 2002b, Gurgul et al., 2002, Rollins
(Davenport and Prusak, 1998, Thierauf, 1999, Hahn and Halinen, 2005), which illustrates how
and Subramani, 2000), strategies and trends. One Communities of Customers (CoC) works. There are
classification of knowledge is based on the source several similarities between CoC and Communities
from which knowledge has been elicited (external of Practice (CoP), which mean that CoC is rooted in
and internal) (Davenport and Prusak, 1998). the traditional KM (Lesser et al., 2000, Gurgul et al.,
External knowledge (CK) can be regarded as one of 2002). In CoC, peer customers share opinions and
the most organisations valuable types of knowledge insights about products and services, which can lead
(Osterle, 1995), which is gathered at the customer to more effective and guided decisions made by any
point of contact under a great deal of time pressure member of this community (Gibbert et al., 2002).
(Lesser et al., 2000). The Utilisation of CK helps From CRM point of view, storing and analysing
organisations to grow, innovate and compete against customer historical transactions leads to
competitors (Gebert et al., 2002a, Gebert et al., understanding customer buying patterns, and hence
2002b), but it is not easy to gather, identify, interpret leveraging up-selling and cross-selling opportunities
and integrate, because it flows to the organisation (Tiwana, 2001, Pan and Lee, 2003). Amazon case
study involves KM and CRM aspects that can be

143
ICE-B 2008 - International Conference on e-Business

benchmarked by efforts devoted to evaluating E- Section 3 described the experimental platform.

CKMS. Interactive systems can be seen as one of the Design of the empirical study is shown in Section 4.
CRM components that enable E-CKM by In Section 5, we presented results and discussion.
establishing learning relationship (learn while Conclusion is provided in Section 6. Finally, we
interact) (Massey et al., 2001). Interacting with described future work in Section 7.
customers in real-time and adopting CoC can be
counted among several proposed approaches that
improve the elicitation (Gibbert et al., 2002), expand 2 RELEVANT WORK
the exploitation (Lesser et al., 2000) of CK, reduce
cost, and hence replace the traditional approach that
Researchers in the CKM field tend to analyse CKM
rely on understanding of sales representatives or case studies, identify cultural, structural and
results market research. managerial mechanisms that facilitate CKM success
There were several challenges to E-CKMS that
and propose models and frameworks for theory and
can be tackled using the interactive multimodal practice. García-Murillo and Annabi (2002)
metaphors. Knowledge hoarding (Davenport and proposed a model of social relationships, and put
Prusak, 1998, Gibbert et al., 2002), for example, can
emphasis on interpreting knowledge directly by
be seen as the lack of customer willingness to share human (García-Murillo and Annabi, 2002). Some
knowledge, and solved by several approaches that other authors consider structural aspects, such as
include optimising customer-company dialogue
organisational boundaries, culture, structure and
(Interaction) (Massey et al., 2001, Gurgul et al., climate (Bose and Sugumaran, 2003, Dous et al.,
2002, Gibbert et al., 2002). Gibbert et al. (2002) 2005, Bueren, 2005). Others look at rewards
stated that trust and knowledge hoarding could be
systems, incentives, management support and
tackled by not only establishing continuous two-way perception (Gibbert et al., 2002, Gurgul et al., 2002).
dialog with customers, but also employing
In addition, there were several authors who
interactive multimedia systems. Another challenge is
discussed CK characteristics, approaches,
that customers when interacting with E-CKMS applications and relations (Feng and Tian, 2005,
needs to have their ideas well-structured and
Lesser et al., 2000, Rowley, 2002, Skyrme, 2001).
organised, which can be aided by what so-called Business Engineering (BE) (Osterle, 1995)
Customer Innovation Toolkits (CIT) (von Hippel, perspective suggests separating strategy, process,
2001b, Von Hippel, 2001a). CIT can be incorporated
system and change levels (Bose and Sugumaran,
into E-CKMS in order to facilitate an optimal 2003, Dous et al., 2005, Bueren, 2005). Several
transformation of customer expertise and studies presented styles of CKM and types of CK
expectations (knowledge sharing) into valuable
(Gibbert et al., 2002, Gurgul et al., 2002, Feng and
suggestions, which can be afterward used to offer Tian, 2005, Rowley, 2002). Although, these studies
customised and personalised products and services. commonly studied CK based on several perspectives
This context involves knowledge sharing, and hence
and points of view, other than empirically examining
encouraging customer to share knowledge is needed, the role of multimodal interaction, it provided
which leads to the assumption that there is a insights into the underlying principles and
potential role for multimodal interaction metaphors
theoretical foundations of E-CKMS.
to play. Another challenge is information overload There is little known about efforts in CKM field
and relevancy, which is, actually derived the web- has been devoted to evaluate the potential role that
based environment. Authors (Davenport and Prusak,
multimodal metaphors can play in E-CKMS.
1998, Bueren, 2005) in E-CKMS field raises Nevertheless, a great deal of studies in several fields
information overload and relevancy as a concern,
of study have been conducted to evaluate such role,
and include it with content issues, but some mention
and found that user interface can be improved by the
identification of relevant knowledge (Urban and von augment of speech (Kehoe and Pitt, 2006)
Hippel, 1988). Brewster (1997) argued that this
(synthesised and recorded speech) and non-speech
could be addressed by enhancing the text with sounds (earcons (Rigas et al., 2000, Rigas and Alty,
graphics only manner of information display with 2005) and auditory icons (Gaver, 1997, Cohen,
means of auditory metaphors (Brewster, 1997). To
1993). In software engineering, There were several
sum up, multimodal interaction is anticipated to studies that support this view, such as (Sonnenwald
address trust, structure and content questions in E- et al., 1990, Cohen and Ludwig, 1991, DiGiano et
CKMS.
al., 1993, Rigas et al., 1997, Rigas and Alty, 1998).
The remainder of the paper is organised in seven Overall, user satisfaction and other usability
sections. In Section 2, we introduced relevant work.

144
 THE ROLE OF AUDIO-VISUAL METAPHORS IN AIDING THE COMMUNICATION OF CUSTOMER
KNOWLEDGE - User Satisfaction Prespective

attributes can be improved in general Information 3 EXPERIMENTAL PLATFORM

Systems (IS) by employing multimodal interaction.
The experimental platform developed for this
Table 1: E-CKMS visual and auditory metaphors. research provided typical functions of web-based
VCKMS MCKMS mobile phones retailing systems, and included an
CK category T G T G S E A R additional function labelled as co-production, which
Trends (top 10) √ √ √ defined by (Gibbert et al., 2002) as the manner in
Customer which customers practice New Product
√ √ √ √ Development (NPD). Typically, E-CKMS consists
review
Customer rating √ √ √ of three main components: CKM, infrastructural and
Website advice √ √ √ √ user interface components, and can incorporate any
Co-production √ √ √ √ √ √ additional function, such as the one included in this
Product study (co-production). This study assumed that
√ √ √ √ √ √ √ CKM and infrastructural components were
features
previously implemented.
Burke et al. (2006) carried out meta-analysis Co-production function facilitates the dual role
investigation into the effectiveness of multimodal of customers (producer and consumer) by offering
interaction, in forty three studies, and found that CIT that aid customers to manipulate elements in
audio-visual metaphors have a significant role to solution space to test new products (billing scheme).
play in improving user performance in IS compared In fact, solution space included several elements, but
to the visual-only display. In addition, this finding the scope of this study limited these elements to
was supported by the results obtained from two loyalty, billing and taffies schemes. Furthermore, co-
experiments carried out by (Rigas and Memery, production function offered a trail-and-error engine
2002), who investigated the use of auditory stimuli that enabled experimental NPD, allowed its
(speech, earcons and auditory icons) to communicate repetition until final product design was reached.
information to users in both email and stock control This engine received customised schemes from CIT,
applications. Additionally, two studies (Rigas and sent it to billing engine, received customised bill,
Alty, 2005, Alty and Rigas, 2005) investigated stored it in trails comparison array and provided
utilising rising pitch metaphors in the comparison of results obtained from other trails in
communication of graphical information, and found order to support customer decision making.
that it was possible for visually-impaired users to This platform was implemented with two
interpret graphical information with the aid of rising interfaces. These two versions were text with
pitch metaphors, even in the absence of a visual graphics only E-CKMS (VCKMS) and Multimodal
display. E-CKMS (MCKMS). In VCKMS, CK was
Since E-CKMS is a web-based environment, it communicated to E-CKMS users using text with
can be linked to similar fields of study, such as web- graphics only, while the communication method was
based browsing and email applications (Rigas and audio-visual in MCKMS. The communication of CK
Memery, 2002, Rigas and Memery, 2003, Rigas, required classification of CK and auditory and visual
2003). In web-based browsing systems, a prototype metaphors, and utilisation of a wide range of
has been develop as an online help system with technologies. First, Types of CK were organised into
sound support (Kehoe and Pitt, 2006), and extended six categories (trends, customer reviews, customer
later by incorporating no-speech sound and other ratings, website advices, co-production CK and
auditory metaphors (Kehoe et al., 2007). In addition, product features). Some of these CK were
another web-based browsing prototype was built to communicated visually, auditory or simultaneously.
browse musical notes with the help of sound, and Second, the visual metaphors employed were text
proven successful performance improvement (T) and Graphic (G), whilst the auditory metaphors
(Fernström and McNamara, 2005). In email were synthesised speech (S), earcons (E), auditory
applications, several experiments (Rigas and icons (A) and recoded speech (R). Table 1 shows
Memery, 2002, Rigas and Memery, 2003, Rigas, each CK category and the way by which it was
2003) were conducted to evaluate the potential of communicated.
audio-visual metaphors in reducing visual
complexity and tackling information hiding, and
concluded that this hypothesis was true, besides that
visual display have to be synchronised with means
of auditory stimuli.

145
ICE-B 2008 - International Conference on e-Business

4 DESIGN OF THE EMPIRICAL ease of the system (EOS), extent of user confusion
(EOC), extent of user frustration (EOF), ease of
STUDY navigation (EON) and overall comfort (COM).
Measuring user attitude towards the system appears
The research undertook explored customer to be difficult. However, asking to specify the
satisfaction aspects related to the use of audio-visual extent, to which the user agree or disagree with a set
metaphors in three levels of task complexity. of statements, tend to support the pursuit of this
measurement (Jordan, 1998).
Table 2: Summary of task levels influential factors.
Table 3: The mode and frequency values of the five
CKMA Complexity factors aspects of customer satisfaction.
Task

PSA TSA PRA CKI CI NR NS Aspects of customer satisfaction

Group Value
E A1 A2 L L 10 40 EOS EOC EOF EON COM
Mode 4 3 3 5 5
M A3 A4 A5 M M 15 17 VCKMS
Freq. 50% 40% 45% 85% 55%
D A6 A7 A8 H H 17 4 Mode 5 2 2 5 5
MCKMS
Freq. 70% 55% 50% 65% 65%
The three levels are task easy (Task E), task
moderate (Task M) and task difficult (Task D). In User agreement and disagreement utilised a six-
this study, task levels were created based on six point scale ranging from agree strongly to disagree
influential factors: number of task requirements strongly (Salkind, 2006a). The values of the scale
(NOTR), number of available selections (NOAS), were six for strong agreement, five for moderate
number of CKM activities (CKMA), customer agreement, four for slight agreement, three for slight
Interaction (CI) and CK intensity (CKI). If the task disagreement, two for moderate disagreement and
was to be designed as difficult, NOTR, CKMA, CKI one for strong disagreement. After completing all
and CI needed to be increased, while NOAS was to user satisfaction questions, responses were summed
be decreased. In CKMA factor, there have been a up to generate an overall score for user satisfaction,
various types of CKMA: phone selection activity based on the system usability scale (SUS) (Brooke,
(PSA), tariff selection activity (TSA) and Co- 1996).
production activity (CPR). In CKI and CI factors, It was noteworthy that multimodal interaction
there were three important levels: low (L), moderate reduced customer response time, which led to
(M) and high (H). Table 2 reviews the task greater customer satisfaction and eventually loyalty.
complexity influential factors, and illustrates the Participants in the experimental group expressed
association between tasks and CKMA. More interest in CK communicated aurally more than
information on task levels, types and workload is those in the control one. Our experience with this
provided on (Burke et al., 2006). platform suggested that users tend to be more
Forty subjects (all were students at University of comfortable with aural communication when sounds
Bradford, and regular internet users) were selected conveyed more rapidly than for the first time. At
randomly, based on the non-probability sampling first glance, the mean value of customer satisfaction
strategy (convenience-sampling method) (Salkind, for using MCKMS (77%) was higher than that for
2006b). Subjects were divided into two groups (20 VCKMS (63%). Significance of the difference
each): control and experimental, and then offered a between the two conditions was tested (at 0.05
short training session on the corresponding version significance level) using the t-test. The difference
of E-CKMS. Subsequently, subjects were asked to was found significant (t38 = 4, CV= 2.03 P < 0.05).
perform the three tasks and then fill a questionnaire Table 3 shows the mode values for the aspects of
devised for this study. The order of tasks was user satisfaction with the values for using VCKMS
counterbalanced between participants in order to and MCKMS, in addition to the frequency of the
neutralize possible task learning effect. mode. User responses suggested that the multimodal
E-CKMS is easier to use, less confusing and less
frustrating. In fact, 70% of users agreed moderately
5 DISCUSSION OF RESULTS that the multimodal system was easy to use
compared to 50% agreed slightly that the text with
Satisfaction was measured by a set of user provided graphics only version was easy to use. In user
answers to questionnaire questions, which include confusion and frustration, half of the sample agreed

146
 THE ROLE OF AUDIO-VISUAL METAPHORS IN AIDING THE COMMUNICATION OF CUSTOMER
KNOWLEDGE - User Satisfaction Prespective

moderately that the multimodal E-CKMS is neither diagrams: the role of context. International Journal of
confusing nor frustrating, in comparison to 40% and Human-Computer Studies, 62, 21-40.
45% of users disagree slightly that the text with Becerra-Fernandez, I., Gonzalez, A. J. & SABHERWAL,
graphics only version was confusing and frustrating R. (2004) Knowledge management : challenges,
solutions, and technologies, Upper Saddle River, N.J.,
respectively. Furthermore, Mann-Whitney statistical Pearson/Prentice Hall.
test was performed, and showed that there were Bose, R. & Sugumaran, V. (2003) Application of
significant differences in all aspect of customer knowledge management technology in customer
satisfaction expect in the ease of navigation. relationship management. Knowledge and Process
Management, 10, 3-17.
Brewster, S. A. (1997) Using non-speech sound to
overcome information overload. Displays, 17, 179-
6 CONCLUSIONS 189.
Brooke, J. (1996) SUS: a quick and dirty usability scale.
The role E-CKM has become increasingly important Usability Evaluation in Industry, 189-194.
to public and private organisations due to the cost Bueren, A. (2005) Improving performance of customer-
savings it offers. However, it is considerably processes with knowledge management Adrian
complex, vague and challenging discipline due to Bueren, Ragnar Schierholz, Lutz M. Kolbe, Walter
the many aspects involved. This study shed light into Brenner The Authors. Business Process Management
Journal, 11, 573-588.
three challenges to E-CKMS (trust and knowledge
Burke, J. L., Prewett, M. S., Gray, A. A., Yang, L.,
hoarding, structured knowledge transformation and Stilson, F. R. B., Coovert, M. D., Elliot, L. R. &
content and relevancy issues), which can be tackled Redden, E. (2006) Comparing the effects of visual-
by utilising multimodal metaphors. This hypothesis auditory and visual-tactile feedback on user
was tested empirically by two independent groups in performance: a meta-analysis. Proceedings of the 8th
two E-CKMS. Subjects took part in this research international conference on Multimodal interfaces,
were satisfied and expressed interest in the use of 108-117.
audio-visual metaphors. The significance of this Cohen, M. (1993) Throwing, pitching and catching sound:
approach was found in all aspects of customer audio windowing models and modes. International
Journal of Man-Machine Studies, 39, 269-304.
satisfaction, except in the ease of navigation.
Cohen, M. & Ludwig, L. F. (1991) Multidimensional
Therefore, it can be concluded that audio-visual audio window management. International Journal of
metaphors has contributed positively towards the Man-Machine Studies, 34, 319-336.
improvement of customer satisfaction. In summary, Davenport, T. & Prusak, L. (1998) Working knowledge :
results showed that the use of multimodal interactive how organizations manage what they know, Boston,
metaphors in E-CKMS is more satisfactory than to Mass., Harvard Business School Press.
text with graphics display. Digiano, C. J., Baecker, R. M. & Owen, R. N. (1993)
LogoMedia: a sound-enhanced programming
environment for monitoring program behavior.
Proceedings of the INTERACT'93 and CHI'93
7 FURTHER WORK conference on Human factors in computing systems,
301-302.
This experiment reported here proved to be Dous, M., Salomann, H., Kolbe, L. & Brenner, W. (2005)
successful. However, it was noticed during the Knowledge Management Capabilities in CRM:
previous experiments that users tend not to take into Making Knowledge For, From and About Customers
Work. the Eleventh Americas Conference on
account various types of presented CK, such as
Information Systems. Omaha, NE, USA.
customer ratings and reviews. Therefore, further Feng, T. X. & Tian, J. X. (2005) Customer Knowledge
investigation into additional multimodal metaphors Management and Condition Analysis of Successful
is needed to determine how speech with aviator can CKM Implementation. Machine Learning and
promotes further utilisation of customer reviews, and Cybernetics Proceedings of 2005 International
this is expected to yield a set of useful guidelines. Conference on., 4, 2239- 2244
Fernström, M. & Mcnamara, C. (2005) After direct
manipulation---direct sonification. ACM Transactions
on Applied Perception (TAP), 2, 495-499.
REFERENCES García-Murillo, M. & Annabi, H. (2002) Customer
knowledge management. Journal of the Operational
Alotaibi, M. & Alzahrani, R. (2004) Evaluating e- Research Society, 53, 875-884.
Business Adoption: Opportunities and Threats. Journal Gaver, W. W. (1997) Auditory Interfaces. Handbook of
of King Saud University, 17, 29-41. Human-Computer Interaction, 1, 1003-1041.
Alty, J. L. & Rigas, D. (2005) Exploring the use of Gebert, H., Geib, M., Kolbe, L. & Riempp, G. (2002a)
structured musical stimuli to communicate simple Towards Customer Knowledge Management.

147
ICE-B 2008 - International Conference on e-Business

Proceedings of the 2nd International Conference on EUROMICRO-97, New Frontiers of Information

Electronic Business (ICEB 2002). Technology, 78–84.
Gebert, H., Geib, M., Kolbe, L. M. & Riempp, G. (2002b) Rigas, D. I. & Memery, D. (2002) Utilising audio-visual
Towards Customer Knowledge Management– stimuli in interactive information systems: a two
Integrating Customer Relationship Management and domain investigation on auditory metaphors.
Knowledge Management concepts. The Second Information Technology: Coding and Computing,
International Conference on Electronic Business 2002. Proceedings. International Conference on, 190-
(ICEB 2002), 296-298. 195.
Gibbert, M., Leibold, M. & Probst, G. (2002) Five Styles Rigas, D. I. & Memery, D. (2003) Multimedia e-mail data
of Customer Knowledge Management, and How browsing: the synergistic use of various forms of
Smart Companies Use Them To Create Value. auditory stimuli. Information Technology: Coding and
European Management Journal, 20, 459-469. Computing [Computers and Communications], 2003.
Goh, A. L. S. (2005) Harnessing knowledge for Proceedings. ITCC 2003. International Conference on,
innovation: an integrated management framework 582-586.
Andrew LS Goh The Authors. Journal of Knowledge Rigas, D. I., Memery, D., Hopwood, D. & Rodrigues, M.
Management, 9, 6-18. A. (2000) Using non-speech sound to communicate
Gurgul, G., Rumyantseva, M. & Enkel, E. (2002) information in user interfaces. Applied Informatics
Customer Integration–Establish a constant bilateral 2000, 357–362.
knowledge flow. Diskussionspapier des Research Rollins, M. & Halinen, A. (2005) Customer Knowledge
Centers Knowledge Source der Universität St. Gallen. Management Competence: Towards a Theoretical
Hahn, J. & Subramani, M. R. (2000) A framework of Framework. System Sciences, 2005. HICSS'05.
knowledge management systems: issues and Proceedings of the 38th Annual Hawaii International
challenges for theory and practice. Proceedings of the Conference on, 240a-240a.
twenty first international conference on Information Rowley, J. E. (2002) Reflections on customer knowledge
systems. Brisbane, Queensland, Australia, Association management in e-business. Qualitative Market
for Information Systems. Research: An International Journal, 5, 268-280.
Jordan, P. W. (1998) Measures of usability. An Salkind, N. J. (2006a) Methods of measuring behavior.
introduction to usability. First ed. London, Taylor & Exploring research, Sixth edition, 123-144.
Francis Ltd. Salkind, n. j. (2006b) sampling and generalizability.
Kehoe, A., Neff, F. & Pitt, I. (2007) Extending traditional Exploring research, Sixth edition, 85-98.
user assistance systems to support an auditory Skyrme, D. J. (2001) Capitalizing on knowledge : from e-
interface. Proceedings of the 25th conference on business to k-business, Oxford, Butterworth-
Proceedings of the 25th IASTED International Multi- Heinemann.
Conference: artificial intelligence and applications Sonnenwald, D. H., Gopinath, B., Haberman, G. O., Keese
table of contents, 637-642. III, W. M. & Myers, J. S. (1990) InfoSound: an audio
Kehoe, A. & Pitt, I. (2006) Designing help topics for use aid to program comprehension. System Sciences,
with text-to-speech. Proceedings of the 24th annual 1990., Proceedings of the Twenty-Third Annual
conference on Design of communication, 157-163. Hawaii International Conference on, 2.
Lesser, E., Mundel, D. & Wiecha, C. (2000) Managing Such, A., Principles, K., Methods, K., Roles, I., Mapping,
Customer Knowledge. Journal of Business Strategy, K. & Reasoning, K. P. R. (2001) Linking e-business
21, 34-7. and operating processes: The role of knowledge
Massey, A. P., Montoya-Weiss, M. M. & Holcom, K. management. IBM Systems Journal, 40.
(2001) Re-engineering the customer relationship: Thierauf, R. (1999) Knowledge management systems for
leveraging knowledge assets at IBM. Decision Support business, Westport, Conn.; London, Quorum.
Systems, 32, 155-170. Tiwana, A. (2001) The essential guide to knowledge
Osterle, H. (1995) Business in the Information Age, management : e-business and CRM applications,
Springer. Upper Saddle River, NJ. London, Prentice Hall PTR ;
Pan, S. L. & Lee, J. N. (2003) Using e-CRM for a unified Prentice-Hall International.
view of the customer. Communications of the ACM, Urban, G. L. & Von Hippel, E. (1988) Lead User
46, 95-99. Analyses for the Development of New Industrial
Rigas, D. (2003) Empirically Derived Multimedia Design Products. Management Science, 34, 569-582.
Guidelines for Browsing Large Volumes of E-Mail Von Hippel, E. (2001a) Learning from Open Source
Data. Software. Sloan Management Review, 42, 82-86.
Rigas, D. & Alty, J. (2005) The rising pitch metaphor: an Von Hippel, E. (2001b) User Toolkits for Innovation.
empirical study. International Journal of Human- Journal of Product Innovation Management, 18, 247-
Computer Studies, 62, 1-20. 257.
Rigas, D. I. & Alty, J. L. (1998) Using sound to
communicate program execution. Proceedings of the
24th EUROMICRO Conference, 2, 625–632.
Rigas, D. I., Alty, J. L. & Long, F. W. (1997) Can music
support interfaces to complex databases.

148
 A PERVASIVE NUTRITIONAL MONITORING AND ADVISE
SYSTEM
NutriMe

Vítor Basto, João Varajão and António Cunha

Department of Engineering, University of trás-os-Montes e Alto Douro, Quinta de Prados, Vila Real, Portugal
[email protected], [email protected], [email protected]

Keywords: Nutrition, Pervasive Nutritional Monitoring, Nutritional Advise System.

Abstract: It is well known, widely accepted, scientifically proved and published by major governmental and non-
governmental organizations worldwide (e.g. WHO - World Health Organization), that nutritional
misbehaviour in so called developed countries, is a major cause of diseases, morbidity and death. The
phenomenon is mainly felt in aged populations, but a significant increase has also been detected more
recently in young populations. This paper presents a proposal to tackle serious social and behavioural
problems related to aging and nutrition. NutriMe is presented as a nutritional monitoring and advising
system to help individuals to monitor and correct their behaviours. We also propose NutriMe as the main
component for a public national observatory on nutritional profiles for public health analysis purposes.

1 INTRODUCTION have been prevented by adopting healthier lifestyles,

which includes correct nutrition, regular physical
Several economical, social and cultural factors that activity and non-smoking (WHO, 2006). WHO also
took place in the last decades in developed countries states that: “A change in dietary habits, physical
have strongly influenced human diseases profiles. activity and tobacco control, have a major impact in
Among those factors, globalization and urban reducing the rates of these chronic diseases, often in
prevalent and increasing lifestyles are worth to a relatively short time”.
mention. One of the reflexes turning out from those Nutritional monitoring and advise is therefore
changes is related to nutritional (mis)behaviours important both in an individual point of view for
(Lopes et al., 2006). Although other reflexes are also individual behaviours correction, and in a global
subject of research (e.g. smoking, pollution, etc.), we point of view, essential for global policy definitions
will not take them into account in our study, and for nutritional education planning.
knowing that nutrition is referred as one of the most NutriMe, the system proposed in this paper, is a
important factors. system designed to tackle the above presented
The outcome of several years of research about issues. NutriMe can also be integrated with the
nutritional reflexes on health/diseases, lead to a Smart pantry project - Diet module (Alves et al.,
present common sense assumption that proper 2006), whose features consist of a subset of NutriMe
nutritional monitoring and advise is need, must be features in the context of a smart house project
continuous, rigorous and customised for each focused on accessibility and inclusion.
individual, according to biological, medical, and life In section 2 a general characterization is made of
style parameters (Lopes et al., 2006). the Smart pantry project and explained its possible
The extent of many harmful reflexes (e.g. integration with the NutriMe system, followed by
morbidity and mortality) caused by incorrect section 3 which presents the NutriMe system.
nutritional behaviours on health, have been Section 4 shows a NutriMe prototype and some
estimated in several studies. World Health issues related to the distributed data model are
Organization (WHO) reports that 80% of presented in section 5. Finally, conclusions and
cardiovascular disease cases, 90% of diabetes future work are referred in section 6.
mellitus type 2 and 33% of all types of cancer could

149
ICE-B 2008 - International Conference on e-Business

2 SMART PANTRY PROJECT control system drawing its inspiration from a bucket
chain); another will be a robotic system of the ‘mini-
Developed countries aged populations is raising warehouse’ variety, to be installed in its own
several and severe problems related to home daily compartment – the ‘pantry’ (modular systems of
elder people tasks, such as physical access and greater capacity).
handling of home stored goods, nutritional control An extremely important element is the HCI
and health care, etc. The Smart pantry project (nicely) module (human computer interface module)
(Alves et al., 2006) was developed with the intention that will be programmed according to user profiles
to fulfil some of those needs. Its main features are and specific needs, so that the system can interact
related to stock control, triggering advertises to with each person according to his personal
replace depleted/missing products, create automatic physical/psychological skills and impairments
customized shop lists, make products physically (image-based, text-based, voice based, simplified vs.
accessible to people with reduced mobility, full featured versions).
suggesting menus tailed to user preferences and The diet module has a set of menus that allow the
medical profiles (diets), all supported by a central counselling of users, accounting for different
database system (Barrias et al., 2008). Figure 1 criteria, such as the number of people (family size),
presents a modular architecture for the smart pantry the status of the food stock, or the way for eating
as proposed in (Alves et al., 2006). adequately.
Smart pantry project diet module features are
considered as part of NutriMe project as the home
based monitoring and advising features. Nutrime
intends to cope with ubiquitous nutritional
monitoring also in other contexts like restaurants,
canteens, bars and pubs, vending machines (food,
beverage, cigars automatic selling machines), etc.
Nutrime extends the concept of nutritional
monitoring to ubiquitous individual monitoring and
advising system.

Figure 1: Modular architecture for the smart pantry

Source: (Alves et al., 2006).
3 NUTRIME
In this system, users enter products using the
entry module, which automatically identifies the NutriMe is a distributed software and distributed
product, presents its features to the user and asks data based system that collects nutritional
him for confirmation/validation. After validation, the information from different sources. In addition to
system stores the product according to its storage individual nutritional information gathering, it also
conditions (temperature and humidity), size, expiry links, relates data, ensures consistency and
dates, etc. integration of syntactic and semantic data models
Based on the storage information in the database, from different sources (food suppliers – home,
the stock management module manages the restaurants, vending, etc.). NutriMe uses that
existence of stored products (e.g. expiry dates, so information not only for individual real-time
that they can be disposed or used), a ‘shopping list’ nutritional monitoring and advising, but also for the
is also presented based on predicted needs, on-line purpose of feeding a nutritional national database
shopping with supplier’s systems integration (nutritional ‘observatory’). The observatory
(Cardoso et al., 2007), history of menus, etc. database allows for classification, segmentation and
The storage module consists of a robotic storage prediction of nutritional profiles. Based on data
system, adapted to user needs (accessible) and to analysis and knowledge extraction (data mining)
available space and conditions. It may be based on from this database, public health policies and
rotating shelves, suspended elements, or other strategies can be better supported and deployed.
systems such those employed in large-scale storage Figure 2 shows the components of NutriMe
systems. (conceptual view).
In order to produce a prototype, two basic
solutions are under development: one thought to be
part of traditional kitchen furniture (a numeric-

150
 A PERVASIVE NUTRITIONAL MONITORING AND ADVISE SYSTEM - NutriMe

following major features: provide mobile

applications/code to be downloaded into
customer mobile devices allowing for
customer multimedia interactive menu
selection (e.g. food, drink, desert, etc.), for
delivery selection (e.g. tables and location of
the customers to be selected based on
interactive maps), for customer identification
and profile management (e.g. personal data for
Figure 2: Components of NutriMe (conceptual view). invoice and receipts), detailed and electronic
invoice issuing (detailed info about consumed
Each component relies on different technologies items) sent to the customer mobile device,
and targets different purposes and features of a electronic payments, etc.;
global nutritional monitoring and advising system. ƒ Automatic selling machines (vending)
NutriMe conceptual model, its individual component, this is a generic component
components roles and descriptions, follow: representing any other system component
providing and/or collecting nutritional
ƒ Home desktop component, supports all the information (e.g. beverage and food automatic
features mentioned in the Smart pantry project selling machines). Each of these components
– Diet module. The major features specified must provide mobile code to be run on user
are related to user profile characterisation mobile devices or implement a compliant
(age, gender, physical activity profile, generic communication protocol profile (to be
professional profile, medical profile, etc.), defined) for integration with the mobile
alimentary items characterisation device;
(identification, common description, ƒ Nutritional observatory component, includes
nutritional data composition, etc.), diets, features of data collection from individual
healthy profiles characterisation, nutritional devices (desktop and mobile devices) and
reports generation and nutritional advise; restaurant systems, data synthesis, nutritional
ƒ Mobile device component, implements a mobile profiles analysis based on multiple criteria
version of the Home desktop component and (e.g. age, location/geography, profession,
features for interoperation and integration with etc.), allowing for population risk
all the other system components that provide classification and evaluation concerning
(or collect) nutritional data for the individual, nutritional behaviours and diseases prevalence
like the Home desktop, Restaurant, Automatic analysis.
selling machines and Nutritional observatory
components. It is intended to download and NutriMe intends to promote healthy nutritional
run mobile code made available from the other behaviours by the means of ubiquitous nutritional
system components (e.g. in restaurants), monitoring in an individual and population basis
behave like an extended GUI (graphical user (reporting individual and population nutritional
interface) of those systems and interchange warnings). In addition, it is intended to provide
(collect/provide) information about individual detailed individual nutritional advising (suggesting
user consumed items from those system detailed meals according to nutritional principles
components. The consumed items identified and user preferences), global behaviour synthesis,
and/or described are to be found in the mobile risk evaluation and classification using data mining
device database for nutritional monitoring, techniques.
processing and advise purposes. The mobile Figure 3 shows UML (Unified Modeling
component is also able to look through Language) Use Cases specification for NutriMe.
consumed items description provided by the It shows the system actors, features, their
other system components, map them into the interactions and relationships (Fowler, 2003).
nutritional characterisation and run the
respective nutritional monitoring, evaluation/
classification and advise algorithms;
ƒ Restaurant component, our concept of a “smart
restaurant” system component includes the

151
ICE-B 2008 - International Conference on e-Business

Physical and health

NutriMe “Automatic consumption registration” is
possible. “Automatic consumption registration” is
<<include>>
Consumer profile setup
profile setup
<<include>>

Food consumer Diet profile setup

the preferred way of nutritional data gathering in
NutriMe, but “Manual explicit consumption
Nutritional
misbehaviour
Manual explicit
consumption
registration” by “Food consumers” are also
alerts registration
available in NutriMe.
Nutritional monitoring
reporting and advising
Individual nutritional and
NutriMe simplified data model is presented in an
health data analysis and data
mining
Nutritionist
Entity-Relationship diagram in Figure 4. The
conceptual data model will be deployed as a
Nutritional data
Group nutritional and
health data analysis
distributed relational database system as detailed in
observatory feed
Health
observatory
and data mining
Health
observatory
section 5. As a distributed software and distributed
system
data based system that collects nutritional
manager

Restaurant
Setup restaurant
virtual space information from different sources and
interface

Food supplier
manager
heterogeneous technologies, concerns of
manager Virtual
interfaces setup
consistency, correctness, integration of syntactic and
Restaurant systems Food
semantic data models are to be considered.
<<extend>> suppliers
systems

Setup vending Automatic

Vending machines virtual consumption
machines interface registration
Vending
manager
machine
systems

Figure 3: NutriMe Use-Cases Diagram.

Use Case “Consumer profile setup” allows the

user to provide his personal data (e.g. gender, age,
height, weight) to the system, defining nutritional
profiles, diets and preferences (e.g. vegetarian,
athlete, diabetic, etc).
Use Cases “Nutritional misbehaviour alerts”
and “Nutritional monitoring reporting and
advising” allows for automatic warnings from
NutriMe towards the user, allows the user to access
statistics about his behaviour and provides
nutritional advising. In addition, NutriMe provides
extra nutritional expertise and tools for knowledge Figure 4: NutriMe Entities-Relationships Diagram.
extraction (clustering, classification, prediction)
from monitoring data for nutrition professionals An innovative implementation of NutriMe
(“Nutritionist”) decision support in “Individual involving flexible and adaptive information systems
nutritional and health data analysis and data integration based on desktop and mobile devices is
mining” Use Case. presented in section 4.
An observatory component of NutriMe is also
feed by nutritional monitoring data (Use Case
“Nutritional data observatory feed”) for “Group
nutritional and health data analysis and data 4 PROTOTYPE
mining” purposes.
“Virtual interfaces setup” Use Case supports Inclusion and accessibility have been major
user interface, data and functional integration concerns during all development life cycle of
features between NutriMe and “Food supplier NutriMe. Software design and technology selection
systems”. For instance, it is possible for “Food were essential to construct a solution that promotes
consumers” to issue meal orders from NutriMe that device and software mobility, usability, high
are transmitted electronically and processed by availability and continuous monitoring.
“Restaurant systems”. When “Food consumers” Mobile technologies such mobile phones/PDAs
interact with “Food supplier systems” through proved recently to ease cumbersome daily task.

152
 A PERVASIVE NUTRITIONAL MONITORING AND ADVISE SYSTEM - NutriMe

Recent devices reached significant processing robustness allowing not only for horizontal but also
storage and communication capabilities, allowing vertical handoffs.
increasing features and new applications. The Next we present some prototype features and
massive adoption of this kind of devices is graphical user interface of NutriMe. User personal
considered today one of the greatest success of profile form is presented in Figure 6 (left picture) for
information and communications technologies user characterisation features of Definition of
market acceptance (Paes, 2006). Personal and Nutritional Profiles Use Case. And
The above mentioned technical factors, growing interactive consumption registration form is also
market acceptance, easy of use and ubiquity of this presented in Figure 6 (right picture) for
kind of mobile technologies, lead us to select this Consumption Monitoring Use Case features.
computing and communication technologies as the
basic support for the main NutriMe component.
Figure 5 shows the technological solution
proposed for NutriMe system main components,
including physical/logical elements, interoperation
and interactions.

Figure 6: NutriMe personal profile form and Interactive

consumption registration form.

NutriMe user interface also provides friendly

graphical classifications of nutritional and health
ratios, such as the BMI - body mass index (Figure
Figure 5: Technical scenario for NutriMe mobile 6), which are subject of time evolution graphical or
component prototype. text based presentation and analysis.
Supporting distributed relational database
NutriMe software (mobile and desktop) technologies and strategies for NutriMe desktop and
components are hosted in PC like and mobile mobile prototypes are presented in section 5.
devices (e.g. mobile phones/PDAs).
NutriMe desktop and mobile versions have been
developed with Microsoft .NET 2005 Framework
and Microsoft .NET Compact Framework
5 DISTRIBUTED DATA MODEL
respectively. Microsoft SQL Server 2005 and
Available commercial DBMS (e.g. Microsoft SQL
Microsoft SQL Server 2005 Compact were de
Server) allow for flexible data management,
DBMS (database management systems) adopted for
enhancing application robustness, high availability
desktop and mobile support respectively. Data
and performance, being especially important in
replication, integrity and synchronisation are
distributed systems involving mobile devices with
supported by both mentioned DBMS (master-slave
non continuous communication connectivity.
data replication and synchronisation) which is the
Replication strategies are usually adopted (and very
most common strategy adopted for similar scenarios.
effective) to overcome periods of non connectivity.
This strategy shows optimal trade-offs concerning
However, data replication needs specific
quality of service (availability, punctuality, etc.) and
mechanisms for integrity enforcement among the
resources consumption/usage (storage, processing
copies spread throughout the several components of
and communication) within strongly connected,
a distributed system (e.g. mobile and desktop
weakly connected and connectionless situations.
equipment). While subject to isolated processing
Support of multiple communications technologies
during non connectivity periods, data replicas in
(GPRS, UMTS, Wi-Fi, Bluetooth), actually quite
different processing devices can evolve to divergent,
common in mobile devices, enforces applications
eventually inconsistent states. Automatic

153
ICE-B 2008 - International Conference on e-Business

mechanisms must be activated to rollback or roll- analysis of general nutritional tendencies and
forward the replicas to get the overall distributed patterns.
system back to a consistent state. The mechanisms
adopted in our system for this purpose are based in
the two-level master-slave transaction model (Liu et REFERENCES
al., 1999).
One level is ruled by desktop workstations, the Lopes C., Oliveira A., Santos A., Ramos E., Severo M.,
second level is ruled by the mobile devices (one or 2006. Consumo alimentar no Porto. Porto: Serviço de
more devices). Data convergence actions take place higiene e Epidemiologia - Faculdade de Medicina do
when a communication channel is available between Porto.
desktop equipment (master replicas) and mobile WHO, World Health Organization. 2003. Global strategy
devices (slave replicas). on diet, physical activity and health. Available:
First, the master (re)executes all transactions URL:http://www.who.int/hpr/NPH/docs/gs_global_str
made available by the slave, corresponding to all the ategy_general.pdf
Alves J., Cunha A., Trigueiros P., Barroso J., Cruz J.,
actions performed by the slave during non
2006. Sistema de gestão inteligente de uma despensa.
connectivity periods. Second, the master notifies the In DSAI 2006, Proceedings of Conference Software
slave the successful reconciliation transactions to be Development for Enhancing Accessibility and Fighting
committed. Inconsistent transactions are tracked Info-exclusion, Vila Real.
back until its root (causal graph node), and undone Barrias A., Cunha F., Varajão, J., Cunha A., 2008. Smart
until the overall distributed system gets into a Pantry: specification of a management system using
consistent status. UML. In IADIS e-Society 2008, Proceedings of IADIS
The presented two-level master-slave replication International Conference e-Society 2008, Algarve.
strategy seemed to us a suitable solution for the Fowler K., 2003. UML distilled: a brief guide to the
standard Object Modeling Language, Addison-
NutriMe system. It revealed optimised trade-offs
Wesley. 3nd edition.
according to quality of service and resource usage, Liu P., Ammann P., Jajodia S., 1999. Incorporating
in scenarios similar to ours. Transaction Semantics to Reduce Reprocessing
Overhead in Replicated Mobile Data Applications. In
ICDCS' 99, Proceedings 19th IEEE International
6 CONCLUSIONS Conference Distributed Computing Systems, Austin.
Cardoso J., Jaulino R., Varajão J., Fernandes V., Cunha
A., 2007. Smart pantry: integration models with
This paper presents a software based system – electronic commerce solutions, In DSAI 2007,
NutriMe, which addresses ubiquitous nutritional Proceedings of Conference Software Development for
monitoring and advising supported by several Enhancing Accessibility and Fighting Info-exclusion,
cooperating software components distributed Vila Real.
through desktop and mobile devices, databases and Paes C., Moreira F., 2006. Aprendizagem com
applications. The system intends to introduce Dispositivos Móveis: Aspectos Técnicos e
Pedagógicos a Serem Considerados Num Sistema de
increasing computational and communication
Educação.URL:http://grace.tech-x-pert.org/blogs/
pervasive features, assisting in simple daily tasks files/2007/06/aprendizagem-com-dispositivos-moveis-
with low or non intrusive reflexes (low or no na-aspectos-tecnicos-e-pedagogicos-a-serem-
perception or interaction required from users). considerados.pdf
Because aged populations are common in developed
countries, and they have usually special needs,
concerns of accessibility and usability have been of
major importance in the design and implementation
of NutriMe prototype. Individual and public interest
(e.g. health, economic) of this kind of systems are
worth to mention. They constituted our first interest
and justified further work and research on this topic.
Future work milestones include information
gathering and processing in public spaces context,
multimodal interfaces for impaired people,
introduction of data mining techniques for
customised individual user advise, extraction and

154
REPLICATION OF WEB SERVICES FOR QOS GUARANTEES IN
WEB SERVICE COMPOSITION

Dirk Thissen and Thomas Brambring

Department of Computer Science, RWTH Aachen University, Ahornstr. 55, 52074 Aachen, Germany
[email protected]

Keywords: Web services, replication, quality of service, proxy.

Abstract: The concept of web services defines a middleware for implementing distributed applications independent of
used platforms and programming languages. When developing new software systems, re-use of function-
nality of existing services can be done to reduce development time and costs. This process of re-use is called
web service composition. But, current web service standards are not equipped to consider non-functional
requirements, i.e. quality of service (QoS) aspects of a user to a composed service. Thus, capabilities of
composed services cannot be guaranteed. This paper presents an approach to integrate QoS aspects into the
composition of web services by using service replication. At composition time, service instances are chosen
depending on the QoS requirements of a user to the whole service, and it is decided which services in the
composition have to be replicated and which replication strategy to use. Replication ensures that the QoS
requirements are not only considered at service selection time, but also can be granted at service runtime.

1 INTRODUCTION tion of services in the composition due to a user’s

requirements. But, QoS aspects can be dynamic, so a
A web service is some software which is seen as a QoS-oriented selection is only considering a kind of
service it offers, aiming at automatic machine-to- system snapshot. Depending on the time between
machine communication independent of the imple- service selection and the execution of a selected
mentation of the software. A software architecture is service in the composition (basing on the composit-
defined to give standards for service definition ion pattern and the interaction between the services),
(WSDL) and interaction (SOAP). Web services are QoS information used at selection time can be
loosely coupled; the search for services to communi- outdated at service usage time. Thus, the architecture
cate with is done dynamically at runtime using a is enhanced by service replication to guarantee the
service registry (UDDI). The interaction across plat- QoS from selection time also at runtime. A flexible
forms and programming languages enables easy and replication framework was developed to allow for as
fast deployment of new software: complex software well performance-related as fault tolerance- and
systems can be plugged together from existing availability-related replication of services.
services to a collaborating group. This is called web The paper is structured as follows. Chapter 2
service composition. One vision is to achieve an presents a short overview about related work in the
automatic composition and interaction of services. area of QoS and replication in web service compo-
But, the standards are not equipped to consider non- sition. In chapter 3, the principle of the QoS broker
functional demands to a composition, i.e., QoS is explained. Chapter 4 describes the replication
requirements like performance, reliability, or cost. architecture and gives an overview about the current
For acceptance by customers, a business should implementation status. Finally, chapter 5 concludes
provide good quality of its composed services. This the paper and gives an outlook on the ongoing work.
paper focuses on handling QoS aspects in web
service composition. The service registry UDDI was
enhanced to select web services for a composition 2 WEB SERVICES AND QOS
with respect to the QoS requested by a user. A QoS
broker was added to UDDI to manage the services’ Several approaches for the composition of web
QoS information and to calculate the best combina- services exist. A prominent example is the Business

155
ICE-B 2008 - International Conference on e-Business

Process Execution Language (BPEL) for Web tent. If the primary fails, a backup can take over its
Services (OASIS, 2007). But these approaches are role, which improves fault tolerance and availability.
not considering quality of service in the composition There are lot of other replication algorithms, and
process. Though there is a lot of research in web also approaches exist to implement them within a
services and composition, not much is related to web service architecture. E.g., (Ye and Shen, 2005)
QoS. The existing work mostly refers only to a part discuss active replication for web services. But, the
of the whole problem. (Liu et al, 2004) for example focus is only on reliability of web services, and only
present a framework to publish up-to-date QoS active replication is implemented. The same holds
information for web services, but the success of this for (Chan et al, 2007): it is focussed on reliability.
mechanism depends on feedback of users about the WS-Replication (Salas et al, 2006) also uses active
quality of the services they consume. (Zeng et al, replication to achieve high availability, and WS-
2004) present a method to select services that fit to a multicast is used for communication between the
user’s interest (expressed as QoS parameters). Local replicas. WS-multicast is SOAP-based and maybe
optimization and global planning are combined to causes a high overhead. (Osrael et al, 2007) is a
find the best set of services for a composition. But, more flexible approach, implementing passive
in case of highly dynamic QoS parameters, the replication and designing an open system for later
global planning approach might take more time for addition of other replication strategies. Consistency
re-calculation than the execution of the service can be weakened in this approach to reduce the
would need. Thus the approach itself can violate the performance overhead caused by update propaga-
QoS. (Jaeger et al, 2004) propose a mechanism tion. But, till now only a variant of passive replica-
which could be more efficient by using an aggrega- tion is realized, and the focus is on fault tolerance.
tion scheme for QoS aspects. The approach sounds Concluding, the replication approaches either
well but was not implemented nor tested by the focus on only one replication strategy, use multicast
authors. We used this approach as basis for the on SOAP level which decreases performance, or
implementation of an own solution to consider QoS only consider a certain QoS aspect, e.g. availability.
aspects in composition (Thißen and Wesnarat, 2006) Thus we designed an own replication framework for
which is explained more detailed in chapter 3. integration with composition, which offers more
All these approaches have the same weakness: flexibility, see chapter 4.
services for the composition are chosen some time
before execution. If QoS parameters change, during
service execution the QoS demands of a user never- 3 QOS IN SERVICE SELECTION
theless can be violated. Replication is a possible
solution to deal with dynamic QoS parameters on For composing web services under QoS constraints,
performance, high availability, and fault tolerance/ we followed the approach presented in (Jaeger et al,
reliability. Instead of running a single instance of a 2004): a workflow pattern is given, showing the
service, several copies are used. Replication defines relations between services. Aggregation rules are
methods for keeping consistent all copies (called used to combine quality measures assigned with
replicas). The complexity is hidden from the user of single services to come to an overall rating of sets of
a service by a frontend which acts as the service services. We identified relevant QoS information
from the user’s view. A lot of replication algorithms and basic composition patterns (SEQUENCE, AND,
are given. In active replication, all replicas act in the OR, XOR, and LOOP) from which the whole
same way. The frontend uses group communication workflow pattern can be formed. Next, we defined
to distribute a request to all replicas. It decides how corresponding aggregation rules and a selection
to deal with responses of the replicas, depending on mechanism to choose the best service candidates.
the QoS aspect which should be considered. To Given the workflow pattern for a composed service,
ensure service available or to decrease the response aggregation of QoS parameters is done by collapsing
time of a service (performance), the frontend returns the whole composition graph step-wisely into a
the first response to the user. To improve fault single node, starting with the innermost composition
tolerance, it compares and combines all responses. A pattern. By aggregating the properties recursively,
different approach is passive replication. One replica only one node is left in the final state. A set of
is a primary, and the frontend only communicates formulas was defined to model the aggregation of
with this replica. The primary forwards the requests the QoS parameters performance, cost, reliability,
to all other replicas (backups) to keep them consis- and availability. This mechanism enables us to
check the resulting QoS of a set of services. Because

156
 REPLICATION OF WEB SERVICES FOR QOS GUARANTEES IN WEB SERVICE COMPOSITION

4. Select the most

suitable sub-services

1. Request a 2. Request WSDL of

composite Service sub-services
Service QoS UDDI
Requestor Broker
5. Receive 3. Receive WSDL of
information of sub-services
composite service QoS and BPEL
with the selected Registry
sub-services
A. Publish Services

6. Request
for service B. Publish QoS
9. Receive information
Service C. Publish
BPEL file
Monitor

Service
Service Monitor Provider
7. Request service from
Orchestrator selected sub-services
Monitor
Service
8. Receive
service from
Provider
selected sub-
services
Service
Provider

Figure 1: Enhancement of the web service architecture.

we need to find a set of services to be executed, each not; the broker uses the BPEL registry to search for
possible combination of service candidates for the a composition pattern. If one is found, the broker
current composition pattern is evaluated, and the asks UDDI for available candidates to all services in
best ones regarding the user’s demands are selected. the pattern (step 2). Having retrieved a list of all
Multiple criteria decision making and weighting are available candidates (step 3), the broker connects to
used to combine a service set’s aggregations for the QoS registry to get the QoS values for these
different QoS parameters into one value, a quality services. By stepwise aggregation of the values
scores. For a composition pattern, the set of service according to the pattern from the BPEL registry and
candidates with highest quality score is selected, and by selection of the best fitting candidates, a set of
it is done aggregation of the next innermost basic services is chosen (step 4). The requestor gets
back a reference to an orchestrator for using the
composition pattern till a single node (the composed
service (step 6/9). The orchestrator manages the
service) remains with assigned QoS values.
service execution (step 7/8). After execution, it gives
For implementation of this approach we have feedback to the broker. In other requests to the same
designed a prototype which enhances the general composed service, the broker can make use of it.
web services architecture. Apache tomcat was used Not included in figure 1 is the use of the QoS
as web container for the provided web services, monitors. Getting the QoS information for aspects
Apache Axis services as SOAP implementation. like cost is no problem: the values are constant for a
jUDDI was chosen as UDDI registry, for executing longer period of time and can be filled in by the
composed services the Oracle BPEL Process service provider at service setup. But most aspects,
Manager was used. It provides a service orchestrator are dynamic, e.g. like performance. Thus a monitor
which can be assigned a workflow pattern and a set is assigned each service to record its behaviour, to
of basic services; it then manages the execution of compute floating averages, and to forward this infor-
the services due to the pattern. For integrating QoS mation to the QoS registry. To avoid that service
consideration as described before, we have imple- providers have to modify their services, the monitors
mented some more components, see figure 1. are independent components. They get the needed
The central component is the QoS broker which information from so called valves placed on Tomcat
implements the QoS aggregation rules. It involves a engine level. Here, e.g. timestamps can be used to
BPEL registry and a QoS registry. Service providers
get statistics about the queuing time of a request.
as usual register their services with UDDI (step A in
Nevertheless, the QoS broker cannot guarantee
figure 1). To publish QoS information, a monitor is
assigned each service, registering with the QoS the QoS from selection time to be constant at
registry when a service is put into UDDI (step B). runtime, thus we had to enhance this architecture by
When a composed service is deployed, the workflow a mechanism which allows for some control at
pattern is stored in the BPEL registry (step C). execution time of the services.
When a service requestor searches for a web
service, it contacts the QoS broker (step 1). It does
not need to know if a service is a composed one or

157
ICE-B 2008 - International Conference on e-Business

4 REPLICATION FOR QOS resp. the assigned QoS registry via their monitors (b)
GUARANTEES AT RUNTIME as described in chapter 3. The services’ monitors do
not need to know if they are belonging to a
The architecture presented in chapter 3 is only able replicated service or to a simple one, they have to
to consider user demands at selection time. Our next submit the same information as before. The monitors
step was to enhance the architecture by capabilities regularly measure the QoS values of their replicas,
of replication, to control the selected QoS at run- calculate advanced information like floating
time. Because of the disadvantages of existing averages, and deliver the resulting values to the QoS
approaches, we designed an own replication broker.
architecture considering the following goals:
1 2
ƒ Allow for flexible choice of replication algorithm Service  QoS Service 
at runtime. We want to use replication for requestor Broker Directory
4 3
guarantees on several QoS aspects, thus we need
different replication strategies supporting perfor- b a
6 5
mance, availability, and fault tolerance in one
approach. Replicated service
ƒ Open architecture which can easily be enhanced
with new replication algorithms. For the begin- Service  Service  Service 
Provider Provider Provider
ning, we only considered the most prominent
algorithms: active and passive replication.
ƒ Decide at composition time which services have
to be replicated to fulfil a requestor’s demands.
E.g., the use of several replicas may improve the Figure 2: Replication enhancement.
reliability, but may contradict the cost of service
usage if one has to pay for each extra replica. If a service requestor contacts the QoS broker to
Thus, in the selection process a tradeoff is ask for a service (1), the broker interacts with UDDI
necessary between gain and costs of using repli- to find all replicas to the requested service (2 + 3).
cation, including the number of replicas to use. Based on the requested QoS, the broker now can
ƒ Transparently use group communication and select a subset of fitting replicas which seems to be
avoid communication overhead by using SOAP. sufficient to fulfil the requestor’s demands.
Otherwise, replication could contradict the QoS. Simultaneously, it can decide on the best replication
ƒ Automatically generate request and result strategy regarding the requested QoS. If e.g. high
classes for web services from WSDL files. performance is needed primarily, active replication
Reduce the costs and time for integrating the is chosen to reduce response times. If availability
mechanisms into each application newly. has priority, passive replication is more appropriate
to reduce the communication overhead. Based on the
We designed our replication architecture known availability probability of the service, also
oriented at these goals and allowing for easy inte- the number of replicas could be determined.
gration with our QoS-based selection of service Currently, active and passive replication are imple-
candidates in a composed service (chapter 3). In the mented in our prototype, and only a few rules are
following, the components of the architecture and implemented, which strategy to use in which cases.
their interaction are described in more detail. The service requestor gets back a reference for
The QoS broker remains the central component its service (4) and can use it (5 + 6). The detailed
of the architecture. It is enhanced by enabling the information transmitted in these steps depend on the
selection of a replication strategy as well as a set of replication strategy chosen by the QoS broker since
suitable replicas for a service. For simplicity, we the service requestor maybe has to contact a single
started with the consideration of simple services service or maybe a service group.
within the replication process, but oriented at the If the broker chooses passive replication, the
composition architecture for easy integration. requestor only communicates with a single replica.
The interaction of the QoS broker with service In contrast to common passive replication there is no
requestor and the replicas of a single service is fixed primary replica which all the time is contacted.
shown in figure 2. The replicas are all registering Instead, the QoS broker chooses the actually best
with UDDI as usual (step (a) in figure 2). The replica due to the requestor’s demands and returns a
replicas additionally register with the QoS broker reference to this replica to the requestor. The other

158
 REPLICATION OF WEB SERVICES FOR QOS GUARANTEES IN WEB SERVICE COMPOSITION

replicas only serve as backups and are invisible to in its request (step 1 in figure 3). Such a request may
the requestor. The dynamic primary selection allows look as follows:
for a better average QoS level in terms of perfor- Proxy.requestActively(request,
mance because it enables a kind of load balancing READ_ONLY, GET_FIRST, 2000);
between all available replicas. But, one has to keep The original request of the user is passed to the
in mind that using different primaries for different proxy only as one parameter request. The proxy
requests can cause consistency violations. Thus it is able to process this request by using the corres-
depends on the service itself if the weakening of the pondingly assigned replicas. In which way to use the
consistency is useful. replicas, is defined by the other parameters of the
On the other hand, in active replication the user’s call. The proxy implements functions
requestor has to communicate with all replicas requestActively and requestPassively.
simultaneously. Thus he has to communicate with a
Depending on which replication mechanism is
group of services instead with a single service.
chosen the client has to call the corresponding
To hide the different usage for the replication
function. The client gets this information from the
schemas, proxies are used. They encapsulate the
QoS broker as part of the configuration information.
functionality of communication with replicated
The second parameter of the request tells the proxy
services. Only a single interface is offered to the
if the request is read-only or not. In case of read-
requestor. Independent if passive or active replicat-
only, consistency is relaxed, which can improve the
ion is used, the requestor gets back a reference to the
performance of a request. This parameter is followed
used proxy instead of a reference to a concrete
by an information if the first response has to be
service (in step 4 of figure 2) – the proxy itself
forwarded to the client (e.g. for performance or
seems to be the service for the requestor. This
availability aspects), or if the proxy has to wait for
schema intentionally is designed similar to the usage
all responses and to combines them in some way to
of composed services via an orchestrator as descry-
achieve fault tolerance. The last parameter is a
bed in chapter 3, to merge the functionalities of
timeout. It defines how long the proxy has to wait
orchestrator and proxy. The only difference for the
for responses before combining the received results
requestor is that the QoS broker not only sends back
(or before sending an error message back to the
a reference to a service (the proxy), but also some
requestor).
additional configuration parameters the requestor
The proxy now can inform the group
has to use in its request to enforce a certain replicat-
communication component about the needed
ion process (which was chosen by the broker).
communication mechanism (2) and the request
4 3
correspondingly is passed only to a single service or
to a group of services (3). The results which are
Client
coming back from the replicas (4) are passed on to a
Group Com- 2 Requestor 1 Service  message handler (5) which can treat the responses in
munication interface request different ways as described above. If passive
5
replication was used, the proxy immediately uses a
Message Callback
Handler functions callback function to deliver the result to the
6
requestor (6). In case of active replication, it can
forward the first response to the client, or collect all
Client Proxy requests coming in before a timeout and form a
consensus out of them before passing only a single
response to the requestor.
Figure 3: Client proxy.
Also on server side a proxy is needed to
coordinate all replicas corresponding to the chosen
Figure 3 shows the structure of such a proxy.
replication strategy, see figure 4. The request comes
The client only holds a reference to its proxy, which
in over the group communication mechanism (step 1
is capable of performing all replication strategies for
in figure 4) and is forwarded to the message handler
any kind of request. The proxy is informed by the
(2). The message handler in the background interacts
QoS broker about the set of replicas to use for a
with the QoS monitor (a) to allow for statistics about
request. To inform the proxy how to handle a certain
the number of requests per second, response times,
request, the requestor now has to include the
etc which is part of the QoS parameters collected by
configuration parameters chosen by the QoS broker
the monitor. Because in active replication
consistency requires a sorted execution of requests

159
ICE-B 2008 - International Conference on e-Business

from different clients on all replicas, the holdback valves like the monitors). Afterwards, the gain of
queue (3) delays all requests till they can be using replication in the composition again has to be
executed without violating consistency to other evaluated by a number of experiments.
replicas. To do so, requests have to be sorted the Replication only is one way to improve the
same way for all replicas. For this purpose Lamport quality of a service. After finishing our current
Timestamps are used. The requests are sorted into a work, beside integrating more replication strategies
delivery queue (4) which simply implements a FIFO we want to examine if instead executing the same
strategy and executes one request to the service after service several times, also equivalent services of
the other. The requests can be passed on to the web different providers could be used. Also, we plan to
service by using a callback function (5 + 6). The enhance the functionality of the proxies by other
delivery queue also gets back the response (7 + 8) strategies, e.g. load balancing as a mechanism with
and initiates the transmission of this response back weaker guarantees as replication, but on the other
to the requestor (9). Again, the group communicat- hand cheaper if services – and quality guarantees –
ion mechanism takes over the transmission of the have to be paid for.
result to the requestor (and to the backups, in case of
passive replication).
1 10
REFERENCES
Replica server

9 Chan, P.P.W., Lyu, M.R., Malek, M., 2007. Reliable Web

Group Com-
munication Services: Methodology, Experiment and Modeling.
2
8 7 Proc. IEEE International Conference on Web Services
Message
Handler
Holdback
Queue
Delivery
Queue
Callback
function
Web
Service
(ICWS 2007), Salt Lake City, USA.
a
3 4 5 6 Jaeger, M.C., Rojec-Goldmann, G., Mühl, G., 2004. QoS
Monitor Aggregation for Web Service Composition using
Server Proxy
Workflow Patterns. Proc. 8th IEEE International
Enterprise Distributed Object Computing Conference
Figure 4: Server proxy. (EDOC'04), Monterey, USA.
Liu, Y., Ngu, A.H.H., Zeng, L, 2004. QoS Computation
and Policing in Dynamic Web Service Selection. Proc.
Using client and server proxy, the whole 13th International World Wide Web Conference, New
replication is transparent for users and services. York City, USA.
When new replication strategies are implemented, OASIS, 2007. Web Services Business Process Execution
only the proxies have to be enhanced. Language Version 2.0. OASIS Standard,
http://docs.oasis-open.org/wsbpel/2.0/OS/wsbpel-
v2.0-OS.pdf
Osrael, J., Froihofer, L., Weghofer, M., Goeschka, K.M.,
5 CONCLUSIONS 2007. Axis2-based Replication Middleware for Web
Services. Proc. IEEE International Conference on Web
Currently, the architecture as described in chapter 4 Services (ICWS 2007), Salt Lake City, USA.
is finished, and experiments are performed to Salas, J., Pérez-Sorrosal, F., Patino-Martínez, M.,
evaluate the behaviour of the replication framework. Jiménez-Peris, R., 2006. WS-Replication: A
On one hand the experiments should validate the Framework for Highly Available Web Services. Proc.
correctness of the implementation. On the other 15th International World Wide Web Conference
(WWW 2006), Edinburgh, Scotland.
hand (and more important for the ongoing work) the
Thißen, D., Wesnarat, P., 2006. Considering QoS Aspects
evaluations also should help in comparing gains and in Web Service Composition. Proc. 11th IEEE
costs of the replication strategies. These Symposium on Computers and Communications
comparisons are necessary for fine tuning of the (ISCC’06), Cagliari, Sardinia, Italy.
decision rules inside the QoS broker: for which Ye, X., Shen, Y., 2005. A Middleware for Replicated Web
combination of requested parameters which strategy Services. Proc. IEEE International Conference on Web
should by used, and with how many replicas. In Services (ICWS’05), Orlando, USA.
parallel, implementation has started to integrate the Zeng, L., Benatallah, B., Ngu, A.H.H., Dumas, M.,
replication enhancement into composed services. Kalagnanam, J., Chang, H, 2004. QoS-Aware
Middleware for Web Services Composition. In: IEEE
This task is easy because the architecture of the
Trans. Software Eng., Vol.30, No.5.
replication system was oriented at the existing
composition architecture (integration of orchestrator
with client proxy, implementation of server proxy as

160
 AN E-VORTAL FOR THE PORTUGUESE BAKING INDUSTRY
Requirements Model

Jorge Gouveia, Paula Oliveira

Department of Engineering, University of trás-os-Montes e Alto Douro, Quinta de Prados, Vila Real, Portugal
[email protected], [email protected]

João Varajão
Department of Engineering, University of trás-os-Montes e Alto Douro, Quinta de Prados, Vila Real, Portugal
[email protected]

Keywords: e-vortal, Internet, Baking industry.

Abstract: Nowadays, Internet is fundamental to the development of the competitive capability of practically any
industry. Web portals, mainly the vertical portals, have in this context a major role, because they are basic
instruments for the dissemination and search of the information. Despite this fact, it is apparent that no such
Web solution is available in Portugal for the baking industry, in spite of the fact that this is one of the major
industrial sectors in the national context. Believing that the lack of a vertical portal for the baking sector is a
deterrent to its growth, in this paper we present the results from a study developed to identify the main
requirements for a system of this kind.

1 INTRODUCTION conception of an Internet solution (vertical portal),

that gathers information, services and other
The creation of a web portal for the baking industry resources useful to the various intervening parts, by
is, for some years now, a reality in several European identifying its main requirements.
countries and other world markets. The reasons Technological innovations may contribute in
invoked for its creation are most diverse. One of numerous ways to a company’s competitive
them is the easiness and comfort with that the portals advantage, by enabling improvements in satisfaction
supply the access to the information, in a versatile, of demand, cost reduction or quality increase. So,
customizable and personalized way, based on the this kind of portal can act as a basis for new
preferences of the users. Another frequent described economies and be a booster in lifting old economies
reason is the globalization. Globalization causes an to the new market realities (Zirpins et al., 2001).
increase of the competition due to bigger integration Thus, we propose the creation of an e-vortal for
of the markets (Gouveia et al., 2007a). the Portuguese Baking Industry: a portal that
In spite of the fact of the baking industry is one provides to all the community an access point to the
of the major industries in Portugal, the Portuguese information and resources, in a way to potentially
scenario is somewhat different: up until now there is increase the business between the suppliers and the
no such portal (Gouveia et al., 2008). baking industry companies.
The majority of Portuguese food and agriculture In this paper we propose a set of main
companies have an incipient use of information requirements for the development of vertical portals
technologies when compared to their European in baking industry. In section 2 we introduce some
counterparts (SPI, 2001). Though the trend is to fundamental concepts on Internet web portals and
grow, the companies’ use of the Internet to buy and vertical portals and make a brief presentation of the
sell goods and services is relatively negligible (INE, baking industry in Portugal. Section 3 presents
2001, INE, 2003, INE, 2004, INE, 2005). several requirements that should be considered in the
The sector’s lag in this domain, together with the design of new vertical portals for the baking
lack of a dedicated e-vortal, is one of the main industry. The paper ends with some final remarks in
motivations to this work: to contribute to the section 4.

161
ICE-B 2008 - International Conference on e-Business

2 CONTEXT certain locations or with other common interests

(Zirpins et al., 2001).
In the past few years, the Internet and, particularly Given the increasing difficulty of efficiently
the World Wide Web (WWW) has expanded searching the web using keywords, a new trend in
continuously, in terms of the technology used and in the design of website for specific users appeared: a
terms of dimension, becoming an essential way for verticalization phenomenon, i.e., the creation of
business relationships. From this development, portal-like internet websites specially tailored for a
several instruments of research and data specific subject or area of interest, which allows for
organization appeared, in which the Web portals the reduction of portal size without omitting relevant
have a particular relevance (Gouveia et al., 2007). content.
As opposed to general-purpose search engines,
2.1 Web Portals vertical portals have search tools with adequate
strategies and terminology, focused on the target
The concept of Internet “portal” is relatively recent market (Medeiros et al., 2000).
and it is not consensual yet. One can briefly define a A vortal, an abbreviation of “vertical portal”, is a
portal as an integrating point of access to relevant vertical industry, market or specific group, portal on
information, a convergence point for different users, the Internet. Vortal refers to a website that
with a wide range of information. aggregates varied content and services of interest to
The idea of Internet Portal is comprehensive and a particular industry and makes it available to
may include everything from a simple online industry members. While a portal, such as Yahoo! or
catalogue to a complex intranet solution. However, a AOL, attracts a large number of “netizens” (Internet
common characteristic is its role as a starting point, citizens), offering a wide range of contents and links
offering a doorway into using web services (Zirpins to other sites, vortals are narrower in focus and
et al., 2001). address a specific industry, theme, or interest. The
The fast growth of the WWW and the recognized audience or participants may be smaller, but they
importance of Portal Web, led in few years, to the have a higher interest and are highly targeted
incoming of countless portals, many times with (Vortalbuilding, 2005b).
different characteristics among themselves. The concept of a vertical portal is based upon the
Consequently, several authors proposed different premise of linking customers and vendors together
definitions, many times for the same object of within a focused environment. This environment
interest (Gouveia et al., 2007). provides information, services and other resources,
One of the first classification of portals was which encourage members to remain within the
presented in 2002 (Strauss, 2002), dividing portals boundaries of the vertical portal. Vortals are also
into two large groups according to the depth of their seen as business-to-business communities or
content: horizontal portals and vertical portals. business-to-consumer communities (Vortalbuilding,
Horizontal portals are public websites whose goal is 2005b).
to serve its users with a broad set of services and A well-designed and well-developed vertical
resources in an effort to convince them to make the portal can create a snowball effect. Users visit it
website their homepage. On the other hand, vertical because it has quality information, advertisements,
portals provide information, application and other discussion forums, products, contents, friends, etc. A
resources targeted to a specific community or higher number of website users, imply more
interest group. advertisements, products and participation, which in
In 2003, Clarke and Flaherty added two new turn will attract more users to it. Once the users’
dimensions to this classification: not only the depth loyalty is ensured, they will keep visiting and using
of the portal’s content (vertical/horizontal), but also the portal on a regular basis.
its mission (transaction/information) and its target Companies and individuals who share the some
(public/private). Each dimension should be viewed interests may gather to interact, collaborate and
as a continuum, with all portals expressing varying transact on a digital market. In Portugal, such need
degrees of each element (Clarke and Flaherty, 2003). clearly exist for certain markets, as one can infer
Focusing our attention on the “depth of content” from the recent appearing of industry specific
dimension, vertical portals, or “vortals”, offer portals, like the construction industry.
contents and services targeted to a specific domain
or community: professional classes, people from

162
 AN E-VORTAL FOR THE PORTUGUESE BAKING INDUSTRY - Requirements Model

2.2 The Portuguese Baking Industry The FIPA was constituted in 1987 with the aim
to represent and defend the interests of the
According to the Portuguese classification of Portuguese Food-Agricultural Industry nationwide
economic activities (CAE), the Baking and Pastry and in the European Union. This site is dedicated to
Industry is part of group 158 – Manufacturing of every Portuguese Food-Agricultural industry, as
other alimentary products. Group 158 and others well as those that work directly with them. In spite
make the Food, Beverages and Tobacco Industry of not being an association of the Bread Makers
division. sector, it has yet some connection with it. This site
In terms of gross sales, group 158 is the second can be classified as an informative portal. In it one
largest in its division, just behind group 159 may find, essentially, relevant information for to the
Portuguese Food-Agricultural Industry. Another
(Beverages Industry). The Baking and Pastry
case is the official site of the Trade Association
Industry alone (CAE’s class 1581) contributed, in (ACIP). This association presents itself as the largest
2001, with 49% of group 158’s gross sales (INE, Portuguese association of the Bread Making and
2002a). Bakery sector. This site can be classified as an
The relative weight of class 1581 has been institutional one, once it gathers information about
growing since 1998. That year, according to INE the ACIP as well as news and data of its activities.
(INE, 1999), this economical sector represented just So far, this site isn’t providing any further service to
28% of group 158’s total gross sales. the user (Gouveia et al., 2007a).
If gross sales-wise group 158 is relevant in the
food industry context, in terms of employment it is
even the more so. Group 158 – Manufacturing of
other alimentary products – is by far the most
3 AN E-VORTAL FOR THE
important, representing approximately 50.000 jobs BAKING INDUSTRY
in the year 2000.
Thus, the importance of the baking industry in A well-conceived e-portal may create a snowball
the context of the Portuguese food industry is effect: having good contents, forums of discussion,
evident. products and advertising, people will visit the portal.
As many people visit the portal, the larger will be
2.3 Portuguese Baking Industry the number of advertisers, the participations in the
discussion forums, the production of contents and
Internet Portals the possibilities of cooperation and data sharing,
therefore contributing to the development of
Vertical portals design and development is, for some industry.
years now, a concern and a reality in several The bakery sector is one of the most
countries (Maltz, 2005). representative sectors in the Portuguese industry as a
In Portugal the reality is quite different from whole (INE, 2002a).
other countries. The e-vortal concept is not yet well The lack of Web vertical portal hinders the
established, as we can see by the lack of chances of a bigger development for the baking and
implemented e-vortals focused on the Portuguese pastry industry. So, we propose a new e-vortal for
Industry. If we focus our attention on the specific the Portuguese baking industry.
sector of baking industry, then the situation is When one considers the creation of an industry-
extreme. After an exhaustive Internet search we specific portal for a sector such as the baking
concluded that currently there is no an e-vortal for industry, the processes involving the identification,
the Portuguese baking and pastry industry (Gouveia, analysis, negotiation, description, validation and
2006). requirements management assume a paramount
There are, certainly, a few sites that belong to importance. These processes are decisive in the
industrial associations of the sector which aren’t system development and assume a great relevance as
more than an institutional presence in the web. We a factor of success in its construction (Gomes and
can, as an example, refer to the site of the FIPA – Soares, 2004).
Federação das Indústrias Portuguesas Agro- In this section first we discuss the fundamental
limentares (Food-Agricultural Federation of concepts that should be considered in the
Portuguese Industry), ACIP-Associação do development of an e-vortal for the Portuguese
Comércio e da Indústria de Panificação, Pastelaria e baking industry. Next, we present the requirements
Similares (Baking Trade and Industry Association) that a system of this kind should support.
(Gouveia et al., 2007a).

163
ICE-B 2008 - International Conference on e-Business

3.1 Fundamental Concepts experiences, searching for new solutions and

products in partnership, are examples of possible
An e-vortal solution for the industrial sector should collaboration in the baking sector of industry.
be based on five fundamental concepts, as depicted
in Figure 1 (Gouveia et al., 2008): 3.1.3 Interactivity
ƒ Community;
ƒ Collaboration; Internet websites and portals in particular, offer lots
ƒ Interactivity; of communication opportunities in both directions:
ƒ Contents; they offer interactivity. Stuart Brand defines it as a
ƒ E-Commerce. “Mutual and simultaneous activity on the part of
both participants, usually working towards some
goal, but not necessarily” (Brand, 1987). With an e-
vortal, the user should be able to interact with the
system and find the information she or he needs.

3.1.4 Contents
The content of a vertical portal is very important.
The existence of e-vortals is in part due to a huge
need to reduce the size of portals, making search
easier, while no omitting relevant contents. The
search tools can (and should be) improved and they
must provide the specific terminology and strategies
of searching for a sector or subject (Vortalbuilding,
2005a). The users’ fidelity is one of the greatest
challenges that an e-vortal brings up. The quantity
and quality of its contents is an important item to
Figure 1: Fundamental concepts for an e-vortal solution. achieve that fidelity. If we keep the contents always
Source: (Gouveia et al., 2008). up-to-date with a high level of quality, existing users
will remain faithful to the portal and more new users
3.1.1 Community will visit it, which in turn will make the portal even
more attractive to the participation of different
The main aim of an industry vertical portal is to
agents – partners, announcer and collaborators. The
encourage companies and individuals with interests
increase of participation will allow a higher degree
in a specific industry to interact, collaborate and do
of information sharing and higher possibility of
business within a digital market. This virtual
collaboration among partners.
community will allow actors with a common interest
to meet, to share ideas and information, and to know 3.1.5 e-Commerce
each other better, so they can strengthen their
relationship. e-Commerce is the purchasing or sale of goods
and/or services through electronic networks such as
3.1.2 Collaboration the Internet. A platform which will make
relationships and business transactions easier
It is important to distinguish the words
between partners should be provided in an e-vortal
“collaboration” and “cooperation”. Usually we don’t
for the baking industry. This will improve a number
make a distinction between them (Dillenbourg and
of aspects of a business, both for sellers and buyers.
Schneider, 1995). However, there is a difference in
Being one of the most complex sides of a vertical
the way the activity is performed by the ensemble.
industrial portal, we can consider two phases of its
Cooperation consists in dividing tasks among the
development. In a first phase it can promote the
participants, each person or organization being
relationships and transactions among business
individually responsible by part of the problem
partners. So, an e-vortal will allow several aspects of
solution. Collaboration is characterised by the
the business process, for the one who sells, as well
mutual contract of the participants, who work in a
as for the one who buys. The promotion of new
coordinated effort to solve the problem all together
products, the small effort demanded by the support
(Dillenbourg and Schneider, 1995). Sharing
of catalogues, the quick answer to the customers and

164
 AN E-VORTAL FOR THE PORTUGUESE BAKING INDUSTRY - Requirements Model

the possible increase of sales due to the increase of Table 1: E-vortal main requirements.
the market base, are some of the advantages foreseen Function Brief description
for the sellers. Bakers’ Guide This is an area where all companies
The fastest answer for the search of the best (Buyers’ Guide) registered should be listed. Some groups
solutions, the quick access to new suppliers, and can be: Equipment vendors,
Ingredients, Maintenance & Sanitation,
consequently to better prices are, among others,
Packaging, Shipping, Services and
some of the advantages for the buyer. others
In a first stage of the portal for the Baking Documents Publishing and browsing documents and
Industry, one can say that it will improve the papers related with the bakery sector
communication among several agents. The partners Legislation Browsing important legislation for the
baking industry sector
will be able to decide more easily as they share a Useful Useful information like weather and
common platform. information stock market information
For a second stage, it will be reserved the b2b News News about the industry
transactional platform. The implementation of Event Scheduler Browsing and searching on event
automatic processes of proposal requests, the scheduler
Training Courses Here it should be possible to browse all
submission of proposals and the closure of deals are the training courses available. It should
the aspects taken cared by the platform. be also possible to do the registration in
a course available
3.2 Requirements Recipes Recipes browsing and sharing
Classified ads Browsing and creation of small
classified ads
The design of a new Baking Industry e-Vortal On-line Products browsing and online catalogue
should take in account all the concepts referred in Catalogues
previous section. In this section we identify and B2B platform This platform act as a promoter of the
relationships and transactions between
describe some of the function/components such e- partners. Thus, several aspects of the
vortal should support. business processes will be made easier
A study has been made with the purpose of (for buyers and sellers).
investigating different aspects of vertical Internet Requests for additional information
about equipment, products or services,
portals. It was an interpretative and exploratory delivery or terms of payment, are some
study and consisted, in a first phase, on the examples.
development of a conceptual framework and, in a
second phase, on the identification and analysis of As already referred, web portals can be classified
currently existing vertical portals, trying to according to three dimensions (Clarke and Flaherty,
understand their objectives, characteristics and 2003): ): the portal purpose, contents and the level of
functionalities. The needs of the baking industry access provided. The portal one proposes presents
were particularly studied. these characteristics in several levels, as in Figure 2.
Thus, four dedicated e-vortals to the bakery
industry had been analyzed: Bakery-Net
(www.bakery-net.com); Bakery Online Market
Place (www.bakeryonline.com); Federation of
Bakers (www.bakersfederation.org.uk); e Portal da
Padaria (www.portaldapadaria.com.br).
This research made possible the identification of
various critical elements that were systemized in the
form of a set of requirements that are presented in
table 1.
The display of the main characteristics/functions
for an e-vortal solution for the bakery sector comes
from a conjunction of characteristics found in other
portals, as in portal bakery-net.com, with other
completely new, as a result of the analysis of the
sector needs.

Figure 2: The Bakery e-vortal dimension characteristics.

165
ICE-B 2008 - International Conference on e-Business

The e-vortal main requirements presented on services, and the fact that it is dedicated to a specific
table 1 will be detailed and classified according to community with a common interest – The Bread and
these dimensions in the following sub-sections. Bakery Industry. In table 3 one can see the main
characteristics of the proposed e-vortal, classified
3.2.1 Portal Purpose according to the contents.

According to their purposes, the portals can be Table 3: E-vortal main requirements: content.
divided in transactional and informational. The
solution presented cannot be classified, exclusively, Vertical Horizontal
News Weather information
as an informational e-vortal. It will contain Newsletter Stock market information
characteristics mainly of an informational portal Legislation
with updated news of the sector, newsletter, events Event Scheduler
and training courses, updated legislation, among Courses Agenda
others. It will also have transactional portal Online Catalogues
Products and prices
characteristics. This portal will allow costumers comparison
gathering data of products (characteristics, prices, B2B platform
conditions of delivery, availability, etc.), comparing Recipes
prices and, mainly, promote business transactions. In Classified Adds
Table 2 one can see the main characteristics of the Articles and other
documents
proposed e-vortal according to its purpose. Baker’ Guide
Training courses - browsing
Table 2: E-vortal main requirements: purpose. and registration

Transactional Informational
Online Catalogues News 3.2.3 Level of Access
Products and prices Newsletter
comparison A public portal is a portal of which information
B2B platform Legislation and/or services are available for any Internet user
Training courses - browsing Event scheduler and have no access restrictions. Opposite to these,
and registration
the Private Portals have restricted access to a group
Training courses Agenda
Weather information of users.
Stock Market information The bakery e-vortal will give free access of some
Recipes contents for every Internet users. As an example
Articles and other documents there will be the news, legislation and scheduled
Classified Ads events, for example. The Table 4 presents the e-
Baker’ Guide
vortal characteristics according to its users
3.2.2 Content
Table 4: E-vortal main requirements: level of access
provided.
The horizontal portals are those whose contents are
wide. This kind of portals don’t give access to Public Private
specific community services contents, but on the News -> Registred users
contrary, they dispose information, applications and Legislation Newsletter
Event Scheduler Recipes
other resources, through multiple category of users, Stock market information Browse Articles and other
representing every member of a community. documents
The vertical portals can display contents and Weather information
services directed to a domain or specific community. Classified Ads -> Partners
They can be centered in specific professional Training Courses browsing Browse and maintenance of
online catalogues
communities, people from certain places or Baker’ Guide – List of Products and prices
communities with common interests. companies comparison
This portal is clearly a vertical one, due to the B2B platform
particular content and the fact that it is dedicated to a (proposals/additional
information)
well defined community.
Training courses Agenda
The classification of the proposed solution is, in Training Courses –
this dimension, simpler. The bakery portal is mainly Registration
a vertical one, because it focused content and Baker’ Guide – companies
browsing

166
 AN E-VORTAL FOR THE PORTUGUESE BAKING INDUSTRY - Requirements Model

Some of the contents will be of restricted access are not in the platform, can thus buy the lowest cost
to registered users. This restriction will become it or in more advantageous conditions.
two distinct levels: registered user and partner. The
distinction between these two types of users is the
following one: a registered user will have access the 4 CONCLUSIONS
preferential content, such as Recipes and
Newsletters; a partner, for its side, will not only have The technological advances of the last decades gave
access to the previously described preferential origin to an atmosphere where the organizations are
content, but also to all the services related with the
forced to search new options to reduce costs, while
e-vortal b2b functionalities. We can highlight: on- at the same time to compete within their markets.
line catalogue browsing and maintenance, products This atmosphere needs flexible, capable and
and prices comparison, buy or sell on-line, training
competitive organizations, able to make radical
courses, agenda and registration, among others. changes in the way they do business, employ people
and use technologies (Varajão, 2001).
3.3 Discussion To be more competitive or, as we saw, to
survive, companies, whatever is sector is - and the
The expected impact after the implementation, in baking industry is not an exception -, they will have
Portugal, of an e-vortal for the baking industry can to anticipate, or at least to follow, the technological
be analyzed according two complementary aspects. changes that happen everyday, in the national and
The first aspect is about the competition, it will international markets.
allow, among others things, a bigger integration of The Internet allows any company, big or small,
the markets, the multiplication of commercial to easily enter new markets, to conquer new clients,
exchanges and the increase of agreements and to establish relationships with new suppliers, and to
partnerships between companies. It will help the establish new partnerships, without the material,
various actors of the baking industry to create new geographical and time constraints of the
commercial relations, as well as fortifying the conventional way to do business (Varajão, 2003).
existing relations. In the “new economy”, the The difficulty to find references about the baking
information, the knowledge and know-making industry in Portugal through an Internet search, for
becomes raw material, and the possibility of being example using Google, suggests the need for the
able to access it, in useful time, will be one of the development of an Internet solution for this sector
main strategically weapons of the competitiveness. with vertical portal characteristics.
The companies of the baking sector will have at it This kind of initiative, when it is well done,
disposal a tool that will allow them to face these new allows companies with a wider publicity and
requirements. Thus, we contribute for the expansion of their markets, the keeping and
satisfaction of the necessity of modernization of the attraction of new clients thought different and
sector. One another aspect is the resultant advantage innovative ways, a better response to partners and
of the business growth, as much for who sell as for clients, better services, new services available, and a
who purchase. Reduction of administrative costs, cost reduction (in products, services and support).
increase of productivity and modernization of the An e-vortal for the balking sector can have a
market are some of the testimonials given to the direct consequence on the growing of IT integration
press by company CEOs and company owners when in business processes, making companies and
they are mentioned to the impact of the adoption of organizations more in sync with current times and
an e-vortal tool in its company. One of the reasons trends.
pointed in practically all the testimonials is the We intended with this work to give a
increase of the business opportunities. The chance to contribution and to promote the development of this
negotiate with new companies is an immediate important industrial sector. The requirements
more-value of the electronic platforms. As much for presented in this paper should help the design and
the suppliers as for the customers, this kind of development of vertical portals for this sector.
platform, increases the option as much of purchase
as of sale. The suppliers see the number of potential
customers to increase - the prospection of new
customers is facilitated. Customers can easily to
compare prices of some suppliers and, because these
are normally more competitive of what the ones that

167
ICE-B 2008 - International Conference on e-Business

REFERENCES Spi (2001) Estudo de Diagnóstico e Prospectiva e

Caracterização de Perfis Profissionais para
Brand, S. (1987) The Media Lab: Inventing the Future at Levantamento das Necessidades de Formação no
MIT. New York, Viking Press. Sector Agro-Alimentar. Instituto para a Inovação na
Clarke, I. & FLAHERTY, T. B. (2003) Web-based B2B Formação.
portals. Industrial Marketing Management, 15-23. Strauss (2002) Web Portals and Higher Educations.
Dillenbourg, P. & Schneider, D. (1995) Mediating the Tecnologies to Make IT Personal. Educause
mechanisms wich make collaborative learning Quarterly, 32-40.
sometimes effective. International Journal of Varajão, J. E. Q. A. D. S. (2001) Outsourcing de Serviços
Educational Telecommunications, 1, 131-146. de Sistemas de Informação, FCA - Editora de
Gomes, P. V. & Soares, A. L. (2004) Análise, negociação Informática.
e especificação dos requisitos de um Portal web Varajão, J. E. Q. A. D. S. (2003) O Comércio electrónico.
baseados nos princípios de ANT. Actas da 5ª Revista Informática, Associação Comercial e
Conferência da Associação Portuguesa de Sistemas de Industrial dos concelhos de Monção e Melgaço.
Informação. Vortalbuilding (2005a) Why vortal?
Gouveia, A. J., Oliveira, P., Gouveia, F. & Varajão, J. Vortalbuilding (2005b) What is a vortal?
(2008) E-vortal development framework for the Zirpins, C., Weinreich, H., Bartelt, A. & Lamersdorf, W.
Portuguese baking industry. Iadis International (2001) Advanced Concepts for Next Generation
Conference e-Society 2008. Algarve. Portugal, Iadis. Portals. University of Hamburg, Department of
Gouveia, A. J., Oliveira, P. & Varajão, J. (2007) Portais Informatics, Distributed Systems Group (VSYS).
Web: Enquadramento Conceptual. Conferência Ibero-
Americana www/Internet 2007. Vila Real. Portugal,
Iadis.
Gouveia, A. J., Oliveira, P. & Varajão, J. (2007a) Estudo
de Soluções Internet para o sector da panificação: a
situação em Portugal. ColleECTeR Iberoamericana
2007 - Collaborative Electronic Commerce
Technology. Cordoba. Argentina, ColleECTeR
Iberoamericana.
Gouveia, A. J. G. D. (2006) PortalPan - Um Contributo
para o Desenvolvimento da Indústria da Panificação.
Departamento de Engenharias. Vila Real,
Universidade de Trás-os-Montes e Alto Douro.
Ine (1999) Grupo 158 - Fabricação de outros produtos
alimentares - 1992 a 1998. Estatísticas Agro-
Alimentarares. Instituto Nacional de Estatística.
Ine (2001) Inquérito à Utilização das Tecnologias da
Informação e da Comunicação nas Empresas.
Sociedade da Informação e do Conhecimento.
Instituto Nacional de Estatística.
Ine (2002a) Grupo 158 - Fabricação de outros produtos
alimentares - 1999 a 2001. Estatísticas Agro-
Alimentarares. Instituto Nacional de Estatística.
Ine (2003) Inquérito à Utilização das Tecnologias da
Informação e da Comunicação nas Empresas.
Sociedade da Informação e do Conhecimento.
Instituto Nacional de Estatística.
Ine (2004) Inquérito à Utilização das Tecnologias da
Informação e da Comunicação nas Empresas.
Sociedade da Informação e do Conhecimento.
Instituto Nacional de Estatística.
Ine (2005) Inquérito à Utilização das Tecnologias da
Informação e da Comunicação nas Empresas.
Sociedade da Informação e do Conhecimento.
Instituto Nacional de Estatística.
Maltz, L. (2005) Portals: A Personal Door to the
Information Enterprises. Educause Quarterly.
Medeiros, R. S., Muniz, J. S. & Varela, L. B. (2000)
Portal, Vortal ou Hortal? In Fluminense, U. F. (Ed.)
Informe NDC Universidade Federal Fluminense.

168
 ON EXPLORING CONSUMERS’ TECHNOLOGY FORESIGHT
CAPABILITIES
An Analysis of 4,000 Mobile Service Ideas

Petteri Alahuhta, Pekka Abrahamsson

VTT Technical Research Centre of Finland, Kaitoväylä 1, Oulu, Finland
[email protected], [email protected]

Antti Nummiaho
VTT Technical Research Centre of Finland, Vuorimiehentie 3, Espoo, Finland
[email protected]

Keywords: Mobile service, technology foresight, user innovation, open innovation, lead-user innovation.

Abstract: Lead user driven innovation and open innovation paradigms seek to involve consumers and common people
to innovative product development projects. In order to help developers choose ideas that meet the end
users’ needs, we undertook a massive collaborative research effort and collected 40,000 ideas from 2,150
common people about future mobile services that they would like to use. We inspired each people to
produce tens of mobile service ideas. In this paper we carry out an analysis for 4,000 ideas from the idea
database. We had a particular interest in whether peoples’ ideas can be used in foreseeing the technology
development needs. The results show that end users produce ideas that are conservative more than novel.
Therefore, we claim that consumers’ technology foresight horizon is limited by the existing technological
base. The second finding, linked to the previous one, is that the great majority of the ideas that consumers
expressed could be realised utilizing existing technologies. The implication of this finding is that the idea
database should be an interesting source of ideas for service developers. The third finding of the study,
related to the methodology, is that a vast number of ideas can be collected fairly easily but analyzing them
cost effectively is a challenge.

1 INTRODUCTION Von Hippel (1986, 2005) has been one of the

first authors to promote the use of consumers’
End users of the products and services are technological foresight capabilities to the fullest
instrumental for the success of any innovation. They extent. In his seminal work on lead user driven
will determine which products and services will innovation in 1986, he argues that in the area of
ultimately become successful and which ones will high-tech product development lead users are
fail. The end users’ market behaviour is difficult to actually ahead of the market place and possibly
forecast, and indeed history has shown that markets possess innovative ideas about the product or
and end users may take unexpected turns. As an technology that they use. Von Hippel (2005) argues
example, in the mobile service segment the high also that lead users may be quite willing to share
popularity of SMS (Short Message Service) led the their product related innovations freely and that
operators to forecast that an MMS (Multimedia these ideas are often commercially attractive.
Messaging Service) offering would become an Chesbrough (2003) popularized the concept of
instant hit among mobile phone users as soon as the open innovation to distinguish from the traditional
new technology was made available. For several model of innovation, which he calls a closed
reasons, the adoption rate of MMS has been innovation model. Chesbrough (2003) argues that
significantly lower than the SMS service and in external R&D (Research and Development) can
majority of the mobile service markets the SMS still create significant value and the internal R&D is
keeps dominating the service offering despite of the needed to claim a part of this value. His open
opportunities that the newer technology offers. innovation model relies on building the business
model first. He argues that companies should learn

169
ICE-B 2008 - International Conference on e-Business

to increase the effective usage of a company’s IP attention to identifying possible technology related
(Intellectual Property) either by selling it or offering obstacles in front of successful mobile services, such
it outside of the company’s boundaries by other as limitations in human technology interaction or in
means. Also, learning to acquire IP from companies, communication capabilities. As context awareness
individuals and user communities (Baldwin, has been advertised as a technology to bridge the
Hienerth & Von Hippel 2006) outside, whenever it shortcomings of the mobile HCI (Cheverst et al.
fits the business model, bears significance in 2000, Korpipaa et al. 2006), we also analyzed to
producing innovative products and services. In what extent the ideas suggest that this would be a
Chesbrough’s (2003) terms the open innovation desired development trend.
model forms a new imperative for creating and The paper is organized as follows: the next
profiting from technology. section outlines the research design of the study
Based on Von Hippel’s and Chesbrough’s focusing on research methods, data collection and
concepts of lead user driven innovation and open analysis. Also, the outline of the Idea Movement
innovation, the involvement of consumers in the initiative that the present researchers took alongside
development of future mobile services would seem with the principal high level results of the Idea
to be of paramount importance to guarantee the Movement itself is presented. This is followed by
continuous feed of innovative products and services. the results of the analysis and a brief discussion of
Little is known, however, to what extent the end the implications for the mobile service business and
users are capable of forecasting into the future (Un, research. The paper is concluded with final remarks.
Price 2007). Thus, the potential technological
foresight capability of a mobile terminal user
remains an area of research with little results. 2 RESEARCH DESIGN
When asked about future technology, we have a
tendency to rely on forecasts made by information This section discusses the issues related to research
and technology research companies such as Gartner design. We will first introduce the research method
(www.gartner.com) and Standish Group and settings, and then we will briefly describe the
(www.standishgroup.com/). Their predictions on data collection and means of analysis of the material.
future service development, builds upon technology. Finally, we will give some examples of collected
Markets and end-users are known to take unexpected ideas stored in the database.
turns. They may be even reluctant to adopt new
technologies or services. behaviourTherefore it can
be asked, how accurately are technology research
2.1 Research Method and Settings
companies able to predict future services, as the
This study is an explorative research on the mobile
diffusion of services is dominated largely by other
service needs. The phases of an innovation process
factors than those related to any particular
have been discussed in the literature (e.g., Koen et
technology.
al. 2002). There is a strong need to better understand
In order to help companies and developers meet
the role and potential of end users in foreseeing the
the needs of the growing number of mobile terminal
technology development needs. The problem has
users, we undertook a massive collaborative research
also been discussed by Un & Price (2007). In order
effort and collected 40,000 ideas from 2,150
to explore the possibilities of ideas, we launched a
ordinary people about future mobile services that
national research project called the Idea Movement.
they would like to use. We inspired people not to
The project started in the beginning of 2006. We set
think about the technology, but to express their
the goal to collect at least 35,000 ideas for mobile
concrete needs. Moreover, we did not request a
applications straight from the citizens, and to make
single or few ideas from these people. Rather, we
these ideas accessible for everyone by publishing
motivated people to come up with 30 or more ideas
them in the Internet. Giving companies,
even if they would feel them to be either “silly” or
organizations and individuals the opportunity to
not realistic in their minds.
build on the ideas of thousands of people, we aim to
In this paper, we report the results from a study
accelerate the development and commercialization
where 4,000 ideas from the database were chosen for
of new mobile services.
a technical scrutiny. We are particularly interested in
understanding the technical foresight capability of
an end user. We also wanted to know if by analyzing
2.2 Data Collection and Analysis
users’ ideas, one could identify specific technology
Ideas were collected systematically in 31 workshops
development needs. Therefore, we paid particular
and events organized around Finland. Altogether

170
 ON EXPLORING CONSUMERS’ TECHNOLOGY FORESIGHT CAPABILITIES - An Analysis of 4,000 Mobile
Service Ideas

2,150 people participated in the events. The majority identified. Furthermore, we explored what kinds of
of the participants were university students, integrated or interconnected devices were proposed
schoolchildren and elderly people. Workshops were to be part of mobile services. The analysis was
also organized in schools, at workplaces and even in carried out by reviewing a selected portion of the
a shopping centre. Furthermore, there was a ideas and categorizing them based on the defined
possibility to post ideas on the Idea Movement categories. Each idea was classified based on its
website, and about 200 people communicated their most obvious way of implementation. For this paper,
ideas in this way. Table 1 summarizes the division we selected 4,000 ideas (i.e., 10% of all ideas) for a
of ideas according to different age groups. detailed inspection. The selected dataset consists of
2,000 ideas produced by high-school students (the
Table 1: Distribution of ideas in the Idea database. age group of 17-18 years) and 2,000 ideas from
university students and employees (the age group of
Age group number of Proportion of
(years) ideas ideas
20-50 years).
7-12 1,800 4%
13-16 6,500 16% 2.3 Data Examples from the Idea Base
17-19 4,300 10%
20-50 21,000 51% The themes of covered a large variety of different
50-90 4,300 10% topics including Public transportation, Travelling,
Unclassified 3,100 8% Work and Learning, Culture and entertainment,
Total 41,000 100% Hobbies, Wellbeing and health, Shopping and
service, Family, Friends and relatives, Household
Each workshop was kicked off with a short management and living and Everyday activities.
introduction to idea generation techniques, followed In order for the reader to comprehend the nature
by brainstorming sessions both individually and in and type of the mobile service ideas, we have listed
groups. At the beginning of each event we asked the below some ideas classified to the four suggested
participants to produce ideas about a given topic, but categories: Excellent ideas, Interesting ideas,
we also accepted ideas that did not match the Conventional ideas and Not a mobile service idea.
original subject of the particular brainstorming These categories were created in an early phase of
session. We instructed each participant to produce the analysis in order to quickly have an exploratory
20 ideas individually and then form groups of 3-4 view of the ideas in the database.
people. The ultimate goal of idea generation was that Examples of ideas classified as Excellent are:
each group would deliver together 100 ideas, or • Warning if parking time is running out.
more. Possibility to automatically buy or order
Ideas are 1-2 sentence descriptions of a mobile additional parking time.
service idea or an expression of a need that they • Location-based filtering of incoming calls. E.g.,
think could be fulfilled utilizing mobile technology. No work-related calls at home.
The progress of a brainstorming session has been • Locating of friends (if they allow).
described in more detail by Leikas (2007). Examples of Interesting ideas are:
Idea analysis was carried out in two phases. First • In a rally, one can get information about the
a group of five reviewers did a qualitative analysis condition of a car. A spectator will know if the
of 2,000 ideas so that we got an understanding of the car is going to break down.
potential of the ideas. According to the reviewers’
subjective opinion, each idea was classified into one • Service that tells you how strong the punch in
of the following four categories: Excellent, your glass is.
Interesting, Conventional and Not a mobile service • Opportunity to order "good explanations" when
idea. coming home late.
We also carried out an analysis of the technical
aspects of the collected ideas in order to get an Some ideas classified as Conventional are:
understanding of the technologies that are required • The controlling of a home automation system
for implementing these ideas. Also, the essential by a mobile handset. Being able to switch on/off
constraints that may currently prevent the lights, heating, etc.
implementation were considered. • Message from library if a new book by your
For the analysis three technology areas (access favourite author is available.
technology, context information, human-technology
interaction) and major technical constraints were • Service for getting current hit music to your
mobile phone.

171
ICE-B 2008 - International Conference on e-Business

Finally, examples of ideas that are not a mobile advertisement. Locating missing objects and persons
service idea include: representrepresented 9.2% and communicating, for
• Mobile phone making a cup of coffee example, with social communities 7% of ideas.
• 3" nails inside the mobile phone Service requests (4.7%), mobile content production
(4.3%), payment (4%) and identification (3.4%)
• Rat trap representrepresented a perhaps surprisingly low
popularity in the analyzed set of ideas.
24% of the ideas fell into the category others. We
think this indicates that there is a versatile set of
3 RESULTS activities in everyday life where mobile technology
could provide value to the users.
In this section we will describe the results of the
technical analysis of 4,000 ideas. At first the 3.2 Technical Constraints
application categories are introduced, followed by
technical constraints in five major areas suggested In this paper we want to explore the nature of user
by the literature. innovation ideas from the technical viewpoint. We
are particularly interested in what kinds of
3.1 Application Categories technologies are needed to realise the ideas that
users have proposed. More specifically we are
We first classified mobile service ideas into eight interested in
distinct categories based on the idea clusters a) what seems to be the main constraints for services not
identified in the data and had one category for being developed,
miscellaneous ideas. The results have been b) what communication and access technologies are
summarized in Table 2. required for the ideas,
c) what kind of context information, if any, is required
Table 2: Distribution of ideas in application categories. for services that users propose,
d) what are the challenges in Human Technology
Category Description Total
Interaction (HTI) technologies, and
Information pull Retrieving information for 30%
e) what kinds of integrated or interconnected devices are
some purpose (possibly
required for the ideas.
based on location).
Information Receiving information 14% 3.2.1 Analysis of Major Constraints for
push automatically (possibly
Adoption of Mobile Services
based on location).
Locating Locating or following 9.2% One of the main findings of our technical analysis is
(persons / some (nearest) person or that there are no major technical constraints in
objects) object.
developing most of the mobile service ideas into
Communication Social discussion channel. 7.0%
actual mobile services for people. However, we can
Service request Ordering a personal 4.7%
pinpoint some constraints that are quite typical for
service (possibly based on
mobile services and well known by the literature.
location).
Even if these constraints are not immediate road
Content Producing content. 4.3%
blocks for services, they weaken the user experience
production
so much that eventually users may not start using the
Payment Using mobile device as a 4.0%
means of payment.
services in the first place or they stop using the
Identification Using mobile device as a 3.4%
services because of the poor experience.
identification device. In Table 3 technical factors that may hinder user
experience have been summarized. Small screen size
Others Applications that do not fit 24% and low bandwidth are perhaps the most important
into other categories. constraints in mobile services. Other factors that
make it difficult to realise some of the ideas are low
According to technical analysis people see processing power, high power consumption of the
mobile services largely as an information channel, terminals and a limited amount of memory available
which can be used when ever they need to know in the terminal.
something. 30% of ideas fall into this category. Some of these constraints will be overcome in
muchSignificantly fewer ideas proposed automatic time. Such , such as bandwidth, processing power
information delivery to users’ terminals. Only 14% and amount of memory. Screen size and power
of the ideas proposed push services such as

172
 ON EXPLORING CONSUMERS’ TECHNOLOGY FORESIGHT CAPABILITIES - An Analysis of 4,000 Mobile
Service Ideas

consumption, however, are harder obstacles for Table 4: Required access technologies.
developers. The trend towards extensive multimedia Category Description Total
communication requires as large a screen as possible 3G Text and multimedia 67%
and higher processing power that leads to higher based services.
power consumption. 2G Mainly text based 60%
services.
Table 3: Major technical constraints identified in collected Short-range Communication over a 13%
mobile service ideas.
communication few meters or by touch.
Category Description Total (Bluetooth, NFC)
Screen The user interface requires a 3.5% WLAN Mainly indoors or in areas 1.5%
size large screen. with lot of people.
Bandwidth A lot of network traffic occurs. 2.9% Broad-casting Sending same information 1.1%
Processing Processing power is especially 1.0% to all cell phones.
power important. Satellite Worldwide access may be 0.47%
Amount of A lot of multimedia or other 0.96% needed everywhere
memory memory expensive data is including mountains and
handled. seas.
Battery A lot of power is needed. 0.91% Combination of Non-trivial combination 0.44%
duration two or more of multiple access
Touch The user should be enabled to 0.32% access technologies.
screen interact with the application by technologies
touching pictures or words on Cell-casting Sending same information 0.22%
the screen. to all cell phones within a
Keyboard The user must type lots of text. 0.30% base station.

It is also to note that the total percentage of ideas 3.2.3 Requirements for Human Technology
having major technical constraints is very low. In Interaction (HTI) Technologies
other words, technology is not the main constraint in
implementing most of the ideas. Earlier in this paper we discussed the major
technical constraints when considering the
3.2.2 Analysis of Required Communication implementation of mobile service ideas.
and Access Technologies
Table 5: Human Technology Interaction technologies
In the technical analysis we wanted to consider what required in mobile service ideas.
communication and/or service access technologies
are required for realizing the ideas. Table 4 Category Description Total
summarizes the analysis of required access Image / Video Extracting information from
1.8%
analysis image or video.
technologies.
Speech Ability to produce sound
In the table we can see that the vast majority
synthesis that resembles human 0.86%
(67%) of foreseen services can be realizedrealised
speech.
utilizing existing access technologies such as
Augmented The idea that an observer's
GSM/GPRS (2G) or 3rd generation cellular networks
reality experience of an
(3G). 13% of ideas required short-range environment can be 0.76%
communication solutions, such as Bluetooth or Near augmented with computer
Field Communication (NFC). Quite few services generated information.
really required such communication technologies as Speech The ability to recognize and
satellite communication, cell-casting or a recognition carry out voice commands 0.69%
combination of different communication or take dictation.
technologies. Audio analysis Extracting information from
During the time of collecting the ideas (in 2006) 0.49%
audio.
there were quite lively debates about the promised Gesture The ability to recognize
break-through of mobile-TV. Our study did not recognition human gestures, usually 0.22%
support the need of these technologies and it seems hand motion.
that users have not adopted to broadcasting-type Gait pattern The ability to recognize
0.02%
access technologies beyond FM-radio-receivers in recognition gait.
their mobile handsets.

173
ICE-B 2008 - International Conference on e-Business

HTI-technologies do not seem to be a major Table 7: Needs for integrated or interconnected devices.
obstacle in the track of creating mobile services. Category Description Total
However, we know that poor usability and Meters/Sensors Measuring a property. 6.8%
restrictions in HTI-technologies weaken the user
Controllers Controlling a device. 3.3%
experience (Hartmann, Angeli & Sutcliffe 2008). Others Mobile phone being or 9.9%
Developers are investing lots of effort creatingto using other devices.
create solutions thethat make mobile interaction
more intuitive. We wanted to look at a number of In the analysis, we identified two clear
HTI-technologies in order to see if end users categories, namely Meters/Sensors and Controllers.
generating service ideas are in need of novel HTI The first category stands for ideas that have some
technologies. Table 5 summarizes the analysis. kind of sensing or measuring functionality. Ideas in
Based on the analysis we can say that users this category include, for example, a mobile phone
either can not imagine or do not seem to hunger for with a step counter or an alcometer. The second
services with novel HTI capabilities, such as speech category includes ideas related to controlling other
synthesis, augmented reality, speech technologies or devices using a mobile phone, for example, the
gesture recognition. possibility to open a home door for someone from
distance using a mobile phone. In addition to these
3.2.4 Analysis of Context Information categories, we found a versatile set of ideas
Required for Services Ideas Users proposing that a mobile phone includes or uses
Proposed another device, such as a laser pointer or a projector.

The use of context information in mobile services

has been an active research discipline for already a
decade. Researchers and developers have proposed a 4 DISCUSSION
number of different context-aware mobile services
and applications. Literature claims that people may have unique,
interesting and potentially commercially attractive
Table 6: Context Information required in mobile service ideas about technologies (Von Hippel 2005).
ideas. Literature also holds that bright ideas may emerge
from inside an organization as well as from the
Category Description Total
outside (Chesbrough 2003). We challenged these
Location Location of the user or some 25%
arguments and collected a vast amount of mobile
object.
service ideas from students, working-age people,
Time Time that is relevant in a 4.7%
non-trivial way.
school-children and elderly people. We argue that an
Activity What the user is doing. 1.8%
in-depth analysis of this material will provide us
Identities of Identities of people that are 1.8%
valuable insight about users’ everyday needs and
nearby people close-by. wishes concerning mobile and ubiquitous
Emotion / How the user or some other 1.6% technologies and services.
Mood party feels. In this paper we were particularly interested if
user-generated mobile service ideas could be used as
Our analysis, summarized in Table 6, confirms a tool in foreseeing the need for technological
the fact that location seems to be the most valuable development. The reasoning behind this idea is as
and versatile context information to be used in follows: if users would propose needs or ideas that
mobile services. Even 25% of service ideas require cannot be realised using existing technologies, it
location information. Time (4.7%) is another would create a potentially attractive target for
obvious context information. Identifying the activity technology developers.
of the user, the social context (people nearby) or an The discussion is organized around three
emotional situation do not seem to be as important practical implications of the study. First, we claim
factors for the proposed mobile service ideas. that it is relatively simple to collect a large amount
of data about people’s ideas or needs. The real
3.2.5 Analysis of Needs for Integrated or challenge is to analyze these ideas cost effectively.
Interconnected Devices Second, we claim that the adoption of a vast variety
of mobile services is not primarily limited by
We also wanted to look at what kind of integrated or
missing technologies. In fact, the majority of all
interconnected devices people wanted to have in
ideas can be realised with existing technologies. And
their mobile terminals. The results are summarized
third, we claim that people are not likely to propose
in Table 7.

174
 ON EXPLORING CONSUMERS’ TECHNOLOGY FORESIGHT CAPABILITIES - An Analysis of 4,000 Mobile
Service Ideas

very futuristic service ideas. On the contrary, they (mobile phone) with a browser or a dedicated
are tightly limited to existing understanding of the application, wireless network access and a server
capabilities of mobile systems. We could say that with service specific functionality.
people are hindered by the technological frame of According to our analysis only 5% of proposed
today. mobile service ideas had major technical obstacles.
One such technical obstacle is, for example, small
4.1 Consumers Freely Share their screen size, which is particularly problematic in
Ideas navigation and multimedia services. Another
technical challenge is limited bandwidth of mobile
During the course of the Idea Movement project we access. Bandwidth limitation causes troubles in
have seen that collecting a large amount of idea multimedia intensive services such as video-
material is not an overwhelming task. People were conferencing and multimedia streaming. Power
willing to participate and reveal their ideas in the consumption and battery technology may also limit
brainstorming sessions that we organized in various possibilities in few service ideas.
schools, universities and companies. In these events The development of technologies takes some of
we offered participants refreshments, such as coffee the current obstacles into history. Some of the new
and snacks, but the participants were not offered any mobile terminals from various vendors have much
financial or material rewards for participation. This larger displays than a couple of years ago. Also, new
seems to confirm the claims that economic factors network technologies are provided with improved
are not the primary source of motivation for people bandwidth and the computing power will increase,
who share their ideas (Lüthje 2004). In the as well as the size of memory.
beginning of the process of collecting these ideas we In fact, we could argue that all proposed mobile
thought that IPR and ownership of the ideas would service ideas could be realised using existing
become an important topic of discussion. technology, but the user experience and cost
Surprisingly, only a few people raised the issue. structure might not be quite satisfactory for
Perhaps this was due to the policy to publish all the commercial deployment.
ideas, and the fact that these ideas are mainly very
short descriptions of a need or service – not detailed 4.3 Consumers’ Technology Foresight
business ideas. is limited to Existing Technologies
Collecting a large amount of ideas is not as and Paradigms
difficult as it may sound. Instead (cost) effective
processing of these ideas is much greater challenge. While we suggested that approximately 95% of all
We have tried various automated systems for ideas can be realised with existing technologies, it
analyzing the data, but they do not seem to work can be said that this method and project produced a
very well. This is mainly due to the free format of large amount of valuable data for companies in quest
ideas. The same idea can be expressed in various of new services right now. Even though researchers
ways with different kinds of language, i.e., using and developers may be interested in technology
standard language, dialect, spoken language or even forecasts, this material may not readily reveal the
slang. future developments of mobile technologies..
An interesting avenue to pursue for processing Instead, we suggest that extensive pre-processing
these ideas is to distribute the workload to large may enable the discovery of novel patterns not yet
groups of people using the Internet. In this approach identified. We should try to identify some “weak
persuading and motivating people to work on the signal” phenomena from the set of ideas instead of
ideas requires further consideration. We have done stronger trends. These stronger trends tend to be the
some experiments on carrying out Internet based conventional ideas that are familiar to everyone.
analysis, but the results fall out of the scope of this Conservative ideas may be due to the lack of
paper and thus will be published in the future. consumers’ understanding on the new possibilities
of mobile technology or that the majority of users
4.2 95% of the Ideas can be are so tied with their current context that they cannot
Implemented imagine new paradigms and revolutionary ideas.We
are far from being disappointed with the evidently
In our study it became evident that the vast majority short technology horizon of people. When we
of mobile service ideas can be implemented using consider these ideas from a business point of view,
existing technologies. Technical components of we see that the ideas deemed Conventional represent
mobile services are typically a mobile terminal needs of a large group of potential customers that

175
ICE-B 2008 - International Conference on e-Business

have not been met by the service providers. We see Movement, therefore, makes a link between the
potential breakthrough opportunities by new intellectual capital of ordinary citizens and the
offerings formed from thousands of raw ideas. technology business know-how of companies which
We were also expecting to see more ideas related results in a potential win-win situation.
to new developments in the Internet such as social
networking and the creation of mobile user-
generated content. Another topic that we expected to ACKNOWLEDGEMENTS
see more was commerce, payment and identification
of users using mobile technology. Both of these We acknowledge Idea Movement Partners, Maaretta
developments are active in the Internet, but
Törrö and Anssi Öörni.
participants of these events did not see the need for
carrying out these tasks with their mobile systems.
Many of these services, however, require that
new revenue-models should be defined, which REFERENCES
would enable the penetration of mobile services
based on the content and interest rather than merely Baldwin, C., Hienerth, C. & Von Hippel, E. 2006, "How
on data traffic costs. user innovations become commercial products: A
theoretical investigation and case study", Research
Policy, vol. 35, no. 9, pp. 1291-1313.
Chesbrough, H.W. 2003, Open Innovation: The New
5 CONCLUSIONS AND FUTURE Imperative for Creating and Profiting from
Technology, Harvard Business School Press.
WORK Cheverst, K., Davies, N., Mitchell, K., Friday, A. &
Efstratiou, C. 2000, "Developing a context-aware
The 40,000 ideas for new mobile services expressed electronic tourist guide: some issues and experiences",
by the participants are far from well-defined Proc. of CHI '00 ACM, New York, NY, USA, pp. 17.
business concepts ripe for commercialization. The Hartmann, J., Angeli, A.D. & Sutcliffe, A. 2008, "Framing
ideas still require further development and the user experience: information biases on website
professional elaboration before they can be quality judgement", Proc. of CHI '08 ACM, New
introduced to the market as products or services. York, NY, USA, pp. 855.
Koen, P.A., Ajamian, G.M., Boyce, S., Qamen, A., Fisher,
In this study we presented the findings based on
E., Fountoulakis, S., Johnson, A., Puri, P. & Seibert,
a technical analysis of 4,000 ideas. The findings R. 2002, "Fuzzy Front End: Effective Methods, Tools,
were grouped into three principal practical and Techniques", The PDMA ToolBook 1 for New
implications: 1) Consumers share their ideas freely, Product Development.
2) vast majority of the ideas can be implemented by Korpipaa, P., Malm, E.J., Rantakokko, T., Kyllonen, V.,
means of existing technologies, and 3) consumers’ Kela, J., Mantyjarvi, J., Hakkila, J. & Kansala, I. 2006,
technology foresight appears to be restricted by their "Customizing User Interaction in Smart Phones",
experience with current technology and paradigms. Pervasive Computing, IEEE, vol. 5, no. 3, pp. 82-90.
In order to advance this development we plan to Leikas, J. 2007, Idea Movement of Aging Citizens:
Lessons-Learnt from Innovation Workshops.
carry out a set of analyses for the idea database. One
Lüthje, C. 2004, "Characteristics of innovating users in a
particular topic that we are interested in is consumer goods field: An empirical study of sport-
comparing the differences in ideas of people of related product consumers", Technovation, vol. 24, no.
different age. We want to explore this further in 9, pp. 683-695.
future studies. We will continue to expand the Idea Un, S. & Price, N. 2007, "Bridging the gap between
Movement to other countries, cultures and technological possibilities and people: Involving
nationalities in the near future. The Idea people in the early phases of technology
Movement’s (www.idealiike.fi) idea bank is now development", Technological Forecasting and Social
open in Finnish for all commercial, research and Change, vol. 74, no. 9, pp. 1758-1772.
Von Hippel, E. 2005, Democratizing innovation, MIT
educational purposes. The ideas will be opened up in
Press, Cambridge, Mass.; London.
English as well. This enables the development of Von Hippel, E. 1986, "Lead Users: A Source of Novel
these ideas in global context. Product Concepts", Management Science, vol. 32, no.
We predict that the actual value of these ideas is 7, pp. 791-805.
generated in the idea refinement phase where
hundreds of ideas are combined and enriched
through the conceptualization process. This is a task
reserved for companies and organizations with the
ability to efficiently commercialize the services. Idea

176
 AN ANALYSIS OF CONTEXT-AWARENESS
IN COMMERCIAL MOBILE SERVICES

Ana M. Bernardos, Daniel Marcos and José R. Casar

ETSI Telecomunicación, Universidad Politécnica de Madrid, Ciudad Universitaria s/n, 28040, Madrid, Spain
{abernardos, dmarcos, jramon}@grpss.ssr.upm.es

Keywords: Mobile data services, context-aware applications, location based services, mobile business.

Abstract: This contribution aims at analyzing context-awareness from a commercial point of view, studying how the
utilization of context descriptors and features in mobile services has evolved during the period 2003-2008.
The analysis is based on the information collected during a Technology Watch activity; this methodology
has provided us with a large database of mobile services, built from monthly updates of mobile novelties
and commercial launches. Services in the database have firstly been categorized regarding their
functionalities, in order to get the big picture of the mobile ecosystem in this period of time. Afterwards, we
have identified a list of descriptors (personal, physical or activity related) and features (related to resources
discovery, management and communications and also to advances HCIs) which are usually identified in
context-aware applications and systems prototypes. The use of these descriptors and features has been
evaluated for each service and some trends have been detected. Our general conclusion is that nowadays
few commercial mobile services can be considered “context-aware”, although isolated features (mainly
related to personalization) are perceived in many applications. Due to their functionalities, location aware
services and mobile social networks are leading the use of context parameters.

1 INTRODUCTION of use study on mobile services. Both types of

analysis are already common in (academic and
Simultaneously with the evolution of commercial corporate) literature. With respect to forecasting,
mobile technologies and markets, a lot of research there are some broad-scope proposals that aim at
has focused on bringing to reality the paradigms that predicting the penetration and the traffic in cellular
are supposed to sketch the future of mobility (Mohr, mobile networks –eg. (Arvidsson et al., 2007)- and
2008). Ubiquitous and pervasive computing (Weiser also studies focused on niche services or particular
et al., 1999) inspire the concept of ambient markets –eg. (Funk, 2007). On the other hand,
intelligence (ISTAG, 2001), a smart and sensitive mobile services adoption is attracting great interest:
scenario where context-aware applications for example, Kelleher (2007) analyzes four studies
(personalized and status adaptive services) make on this issue, Bouwman et al. (2006) goes deeper on
easier everyday activities (Schilit and Theimer, barriers and drives that condition the services’ use
1994). and Bina and Giaglis (2005) and Gilbert and Han
To date, context-awareness has inspired many (2005) analyze users’ preferences taking into
research prototypes in different application fields account their life styles, needs and demographical
(museums, airports, hospitals, smart homes, etc., see characteristics. Verkasalo (2007) studies the users’
Section 2). From an analysis on the evolution of preferences from the analysis of the data traffic
commercial mobile services during 2003-2008, this generation of a group of representative applications.
contribution aims at shedding some light on how Using a complementary perspective to mobile
context-aware research concepts and features are data adoption studies, our contribution is firstly
being transferred into market applications. Our final conceived to analyze the generation of mobile
motivation is to identify the hindrances (technical, services. With this purpose, during the last five years
privacy and business related issues) which need to (2003-2008), we have followed a Technology Watch
be overcome to generalize the use of context methodology which has provided us with a monthly
information in commercial mobile services. updated data base of new mobile services. In this
It is important to remark that this is neither a contribution, we elaborate on the long series of
forecasting work nor a study on adoption and pattern

177
ICE-B 2008 - International Conference on e-Business

gathered data in order to show how mobile services combined with context-awareness to achieve a
and context-awareness presence has evolved. new user experience.
The paper is organized as follows. Section 2 - Productivity applications for working spaces
goes deeply into the concept of ‘context-awareness’ such as classes, campuses (e.g. ActiveCampus,
and gathers a review of its areas of application. Classroom 2000 or eClass projects) and offices
Section 3 presents the methodological approach used (e.g. Context-Aware Office Assistant).
in this study. From the empirical analysis, Section 4 - Fieldwork applications for environments where
provides a view of the mobile ecosystem, to frame recording and filtering data are prior tasks, such
Section 5, which elaborates on the level of “context- as laboratories (e.g. Labscape) or archaeology
awareness” that current mobile applications have. areas (e.g. the Context-aware Archaeological
Section 6 concludes the paper with some open issues Assistant).
for the generalization of context-awareness in - Smart homes are highly pervasive scenarios
commercial mobile services. with sensors, actuators, wireless networks, etc.
(Meyer and Rakotonirainy, 2003); in particular,
there is a growing interest in applications aimed
2 FROM LOCATION BASED at supporting daily living activities and well-
being (especially of elderly and disabled people)
SERVICES TO (e.g. Wireless Wellness Monitoring and Howel
CONTEXT-AWARE projects).
APPLICATIONS - Health care environments, such as hospitals -
Bricon-Souf and Newman (2007) contains a
From the late nineties on, location has been the survey- or even operating rooms (Agarwal et
enabler of a number of commercial “precontext- al., 2007).
aware” services, such as family finders, location - Horizontal applications such as memory aids,
based advertising, area billing, pervasive games, location annotation software (e.g. GeoNotes) or
trackers, real time location systems (RTLS), etc. context-aware telephony (callers are provided
(Bernardos et al., 2007). After a complicated take off with context information about the receivers
of LBS (Kaasinen, 2002), standalone navigation and (Khalil and Connelly, 2006)).
tracking applications have nowadays become - Other challenging areas of application are
popular, partly due to some device manufacturers’ context-aware mobile learning (e.g. Mobilearn
efforts to promote the use of GPS enabled devices. project), context-aware mission critical support
Meanwhile, innovative mobile services related to (e.g. Siren project for firefighting) or ubiquitous
Web 2.0 have shown up (it is the case of mobile mobile gaming (e.g. Botfighters).
social software or applications making easier content Context-aware services share requirements with
geotagging, for example) and some analysts expect general mobile ones (eg. with respect to usability
them to contribute to widen the adoption of LBS. and interaction mechanisms, device requirements -
Simultaneously to LBS evolution, research in screen sizes or power consumption-, network speed,
context-aware systems and services has gone ahead. etc) but, the same as location based services, are
Since the pioneers Active Badge or PARCTab especially dependant on:
projects, advances and challenges in positioning 1) Stable location mechanisms. Location is a
techniques, semantic context representations or relevant descriptor of context that is used as a filter
software architectures for context-awareness have in most context-aware applications. The quality of
evolved and a great variety of application the location estimation has a direct impact on the
environments have been explored. By way of user experience. So, reliable, transparent and
illustration, following there is a short list of some of latency-controlled location mechanisms are still
them: needed to offer an acceptable user experience. To
- Context-aware mobile guides: many initiatives date, GPS (and its variants, such as A-GPS) is the
(such as the breaking Cyberguide or GUIDE) most used positioning mechanism outdoors. But
have focused on the development of tourist GPS does not perform well indoors, where the
context-aware mobile guides. Museums (e.g. multiplicity of networks make possible to configure
Sotto Voce or Exploratorium projects) and diverse solutions. Roaming among different
exhibition centres (e.g. Hippie prototype, positioning technologies is a challenge to be solved.
mExpress project or XGuide application) have 2) Reliable privacy management. Context-
also been inspiring environments. Augmented awareness means acquiring and handling identity,
reality techniques are nowadays being real time location or activity data. Users must be

178
 AN ANALYSIS OF CONTEXT-AWARENESS IN COMMERCIAL MOBILE SERVICES

aware of which personal data are being managed and This way, our time-indexed database gathers
have control over the acquisition mechanisms. The information of 735 mobile services, from May 2003
combination of security, pseudonym use or cloaking to March 2008. As the technology watch process has
zones may confer the user with a sufficient feeling not been to collect every new service, but the most
of control. representative or new, the quantitative analysis
3) Sustainable business models: clarifying the happens to be done on critically filtered data.
“ownership” of context data. Mobile operators are Besides, the services are mainly focused on
nowadays handling a lot of personal information European and North America markets (as it is
(such as rough cell-based location data) and difficult to directly access Asian information
providing third partners with the tools to access it. sources, due to language restrictions).
At the same time, hardware manufacturers are In this contribution, the services’ information has
equipping mobile device with autonomous sensors been submitted to further analysis, considering the
(eg. GPS or NFC readers), capable of acquiring aspects of user experience, technology evolution,
context information directly from the device. Some business case, and contextual functionalities
of them are also partially shifting their revenue developed in the following sections.
sources to the provision of applications that use the
acquired parameters. These two different approaches
to context-data acquisition may determine the 4 BUILDING THE BIG PICTURE
development of context-aware services.
“Fully-compliant” context-aware applications OF THE EVOLUTION OF
seem to be in a prototype stage yet. But some mobile MOBILE SERVICES
services are already intensively using some context
descriptors. In the next pages, we intend to analyze In this section, we comment on the mobile
to which extent context-awareness is implemented in ecosystem evolution between 2003 and 2008, the
current commercial mobile services and how it has considered period of analysis. The objective is to
evolved in the last few years. provide the reader with a general (not complete)
picture on how the situation has changed in the last
five years, supporting our statements with some
3 METHODOLOGICAL results of the analysis of the collected data.
APPROACH a) Entertainment and self content generation
together with information services has driven the
Technology Watch, according to the Standard UNE production of mobile applications.
166006-EX, is the continuous process of obtaining To analyze what the present offer of mobile services
and selecting information from a scientific and is and how it has evolved during the last five years,
technological environment, in order to turn it into we have inferred a functional classification on the
useful knowledge by analyzing and spreading it, so stored data. It is composed of 12 categories; we
that organizations can take advantage of it and make consider that a potential category becomes a formal
better decisions by being ahead of changes. one when it groups 20 services as a minimum. The
From 2003 to date, a Centre for Technology classification is as follows (note that some services
Diffusion in the UPM, CEDITEC, accomplishes a may match more than one category):
Technology Watch Programme in ‘Wireless - Location and tracking: of people or mobile
Technologies and Mobile Applications’, which assets (94 services).
includes the generation of a monthly newsletter and - Information: news, weather reports, etc. in
a technology watch report on this issue. The watch addition to eLearning and language translation
process starts with the selection of critical services (130).
information sources related to mobility, more than - m-commerce: mobile payments, financial
60 sources ranging from technical publications to services and m-ticketing (62).
mobile market focused weblogs, also including - Browsing and searching: in the Internet (41).
general press. Afterwards, a methodical process of - Connectivity and communications: including
data collection allows expert filtering to elaborate a VoIP, SMS, MMS, IM or videoconference (96).
selection of the most relevant information in the - Applications and productivity: e-mail
month: news, reports, publications, patents, products applications, document formatting, and remote
and also an average selection of the 15 newest and access to computers or mobile phones (47).
most representative mobile services found in that - Security and safety: data security and
period of time. surveillance applications (20).

179
ICE-B 2008 - International Conference on e-Business

- Entertainment and self content generation: It is noticeable that during 2003-2005, the
photos, video, music, and games downloading, content offer was mainly focused on data
creating, publishing and sharing (155). downloads, whereas from 2006 on, an increasing
- Personal information management: calendars, trend in services supporting the creation and sharing
reminders, customization, settings (78). of mobile content is detected. At the same time, the
- Social networking (49). presence of other categories, almost undeveloped
- Mobile experience enhancers: new interfaces, before, are intensified: it is the case of browsing and
content adaptation, voice to text conversion, 2D searching or doing social networking.
codes reading (72).
c) A technological view: enhanced networks and
- Miscellanea (32).
sensing mobile devices.
Most categories include some subtypes of
There are two key points regarding technology that
services which share the final objective but differ in
shape the changes of the mobile ecosystem during
their functional implementation. It is the case of
2003-2008: a) mobile devices have evolved from
multimedia content and information services. Both
traditional mobile phones to smart devices with
have driven the commercial production of mobile
increasingly embedded technology and b) the
applications in the last years (accounting for 21%
communication infrastructure is not based on
and 18% over the total).
cellular technology any more. Cellular
The other main categories are connectivity and
communication networks have enhanced their
communications (13%), location and tracking
performance and availability, but alternative
services (12.7%) and personal information managers
technologies have shown up and increased their
(10.6%).
penetration rate.
b) Towards mobile 2.0: downloading, but also With respect to mobile devices features, Figure 2
sharing and uploading contents. illustrates an approximated timeline on how new
“Mobile 2.0” is the way to refer to the convergence technologies have been embedded into mobile
between the social web (or Web 2.0) and the basics devices. The arrows point out the early introduction
of mobility (personal, localized and always-on). of a technology in a commercial phone in the
From a practical point of view, Mobile 2.0 is about general market, although its technical availability
connecting your phone to download your favourite may be previously detected and its availability as a
podcasts, read your RSS feeds, do a one-click image commercial common feature may not have happened
upload to an on-line photo management and sharing until several months later.
application, consult the location map while on the
road, tag your streamed videos or update your
moblog.

Figure 2: Approximate timeline of technologies in mobile

devices.

On the other hand, a very general picture of

Figure 1: Evolution of content services (percentage of
communication infrastructures shows that:
services within the total of each quarter). - Cellular networks have increased their data rate,
through UMTS/WCDMA technologies
Using Mobile 2.0 applications needs broadening (384kbps) and its enhanced versions: HSDPA
the type of user operations, going beyond simple (14.4Mbps), HSUPA (5.6Mbps), HSPA+ (42 /
downloading. This trend is somehow shown in 11.5Mbps) and the next LTE – 3GPP Release 8
Figure 1, which describes the evolution of content (>100 / >50Mbps).
services according to our data. To compose Figure 1, - Growing production of standards: WiFi (IEEE
we have divided the services in the “Entertainment 802.11g, June 2003; next IEEE 802.11n in
and self generation content” category between 2009); WiBRO (TTA, late 2005), WiMAX
“downloading” and “uploading” applications. (ITU, October 2007), LTE (GSA, January
2008).

180
 AN ANALYSIS OF CONTEXT-AWARENESS IN COMMERCIAL MOBILE SERVICES

- Great take off of WiFi networks, with 89% mobile devices, whereas in the last six quarters this
growth of WiFi business Hotspots (those market has surrendered to Internet software service
located in airports, hotels, retailers, etc.) over providers.
H2 2007 (iPass, 2008).
- Fixed to Mobile Convergence (FMC) has been a
well-known strategy of some operators, 5 CONTEXT-AWARENESS IN
adopting UMA and IMS in order to provide
better indoors mobile coverage. COMMERCIAL MOBILE
- Enhancements in location technologies: E-911 SERVICES
in USA, market-driven deployment of location
based services in Asia-Pacific zone (DTI, 2004), 5.1 A List of Context-Aware Features
growing integration of GPS receivers in mobile
devices all around the world. From the classification of mobile services proposed
d) Business: mobile service providers shift from in Section 4.1, the reader might conclude that many
operators towards software companies. of the considered services are not context-aware at
Generally, the development and commercialization all. But some of them –although far from being truly
of a mobile service implies the interaction of a context-aware applications– could have some
number of stakeholders. Obviously, the value chain ‘context-aware’ features; intuitively, it is the case,
(or network) composition is very dependent of the for example, of those services grouped into the
final service. For example, location based services ‘location and navigation’ or ‘social networks’
value chains are generally complex: in the case of an categories.
A-GPS mobile location server, it is possible to Measuring the context-awareness level of a
identify at least ten roles doing their part to build the mobile service is not an easy task. For our analysis,
final service (Bernardos et al., 2007), and it will be we have identified a number of basic questions that a
the mobile operator the one offering and billing the context-aware service should be able to answer:
service. On the other hand, a Bluetooth based social 1) Who the target entity/user is and what does
application, offering discovery of peers and it/he/she want/like? Some services will need to
communications will only require an application handle personal descriptors, these understood as
provider and, indirectly, a handset compatible identity, profile, preferences and group membership
provider. In this case, the user will get the related issues. Personalization and services’
application directly from the service provider adaptation to the user’s profile are key for many
website, and will be not charged for using it. applications.
2) Where the user is? Location is a physical
descriptor that usually provides significant
knowledge about the user’s context. Location acts as
a situational filter that is influenced by the
positioning system accuracy. Being the basis of
navigators and trackers, location availability is
making possible some innovative services such as
mobile locative social software, pervasive games or
geotagged content making (photos, blog posts, etc.
may be georeferred).
Figure 3: Comparison of mobile services’ providers. 3) How the user is? Environmental and
biometric data use. Apart from location, context-
On the whole, operators start to change their aware services may use other physical descriptors to
walled-garden strategies (Wieland, 2007) towards describe the environment where entities are plunged
more open ones, in part to incentive data traffic (this may be useful, for example, for environmental
through popular 2.0 applications and to let the users and agriculture surveillance applications and
browse in a more familiar way. Well-known web domotics), or even to monitor their biometric
services providers that have mobilized their parameters, in order to infer information about the
applications (p.e. Google, Yahoo! or Skype) are target entity physical state.
acting as users’ attractors to mobile data services. 4) What is the current user’s activity? Not
From our analysis of data, we get the following independent of the previous issue, information about
picture (Figure 3): mobile operators started the current activity and logical state (presence or
dominating the provision of content and services to connectivity, for example) of a target entity may

181
ICE-B 2008 - International Conference on e-Business

complete the inferences about its situational Elaborating on these questions, Table 1 gathers
condition. Features such as presence announcement some features enabling context-awareness. From
and state dissemination or availability notifications their combination, mobile services may be built:
are considered in this point. features 1, 2, 3, 5 and 7 may be identified in a
5) What does the user intend? This feature is mobile social software application, while a sport
related to personal or professional activity data monitoring service may need to handle
gathering, under the shape of calendars, schedules or characteristics 1, 3, 4 and 6.
notifications, or behavioural patterns. In the next Section we analyze the
6) Who/what is near the user? In certain implementation of these features in the stored data.
applications, building a social group based on
proximity issues with defined privacy levels and 5.2 Some Results
collaboration policies is the service’s leitmotif.
Context-aware services combine physical and - Limited but detectable use of context-aware
situational data with the management of preferences features. 20% over the total of analyzed services are
and interests, creating social networks which using personal descriptors, only 8% are employing
promote direct interaction among peers. Apart from physical parameters and managing resources and
discovering people, discovering and interacting with communications, while 5% of the services are using
daily objects through wireless location (WiFi or BT activity descriptors. At least one of the considered
based systems, for example) and proximity features appears in 47% of the 735 services. Around
technologies (such as NFC or RFID) is also possible. 49% of these services are just implementing one of
7) With whom the user may communicate? the features; in fact, only 9.4% employ more than
Applications which are focused on enabling peer four context-aware characteristics.
communications make possible the formation of - Upward trend in personalization and emergence
proximity-based groups. of sensor-assisted HCIs. Discarding the incomplete
8) How are the user’s interaction series of 2003 and 2008 and aggregating the rest of
mechanisms? Context-awareness aims at the services in two biannual periods (2004-2005 and
facilitating daily living and interaction with the 2006-2007), Figure 4 shows a general increase of
environment. In the last years, mobile devices have context-awareness.
increased the number of embedded sensors they Personalization (identity and profile, preferences
have. Accelerometers, cameras, gyroscopes, etc. and group membership) appears as a general trend.
enable more intuitive interfaces: pointing an object Real time information about the user connectivity
to get information about it or using a 2D sensor (logical state) is also increasingly used. On the other
reader to make easier the information search process hand, peer to peer communications have crossed the
also increase the development of mobile context- barrier of mobility in the second period of time.
awareness. Almost 16% of the new mobile applications also
manage location descriptors. Finally, services using
Table 1: A list of context-aware features. new interfaces have significantly augmented in the
CONTEXT-AWARE FEATURES period 2006-2007.
PERSONAL Sensor-assisted HCIs

1 Identity and profile management Peer to peer communications

Resources allocation
2 Preferences and group membership Social network
management Nearby resources discovery

PHYSICAL Activity data gathering

Logical state info.
3 Positioning capabilities Environmental or biometric data
4 Environmental or biometric data Positioning capabilities
acquisition and management Profile, preferences, group

ACTIVITY Identity

5 Connectivity and presence information 0 5 10 15 20 25 30 35

6 Activity data gathering and storing % 2004-2005 % 2006-2007

RESOURCES MANAGEMENT & COMMUNICATIONS
7 Nearby resources discovery Figure 4: A biannual comparison of context-awareness
8 Resources management and allocation implementation in mobile services. Percentage over the
9 Peer to peer communications total production in the periods of analysis (300 services in
INTERFACES 2004-2005 and 304 services in 2006-2007).
10 Sensor-assisted HCIs
- LBS lead the group of precontext-aware
services, together with information,

182
 AN ANALYSIS OF CONTEXT-AWARENESS IN COMMERCIAL MOBILE SERVICES

communications, personal information managers resources are limitedly using physical filters such as
and social networks applications. Figure 5 position, with the obvious exception of “Location
compare the implementation of different context- and tracking” category.
aware features based on the categories presented in
Section 4.1. It shows that Location and tracking
services are the most personalized and activity aware 6 CONCLUSIONS
ones. Personal information managers do not lag
behind with respect to these two features, although This work explains the results of the analysis of a
in general they do not include location as a large database of mobile services, systematically
descriptor. On the contrary, information services are
updated during the period 2003-2008. Apart from
increasingly using positioning as a filter. Mobile giving a general view of mobile services evolution,
commerce services do not use activity data but are we have focused on studing to which extent context-
personalized and sometimes related to the user
awareness is implemented in commercial mobile
location. In general, “entertainment and self content services. With this purpose, we have identified
generation” services are scarcely considering the use several context-aware descriptors (personal, physical
of location.
or activity related) and features (capability of
Miscellanea managing resources, establishing P2P
Interface communications or implementing advanced HCIs),
Social networking
and evaluated their implementation in mobile
Personal information management
Applications and productivity
services.
Entertainment - In general, few commercial mobile services can
Security and safety be considered fully ‘context-aware’, although
Connectivity and communications isolated features have been detected in about
Browsing and Searching
m-commerce
half of them.
Information - Personalization -this understood as the
Location and tracking capability of handling identity, preferences and
0 10 20 30 40 50 60 70 80 90 group membership information- is the context-
Personal Physical Activity Resources aware feature that most applications implement.
Figure 5: Number of services for each category that
Physical and activity descriptors’ use lags
implement personal, activity, physical and resource related behind.
features. - Mobile services are increasingly incorporating
the use of context-awareness, although there is
not a breaking point in the period 2004-2007.
Again, personalization is the most significant
trend.
- Context-aware features have been detected in
services of all the functional categories.
Anyway, services in the following categories
“Location and Navigation”, “Social networks”,
“Information” and “Personal information
management” are showing the highest levels of
context-descriptors’ use.
- Mobile ‘social networking’ applications may be
considered as highly context-aware. They have
Figure 6: Evolution of context-awareness in the periods remarkably appeared during 2006-2007 and are
2004-2005 and 2006-2007. Data for some of the intensively using personal descriptors while
categories identified.
discovering and managing resources. Location-
based filtering is also included in some mobile
When analyzing what kind of services includes social services.
resources discovery and management capabilities, - The integration of sensors such as gyroscopes or
the “Connectivity and communications” category is accelerometers is making possible new
the one which aggregates the greater number of interaction mechanisms with the environment.
applications, followed by the “Location and Sensors assisted HCIs proposals have
tracking” and “Social networking” categories. As it noticeably grown in 2006-2007.
can be noticed, most of the services handling

183
ICE-B 2008 - International Conference on e-Business

Nowadays favourable boundary conditions - Gilbert, A.L., Han, H. (2004). Understanding mobile data
progressively more connected environments, services adoption: Demography, attitudes or needs?
evolved mobile devices (with new communication Tech. Forecasting & Social Change 72, 327-337.
capabilities, more usable HCIs and embedded GUIDE (1999). The GUIDE Project.
sensors), “literate” users and more established data http://www.guide.lancs.ac.uk/overview.html.
markets – seem enough to make context-aware iPass (2008). Mobile Broadband Index.
services enter in the commercial offer. This http://www.ipass.com/pdfs/2H07_WiFiHotspot_Index
“context-awarezation” of mobile services has .pdf
already started, even if there is still a long way to go. ISTAG (2001). Scenarios for Ambient Intelligence in
2010. European Commision. Available in
http://www.cordis.lu/ist/istag-reports.htm.
Kaasinen, E. (2003). User needs for location-aware mobile
ACKNOWLEDGEMENTS services. Personal Ubiquitous Computing 7, 70-79.
Kelleher, C., Damsgaard, J. (2007). What drives the
This work has been financed by the Spanish innovation, diffusion and adoption of mobile services?
Ministry of Education and Science under grant An analysis of four alternative studies. 6th Annual
TSI2005-07344 (COLOCAME) and by the Global Mobility Roundtable 2007, June 1-2, L.A.
Government of Madrid under grant S-0505/TIC- Khalil, A., Connelly, K. (2006). Context-aware
0255 (MADRINET). Telephony: Privacy Preferences and Sharing Patterns.
ACM Proceedings of the Computer Supported
Cooperative Work Conference. November 4-8, Banff,
Alberta, Canada.
REFERENCES Meyer, S., Rakotonirainy, A. (2003). A survey of
research on context-aware homes. Proc. of the
Agarwal, S., Joshi, A., Finin, T., Yesha, Y., Ganous, T. Australasian information security workshop
(2007). A Pervasive Computing System for the conference, 2003. Vol. 21, pp.159-168.
Operating Room of the Future. Mobile Netw. Appl. Mohr, W. (2008). Vision for 2020? Wireless Personal
12:215-228. Communications 44:27-49.
Arvidsson, Ǻ., Hederstierna, A., Hellmer, S. (2007) Schilit, B.N, Theimer, M.M. (1994) Disseminating active
Simple and Accurate Forecasting of the Market for map information to mobile hosts. IEEE Network,
Cellular Mobile Services. ITC 2007, Lecture Notes in pages 22-32, September/October.
Computer Science 4516, pp. 690-706. Verkasalo, H. (2007). Mobile Data Service Evolution –
Bernardos, A.M., Tarrío, P., Casar, J.R. (2007). A Empirical Observations on Packet Data Service
taxonomy of mobile location-based applications. Adoption. Available in
Proceedings of the 2007 International Conference on http://www.netlab.hut.fi/~hverkasa/verkasalo_dataserv
e-Business, pp. 114-120, Barcelona, Spain. ices_final_submission.pdf, [07/04/2008].
Bina, M., Giaglis, G.M. (2005). Exploring Early Usage Weiser, M., Gold, R., Brown, J.S. (1999). The origins of
Patterns of Mobile Data Services. IEEE Proceedings ubiquitous computing research at PARC in the late
of the International Conference on Mobile Business 1980s. IBM Systems Journal, vol. 38, no. 4.
(ICMB’05), 11-13 July 2005, Sydney, Australia. Wieland, K. (2007). Walled gardens come tumbling down.
Bouwman, H., Carlsson, C., Molina-Castillo, F.J., Telecommunications online, available in
Walden, P. (2007) Barriers and drivers in the adoption http://www.telecommagazine.com/article.asp?HH_ID
of current and future mobile services in Finland. =AR_3373 (April 25th 2008)
Telematics and Informatics 24, 145-160.
Bricon-Souf, N., Newman, C.R. (2007). Context-
awareness in health care: A review. International
Journal of Medical Informatics 76, 2-12.
Dey, A.K., Abowd, G., (1999) Towards a Better
Understanding of Context and Context-Awareness
Source. Proc. of the 1st Int. symposium on Handheld
and Ubiquitous Computing, Karlsruhe, Germany. pp
304-307.
Funk, J.L. (2007). The future of mobile shopping: The
interaction between lead users and technological
trajectories in the Japanese market. Technological
Forecasting & Social Change 74, 341-356.

184
 AN ACCESS-CONTROL MODEL FOR MOBILE COMPUTING
WITH SPATIAL CONSTRAINTS
Location-aware Role-based Access Control with a Method for Consistency Checks

Michael Decker
Institute AIFB, University of Karlsruhe (TH), Englerstr. 11, 76 128 Karlsruhe, Germany
[email protected]

Keywords: Mobile Information Systems, Access Control, Location-based Services, Mobile Security.

Abstract: Some of the most salient challenges that come along with the employment of mobile information systems stem
from security issues: portable devices like PDAs, smartphones and notebooks easily get stolen or lost and
wireless data transmission could be eavesdropped, so that unauthorized individuals gain access to confidential
resources. One approach to tackle these problems is location-aware access control, i.e. based on knowledge
about the user’s position the information system can decide if access to a resource should be granted or not.
For example a nurse using a PDA should only be allowed to access confidential patient data while staying on
the premises of the hospital. In our article we present a data model for location-aware access control based
on the concepts of roles. Using our model it is possible to assign location restrictions to several entities, e.g.
to users, to roles or permissions. We also propose a method to analyze the consistency of spatial constraints
expressed by an instance of our model.

1 INTRODUCTION user just stays on the premises of the company; down-

loading a confidential document to a notebook com-
Nowadays it is possible to build portable computers puter could be denied if the user stays abroad. To de-
like PDAs and smartphones which are more power- termine the location of a mobile device there are sev-
ful than stationary systems were a few years ago. At eral technologies available (e.g. GPS or CellID), see
the same time technologies for wireless data commu- Küpper (2005) or Hightower & Borriello (Hightower
nication like GPRS, WiFi or UMTS were developed and Borriello, 2001) for an overview.
and are nowadays widely available. Based on these To implement location-aware access control a for-
two technologies mobile information systems (MIS) malism is required to express which resources can be
can be realized to make computer support available accessed at which location by which users. Such a
almost anywhere and anytime. These MIS have a formalism is called ”access control model” (ACM).
great potential for many novel application scenarios, In our article we will introduce a location-aware ACM
especially with regard to support mobile workers. But that is based on the notion of role-based access con-
the utilization of mobile technologies comes along trol (RBAC). Location-restrictions can be assigned to
with serious security challenges, because due to their several elements in this model to provide a great de-
portability and small size mobile devices easily get gree of versatility. We also introduce methods to ana-
stolen or lost. If unauthorized individuals get into the lyze model instances based on geometric operations;
possession of a mobile device they can gain access to as far as we are aware this concept is novel and not
confidential resources (e.g. health records, telephone mentioned in pertinent literature yet.
numbers of board members, financial data). The remainder of the article is structured as fol-
In our work we employ the concept of location- lows: We will first give a short introduction to access
aware access control to address this problem field. control models in section 2.1 before we introduce our
The basic idea is that for the decision wether a par- own model and give an example scenario in section
ticular user is allowed to access a given resource the 2.2. A formal presentation of the model will be given
current location of the mobile device is regarded. For in section 3; this comprehends also a simple location
example the access to the customer database using a model. Based on this formalization we describe some
mobile device should only be granted if the mobile ways to analyze model instances by geometric con-

185
ICE-B 2008 - International Conference on e-Business

siderations in section 4. We also implemented a small 2.2 A Location-Aware Access Control

system to visualize the location constraints of model Model based on RBAC
instances and to run the analyzing algorithms; this im-
plementation will be sketched in section 5 before we
In this subsection we will introduce an RBAC-based
conclude in section 6.
model for location-aware access control which is de-
picted in figure 1 as entity-relationship diagram. The
cardinalities in the diagram are given in the form of
2 ACCESS CONTROL MODELS (min, max)-values. For example ”(1, ∗)” as cardinal-
ity of entity ”role” in the ”role-location”-relationship
2.1 Basics means that each role entity occurs at least once in the
relation representing that assignment. Since ”max =
Access Control Models (ACM) are formal models ∗” a role entity may occur as many times as needed in
to describe under which conditions an information that relation.
system should grant access to resources. There are Each user can be assigned to several roles. Roles
three main groups of ACM (Samarati and di Vimer- again are collections of permissions. Permissions in
cati, 2001): mandatory access control (MAC), discre- RBAC-models are usually considered as symbols that
tionary access control (DAC) and role-based access have to be interpreted for the respective deployment.
control (RBAC). In the figure we interpret a permission as a set of oper-
Using MAC systems subjects as well as objects ations (e.g. read, write, update) that can be performed
are classified according to their level of confidential- on a given class of objects. An object class might rep-
ity (e.g. ”top secret”, ”secret”, ”confidential”, ”un- resent a class of documents like ”customer record”.
classified”). Then there are rules that impose restric- For each object class there might be several object in-
tions like ”a subject should only be allowed to read an stances, e.g. ”customer record for Mr. Meyer” and
object if the subject’s classification (or clearance) is at ”customer record for Ms. Miller”. The entity location
least as high as the one of the object.” One prominent represents a location like a city, a room or a country.
example for such MAC is the Bell-LaPadula-model. We will later give a formal definition for what a loca-
DAC models follow the owner principle: each object tion is.
under protection of the access control model belongs Several components of the model have a relation
to an owner. This owner can grant rights to perform to at least one location. A component is only enabled
certain operations on that object to other subjects. Ex- if the respective user stays at that location, e.g. if
amples for DAC-models are those used by operating the role ”manager” is restricted to the locations ”lo-
systems, e.g. on Unix a user can grant the right to read cal branch Munich” and ”local branch Lisbon” this
a file he just created to users of a so called group. For means the role can only be used when the user is in
the article at hand the most important group of ACM Munich or Lisbon. If there should be no location re-
is RBAC (Ferraiolo et al., 2003). The basic notion striction we just assign the location ”universe” (which
is to assign permissions not to individual subjects but covers the whole reference space) to the respective
rather to so called roles. Roles represent job descrip- component. The remaining points for the assignment
tions or positions within an organization. So if Alice of location restrictions are:
is hired as new ”manager” the permissions required • Restricting a user to locations means that the user
for this job (e.g. read payroll file, approve order) don’t can use the system only at certain locations, e.g.
have to be assigned individually; we just have to as- we can restrict a user to ”southern Spain” because
sign the ”manager” role to Alice’s user account. he has only to work in this region for the organi-
Almost all location-aware access control mod- zation or he bought a software licence that covers
els that can be found in literature are extensions of only this part of the country.
RBAC. The basic idea to obtain a location-aware
RBAC model is to switch particular elements of the • Restricting a permission to locations means that
RBAC on or off depending on the user’s location, this permission should only be accessible if the
e.g. roles (Bertino et al., 2005), the role-permission- user is at the specified locations, no matter how
assignment (Hansen and Oleshchuk, 2003) or the powerful his role is. An example is that ”access
user-role-assignment (Chandran and Joshi, 2005). payroll” should only be allowed while being on
the company’s premises.
• It is also possible to assign location restrictions for
individual objects, e.g. the object ”research report
no. 123” is not allowed to be accessed outside a

186
 AN ACCESS-CONTROL MODEL FOR MOBILE COMPUTING WITH SPATIAL CONSTRAINTS - Location-aware
Role-based Access Control with a Method for Consistency Checks

(0,*) (1,*)
Location Object-Instance
(0,*) (0,*) (0,*) (0,*) (0,*) (1,1)

(0,*)
(0,*)
(1,*) (1,*) (1,*) (0,*) Object-Class
(0,*) (0,*) (0,*) (0,*)
User Role Permission
(0,*)
(0,*) Operation
Figure 1: Access Control Model as Entity-Relationship-Diagram.

certain country because espionage is feared.

• Attaching a location restriction to the user-role- p2
assignment means that the given user is allowed to r3
p2
play that role only at certain locations, e.g. a mo- r1 p1 r4
p1
bile worker is only allowed to play the role ”trav-
elling salesman” in his personal sales district.
• Attaching location restrictions to the role- r2
permission-assignment means that the given per-
p2
mission can be only performed using that role p1
when the user is at certain locations, e.g. we could p3
have a role ”manager” that encompasses the per-
mission ”approve order” but only within Europe.
We don’t assume that for a real-world instance of Figure 2: Example for spatial constraints.
our model all six location restrictions will be used;
but having six possible components where location- service technicians): r1 is for the northern part of dis-
restrictions can be assigned to gives the freedom to trict 1, r2 for the southern part of that district. The
choose the points that are most appropriate or natural number of jobs to perform in districts 2 and 3 isn’t
for the respective scenario. If location restrictions are big, so a single role r3 is sufficient for both. r4 is
assigned to more than one point in the model it might the manager’s role who has to visit facilities in all re-
be necessary to calculate the intersection of different stricts, e.g. to handle complaints or to acquire new
location restrictions to obtain the area where the user customers. Therefore this role’s location constraints
is indeed authorized to perform a certain permission. encompass all districts. Role r3 has the permissions
To exemplify our model we give the following ex- p2 and p3 ; this means that a user having r3 can access
ample: The scenario encompasses a company with customer data (p2 ) in the areas where the rectangle
three local branches that sends mobile employees to of r3 intersects with the two rectangles for permission
facilities in a given area (e.g. to deliver goods or to p2 ; the navigation service is only accessible where the
perform some kind of maintenance work). In figure 2 rectangle for r3 intersects with the rectangle for p3 .
the location restrictions for four roles (dotted lines)
and three permissions (solid lines) are depicted as
rectangles: p1 is the permission to access the payroll 3 FORMAL DESCRIPTION OF
file; since this is confidential information this should
only be possible while staying at one of the three lo- THE MODEL
cal branches. p2 is the permission to access customer
data; this permission is restricted to three service dis- 3.1 Location Model
tricts where the facilities are that have to be served by
the company’s employees. p3 is a navigation service To give a formal description of our model and to de-
licensed from an external service provider; this ser- scribe spatial analyzes to be performed on instances
vice is accessible only in an area that covers most of of the model we need a formalization of the concept
the three services districts (licensing for a larger area of a ”location”:
would be more expensive). We denote the area to be covered by the ACM as
Roles r1 to r3 are roles for mobile workers (e.g. reference space or universe. Depending on the scope

187
ICE-B 2008 - International Conference on e-Business

of the MIS universe may be the premises of a com- role r0 then (u0 , r0 ) ∈ UR and if permission p0 is as-
pany, a country or earth’s surface. All other locations signed to role r0 then (r0 , p0 ) ∈ RP. Together these
l like cities, regions, buildings or rooms are real sub- sets form the set of main entities E:
sets of the universe: l ⊂ universe. The set of all lo-
cations l and the universe is named locs: E = U ∪ R ∪ P ∪ UR ∪ RP
The function loc() takes as input one of these enti-
locs = {l|l ⊆ universe}
ties and returns the subset of locations the respective
With regard to the later implementation we demand entity is restricted to:
that all locations are polygons because locations will
loc() : E −→ 2locs \ 0/
be stored in a database system. Since we can approxi-
mate every area by a polygon with arbitrary precision The locations returned by loc() are also denoted as a
this is a weak restriction. If PT denotes the location user’s, a role’s resp. a permission’s area. Example:
”Portugal” and LIS the location ”LISBON” we get:
locs = {universe, PT, LIS, . . .}. loc(u1 ) = {PT, MUN}
For each location or set of locations we can calcu- This means the user u1 is allowed to use the system
late the covered area with the function area(), which in Portugal or Munich. If an entity isn’t restricted to
will return a real number greater than zero: a location at all loc() just returns the universe. If we
area() : 2locs −→ R+ want to express that user u2 can activate role r2 only
in Lisbon this would be:
If two locations of an input set overlap the intersecting
loc((u2 , r2 )) = {LIS}
area will be counted only once.
For two locations l1 , l2 ∈ locs we can calculate
the intersection and the union. When calculating the
intersection of two polygons we may obtain one poly- 4 SPATIAL ANALYSES OF
gon, several polygons or the empty set as result, so the MODEL INSTANCES
intersection operator’s range is the power set of locs:
l1 ∩ l2 −→ 2locs In this section we will describe some approaches to
perform spatial analyses on instances of our model
The union set of l1 and l2 can consist of one location to find inconsistent configurations. For the sake of
(if l1 and l2 intersect or touch each other) or two loca- simplicity we don’t consider location restrictions as-
tions, but the empty set isn’t possible: signed to user-role-assignments and role-permission-
assignments.
/
l1 ∪ l2 −→ 2locs \ 0, |l1 ∪ l2 | ∈ 1, 2
There is also the set C of location classes. Location 4.1 Coverage
classes are semantic categories for locations, e.g. C =
{cities, countries, buildings, . . ., unclassi f ied}. Each Coverage is a function that takes one entity from the
location in loc is mapped to exactly one location class set of all users, roles and permissions as first argu-
in C: ment (pivot entity) and one of the target categories
class() : locs −→ C T = {”user”, ”role”, ”permission”} as the second ar-
gument:
One class in C is ”unclassified” which is the ”dummy
class” to be used if there is no reasonable semantic cover() : (U ∪ R ∪ P) × T −→ 2locs
category for a location. Location classes are a mean
to support human users when browsing through the If the pivot entity is of the type specified by the target
set of pre-defined locations. category then cover() just returns the location restric-
tion of the pivot entity, e.g.
3.2 Core Model cover(u0 , ”user”) = loc(u0 )

We have a set for each of the main entities of the If the pivot entity stands ”more left” than the target
ACM: the set of users U = {u1 , u2 , . . . }, the set of category in figure 1 then the result is the set of loca-
roles R = {r1 , r2 , . . . } and the set of permissions P = tions where the pivot entity can activate at least one
{p1 , p2 , . . . }. For the sake of brevity we don’t con- entity of the target category:
sider the classes and objects behind a permission. • cover(u0 , ”perm”) returns the area where user u0
Further there are the sets UR and RP: if user u0 has can activate at least one permission.

188
 AN ACCESS-CONTROL MODEL FOR MOBILE COMPUTING WITH SPATIAL CONSTRAINTS - Location-aware
Role-based Access Control with a Method for Consistency Checks

• cover(u0 , ”role”) returns the area where user u0 Permission p0 is assigned to both roles. The respec-
can activate at least one role. tive location restrictions are depicted in figure 3. The
• cover(r0 , ”perm”) returns the area where with role coverage is smaller than the location restriction as-
r0 at least one permission can be activated. signed to p0 because there is an uncovered area (the
shaded part of the uppermost permission-area). This
In the opposite case (the pivot element is on the right- part is covered by role r2 . User u2 and u3 are assigned
hand side of the target category) the locations returned to that role, but their location restrictions don’t cover
are the area where the pivot element can be activated the shaded area.
by at least on entity of the target category:
• cover(p0 , ”role”) returns the area where permis- 4.2 Empty Assignments
sion p0 can be activated by at least one role.
• cover(p0 , ”user”) returns the area where permis- We talk about an ”empty user-role assignment” when
sion p0 can be activated by at least one user. for (u0 , r0 ) ∈ UR the following holds:
• cover(r0 , ”user”) returns the area where role r0 loc(u0 ) ∩ loc(r0 ) = 0/
can be activated by at least one user. The intersection between the locations assigned to the
Due to space limitations we will only discuss user and the role is empty, so the respective assign-
cover(p0 , ”user”) in detail, which is also the most in- ment is redundant, i.e. it could be removed without
teresting case: if permission p0 is location-restricted changing the policy.
to certain locations (e.g. area where customers have Further the following case shouldn’t occur:
to be served, area for which a software license was loc(u0 ) ∩ cover(r0 , ”perm”) = 0/
bought) we may want to check if every point of the This means that role r0 was granted to user u0 , but this
permission area is covered by at least one user, i.e. is assignment doesn’t allow him to perform any permis-
there a part of the permission area where no user ac- sion. In the same way we can check for ”empty role-
cording to his roles and his own location restriction permission assignment” for (r0 , p0 ) ∈ RP:
can use that permission? If this case is given it would
loc(r0 ) ∩ loc(p0 ) = 0/
mean that we had customers that cannot be served or
that we bought the software license for a too big area. Each assignment made by the administrator of a
The formal definition is as follows: model should be checked for these two cases; if de-
tected the administrator should be asked if he is sure
cover(p0 , ”user”) = about this assignment.
[
loc(p0 ) ∩ ( cover(r0 , ”user”))
(r0 ,p0 )∈RP
So we have to calculate the intersection of the permis- 5 IMPLEMENTATION
sion area with the union of all coverages for roles that
are assigned to p0 . The coverage of a role again is: We developed an application with a Java/Swing-based
graphical user interface to have a runtime environ-
cover(r0 , ”user”) = ment for instances of the proposed ACM. The appli-
[ cation can perform the spatial analyses covered in the
loc(r0 ) ∩ ( cover(u0 , ”user”))
(u0 ,r0 )∈UR)
last section and is also capable of visualizing location
constraints assigned to individual entities.
Here the location area of role r0 has to be intersected The instances of the ACM are stored in a Post-
with the union of the area of all users that are assigned greSQL/PostGIS database management system which
to role r0 . provides support for working with spatial data. For
If the permission area is smaller than the coverage, the visualization of spatial data we resorted to Open-
i.e. Jump.
area(cover(p0, ”user”)) < area(loc(p0 )) OpenJump can load spatial data from a database
into different layers. We use this feature to load the lo-
we have locations where the permission could be per-
cation areas for different entities of a model instance
formed, but no user is available who could. This is a
into various layers to see how they are related. Fig-
strong hint for an erroneous configuration.
ure 4 is a screenshot from the module of the applica-
To exemplify this we give an example with two
tion that implements the coverage calculation as de-
roles and three users. For the user-role-assignment
scribed in section 4.1. On the right side we can see
we have:
the OpenJump-GUI that displays the covered and the
UR = {(u1 , r1 ), (u2 , r1 ), (u2 , r2 ), (u3 , r2 )} uncovered area of the example given in figure 3.

189
ICE-B 2008 - International Conference on e-Business

u1 u1

r1
r2 u2
p0
p0 u3
r1 r2

u3
p0 u2

Figure 3: Example for permission-coverage.

Figure 4: Screenshot of module for calculation of coverage along with visualization provided by OpenJump.

6 CONCLUSIONS the 6th International Conference on Web Information

Systems Engineering (WISE ’05), pages 361–375.
We started by motivating the need for location-aware Ferraiolo, D. F., Kuhn, D. R., and Chandramouli, R. (2003).
Role-Based Access Control. Artech House, Boston
access control and proposed a RBAC-based model and London.
where location-restrictions can be assigned at several
Hansen, F. and Oleshchuk, V. (2003). SRBAC: A Spatial
components to provide a great degree of flexibility. Role-Based Access Control Model for Mobile Sys-
Based on a formalization of the model itself and the tems. In Proceedings of Nordsec ’03, pages 129–141,
underlying location model we introduced some ap- Gjovik, Norway.
proaches to analyze model instances based on spatial Hightower, J. and Borriello, G. (2001). Location Sys-
calculations; this helps to detect erroneous location tems for Ubiquitous Computing. IEEE Computer,
restrictions. We also sketched the prototypical imple- 34(8):57–66.
mentation of a system as runtime environment for the Küpper, A. (2005). Location-based Services – Fundamen-
model and visualization purposes. tals and Operation. John Wiley & Sons, Chichester,
U.K.
Samarati, P. and di Vimercati, S. D. C. (2001). Access Con-
trol: Policies, Models, and Mechanisms. In FOSAD
REFERENCES ’00: Revised Versions of Lectures Given during the
IFIP WG 1.7 International School on Foundations of
Bertino, E., Catania, B., Damiani, M. L., and Perlasca, P. Security Analysis and Design on Foundations of Se-
(2005). GEO-RBAC: A Spatially Aware RBAC. In curity Analysis and Design, pages 137–196, London,
Proceedings of SACMAT ’05, pages 29–37, Stock- UK. Springer.
holm, Sweden.
Chandran, S. M. and Joshi, J. (2005). LoT-RBAC: A Loca-
tion and Time-Based RBAC Model. In Proceedings of

190
 A PLATFORM FOR INVESTIGATING EFFECTIVENESS FOR
STATIC, ADAPTABLE, ADAPTIVE, AND MIXED-INITIATIVE
ENVIRONMENTS IN E-COMMERCE

Khalid Al-Omar and Dimitris Rigas

Department of Computing, University of Bradford, Bradford BD7 1DP, U.K.
[email protected], D.Rigas@@Bradford.ac.uk

Keywords: Adaptive, Adaptable, Mixed-Initiative, Static, Usability, Effectiveness, Interactive.

Abstract: This paper introduces an empirical study to investigate the use of four interaction conditions: Static,
Adaptable, Adaptive, and Mixed-initiative. The aim of this study is to compare the effectiveness of these
four conditions with regard to the number of tasks completed by all users and the number of users who
completed all tasks. In order to carry out this comparative investigation, four experimental interfaces were
built separately. These environments were tested independently by four separate groups of users, each group
consisting of 15 users. The results demonstrated that in the searching tasks the most effective condition was
the Mixed-Initiative. In the learnable tasks the most effective condition was the Adaptable condition. In
addition, the Static approach was found to be less effective than all other approaches.

1 INTRODUCTION and system and this is called Mixed-initiative

approach which is a combination of adaptable and
Today, software application and e-commerce web- adaptive approach.
based application (Alotaibi and Alzahrani, 2004) is Despite the disagreement in the research
crowded with functions, icons, menus, and toolbars community, there are multiple direct comparisons
(McGrenere et al., 2007). In addition, the web-based between Static, Adaptable, and Adaptive approaches
e-commerce application is crowded in both the have shown different results. In 1985, the first study
Graphical User Interface and content. This is a of adaptation was reported by Greenberg and
phenomenon called ‘Bloatware’ or ‘featurism Witten. They demonstrated an adaptive interface for
creeping’ (McGrenere et al., 2007). This a menu-driven application. In their study users were
phenomenon makes searching for information and novices on the task and the interface (Greenberg and
products within e-commerce web-based application Witten, 1985). In addition, Greenberg and Witten
very complex (Findlater and McGrenere, 2004) (1985) built a directory of telephone numbers that
(Te’eni and Feldman, 2001). Therefore, users can access through a hierarchy of menus. Their
personalising the application to users need and goal is to reduce the number of key-presses buttons.
preferences is essential and becomes very important Their approach is to present items at a level in the
(Findlater and McGrenere, 2004), (Fink et al., 1998). hierarchy according to the number of selection.
Personalisation is a topic of debate between two Greenberg and Witten tested their system against a
communities, the Intelligent User Interface static system in a 26-participant experiment. Their
community favouring adaptability (Shneiderman, B. results showed that subjects performed faster with
and P. Maes, 1997) at the expense of user freedom the adaptive system, and 69% of subjects prefer the
and Human Computer Interaction community adaptive system. In addition, they found that the
favoured adaptability (Shneiderman, B. and P. Maes, adaptive system reduces the search paths for
1997) at the expense of system help. According to repeated names, reduce 35% in time per selection,
McGrenere et al. (2002) there are three potential and reduce 40% in errors per menu. Trevellyan and
ways to personalisation: 1) by users and this is called Browne (1987) replicated the Greenberg and
an adaptable approach. 2) by system and this is Witten’s experiment with a larger number of trails
called An adaptive approach. 3) by both the users because they believe after a large of trails subjects

191
ICE-B 2008 - International Conference on e-Business

will be familiar with the static and they can not slower than the static menu. In addition, it shows
memorised the sequence of key-presses. This would that the adaptable menu was preferable than the
reduce the mean time per menu. However, they static menu and the static was not preferable to
found that the adaptive system is effective and after adaptive menu. Another study examined how
using the system for long period of time users did characteristics of the users’ tasks and customisation
begin to perform better with the static interface. This behaviour affect their performance on those tasks
study did not provide a firm conclusion since the (Bunt et al., 2004). The results confirm that users
total number of subjects in each interface is 4 may not always be able to customise efficiently. The
subjects. results indicate that customisation is beneficial to
In 1989, Jeffrey Mitchell and Ben Shneiderman reduce tasks time if it done right. Also, indicate that
(1989) conducted an experiment to compare an the potential for adaptive support to help users to
adaptive menu that items positions change overcome their difficulties.
dynamically according to frequently clicked item, In 2005, Tsandilas and Shraefel conducted an
with a static menu. Sixty-three subjects assigned empirical study that examined the performance of
randomly tried both menus and carried out the same two adaptation techniques that suggest items in
12 tasks in each menu. Their results showed that adaptive lists. They compared between the baseline
static menu faster than the adaptive menu at first where suggested menu items were highlighted and
group of tasks, and no difference in the second shrinking interface which reduced the font size of
group of tasks. That because, subjects in both groups non-suggested elements. The results indicate that
were able to increase their performance the Shrinking information was shown to delay the
significantly. However, Eighty one percent of the searching of items that had not been suggested by
subjects preferred the static menu. Another study the system. In addition, the accuracy affected the
introduces a system to provide environment for ability of participants to locate items that were
adapting Excel’s interface to particular users correctly suggested by the system. Gajos et al.(2005)
(Thomas and Krogsæter, 1993). The result showed comparing two adaptive interfaces: 1) their Split
that an adaptive component which suggests interface, which is most of the calculator’s
potentially beneficial adaptations to the user could functionality was placed in a two-level menu. 2)
motivate users to adapt their interface. Jameson and Altered Prominence interface, all functionality was
Shwarzkopf (2000) conducted a laboratory available at the top level of the interface. The study
experiment with 18 participants a direct comparison showed user preference for the split interface over
between automatic recommendations, controlled the non-adaptive baseline. Another experiment
updating of recommendations, and no compared the learning performance of static versus
recommendations available. Their comparison dynamic media among a 129 students. Their result
concerned about the content rather than the showed that the dynamic media (animation lessons)
Graphical User Interface. Their results showed that has a high learning performance than the static
there was no difference on performance score media (textbook lessons) (Holzinger, 2008).
between the three conditions. Despite the debate between the two
In 2002 McGerenere et al. conducted a six-week communities, there has been very little work directly
with a 20 participant field study to evaluate their two comparing to either an adaptive or adaptable
interfaces combined together with the adaptive approach with the Mixed-Initiative approach through
menus in the commercial word processor Microsoft empirical studies. On example of a such a
Word 2000. The two interfaces are a personalised comparison conducted by Debevc et al. (1996). They
interface containing desired features only and a compared between their adaptive bars with the built-
default interface with all the features only. The first in toolbar present in MSWord. Their results showed
four weeks of the study participants used the that the mixed-initiative system improved
adaptable interface, then the remaining for the significantly the performance in one of two
adaptive interface. 65% of participants prefer the experimental tasks. Bunt et al. (2007) designed and
adaptable interface and 15% favouring the adaptive implemented the MICA (Mixed-Initiative
interface. The remaining 20% favouring the Customisation Assistance) system. Their system
MsWord 2000 interface. This work extends by provides users with an ability to customise their
Findlater and McGrenere (2004) and they compared interfaces according to their needs, but also provides
between the static, adaptable, and adaptive menus. them with system-controlled adaptive support. Their
Their result concludes that the static menu was faster results showed that users prefer the mixed-initiative
than the adaptive menu and the adaptable menu was support. Also, it shows that the MICA’s

192
 A PLATFORM FOR INVESTIGATING EFFECTIVENESS FOR STATIC, ADAPTABLE, ADAPTIVE, AND
MIXED-INITIATIVE ENVIRONMENTS IN E-COMMERCE

recommendations improve time on task and decrease items. The two algorithms are used by Microsoft
customisation time. (Findlater and McGrenere, 2004) and suggested by
the literature (Findlater and McGrenere, 2004). For
our experiment, once the user clicks the items they
2 THE EXPERIMENTAL will move up to the top of the list (See Figure 1).
PLATFORM 2.3 The Adaptable Platform
An experimental e-commerce web-based platform The layout, content and item position on the list is
was developed to be used as a basis for this changed by the user during the course of usage. Our
empirical study. The platform provided four types of goal was to make the customisation process as easy
interaction conditions: Static, Adaptable, Adaptive, as possible. Therefore, the Coarse-grained and Fine-
and Mixed-initiative. The structure of the platform is grained (Findlater and McGrenere, 2004)
similar to many e-commerce web-based platforms. customisation techniques were utilised by allowing
The difference between the four conditions applied the user to move items to a specific location (See
to the contents, layout, and item position on the list. Figure 2). However, the main page provides two
choices for the user to choose from. The first choice
2.1 The Static Platform is an empty page that is left to the user’s decision as
to which content to add in. The second choice is full
The layout, content, and item position on the list content that has already been suggested. This is
does not change during the course of usage. Our because some of the early studies suggested a need
goal was to design the ideal platform to do the to examine full-featured interfaces versus reduced
required tasks as efficiently as possible. interfaces. However, when the participant started,
four items were displayed as a default in each web
2.2 The Adaptive Platform part of the home page. Subjects can increase the
number of displayed items as many items as they
The layout, content and item position on the list does like and reduce the number of displayed items not
change by system during the course of usage. less than one item. In addition, subjects can sort the
Adaptation helped users to find items by changing web contents by item name, id and price and the user
content to their preferences. Our goal was to design can also search in different sub-categories. Subjects
the most predictable personalised approach as can add new content to the home page, delete, and
possible. move an existing content to different positions.

Lock and Unlock list

Once clicked item moved

up to top of the list.

Move items up
or down to a
specific position

Figure 1: Adaptive list.

Therefore, the adaptive approach algorithm

dynamically determines item position on the list Figure 2: Customisable list.
based on the most frequently and recently used

193
ICE-B 2008 - International Conference on e-Business

2.4 The Mixed-Initiative Platform requirements is more than four and up to six
requirements. The users are guided to the list but not
In the Mixed-Initiative condition the control is the subcategory, so it is the user’s responsibility to
shared. Therefore, our goal was to make sure the search for items in the subcategory.
control is shared as fairly as possible. The Mixed- For the difficult tasks the number of available
Initiative condition algorithm is dynamically items is one item within a list that consists of more
determined based on the most frequently and than 40 items. The items are positioned at the middle
recently used items. However, to allow users to take of the list, to make sure that users can find the item
control, a new function was implemented to lock and even if the list changed. The number of requirements
unlock item movement (See Figure 2). Items will be is more than seven. In the difficult tasks there is no
moved up to the top of the list when clicked three guidance to items, so it is the user’s responsibility to
times, even if the list locked. Initially, when the search for items in all lists and all subcategories.
website is loaded the default content of the home
page is personalised. However, organising the list is
the user’s responsibility along with locking the lists. 4 SUBJECTS
These environments were tested empirically by four
3 EXPERIMENTAL DESIGN independent groups, each group consisting of 15
users. All the groups were asked to accomplish the
The experimental platform was tested empirically by same group of tasks (three easy tasks, three medium
four independent groups, consisting of 15 users. All tasks, and three difficult tasks) and a one learnable
the groups of users were asked to accomplish the task before starting each group. Each user attended
same 12 tasks. These tasks were designed with three a five minute training session about their
complexity levels: easy, medium, and difficult. In environment before doing the requested tasks. A
order to avoid the learning effect, the order of the pre-questionnaire conducted before the experiments
task complexity was varied between subjects. The to obtain users personal information. All users were
number of available items, item position (location) between the ages of 18 and 40. 44 of them were
in the list, number of requirements and guidance was male, while the remaining 16 were female. 70% of
considered when designing the tasks, i.e. more than them were postgraduate students. Most of the
three items available within a list that consists of a participants used the internet for 10 hours or more a
maximum of 20 items. The items are positioned at week. 85% stated that they do not customise new
the top, middle and at the end of the list. Thus users software unless they have to; the remaining 15 stated
can find the item even if the list changes. The that they do so. Also, 32% never used any
number of requirements is less than four. The users customisable web pages, where 17% used it once,
are guided to the list by providing the name of the and just four participants used it every time they
list and the subcategory. went online.

Table 1: Tasks design.

5 RESULTS AND DISCUSSION
Effectiveness was measured by calculating the
percentage of users who completed (learning and
completion) tasks along with the percentage of
(learning and completion) tasks completed by all
users. To compare the effectiveness between the
four conditions, three critical time limits for task
completion was derived for each level of tasks (easy,
medium, and complex). Therefore, a task would be
regarded as successfully completed if users
For the medium tasks, the number of available completed the task within the level critical
items is reduced to two items within a list that completion time.
consists of more than 30 items. The items are However, it was noticed that during the
positioned at the middle of the list. The number of experiment users who participated in the evaluation

194
 A PLATFORM FOR INVESTIGATING EFFECTIVENESS FOR STATIC, ADAPTABLE, ADAPTIVE, AND
MIXED-INITIATIVE ENVIRONMENTS IN E-COMMERCE

Figure 3: Searching Tasks.

of the adaptable and mixed-initiatives were more there was a significant difference between the
confident than the static and adaptive. Also, users numbers of users who completed all tasks between
got confused while participating in the evaluation of the adaptable and static conditions (t (3) = 3.04,
the adaptive and static conditions. This confusion cv=3.1) and between the adaptive and static
made them spend time on understanding what is conditions (t (4) = 4.5, cv=2.7). Figure 1 shows the
happening around them. Overall, just 8 users did not percentage of tasks completed by all users in each of
complete all tasks using the Mixed-Initiative the four conditions. However, the number of the
whereas 23 users did not complete all tasks using the tasks completed by all users was calculated to obtain
adaptive condition. In the adaptable condition, 24 an overall percentage. The result showed that the
users did not complete the all tasks whereas only 2 number of tasks not completed by all users was 8
users did complete all tasks using the Static tasks by using the Mixed-Initiative, 33 tasks by
condition. This shows that the overall number of using the Adaptive, 44 tasks by using the Adaptable,
users who completed all tasks in the Mixed-Initiative and 83 tasks were not completed by using the Static.
is higher than the other conditions. An ANOVA In the learnable tasks, there was a difference
result showed a significant difference in the number between the four conditions (See Figure 4). This
of users who completed the tasks at 0.05 (F = (3, difference was found to be statistically significant at
11), p <0.004). The users who completed the easy, 0.05 by using the ANOVA test. T-Test results
medium and complex tasks using the Mixed- showed that there was a significant difference at
initiative condition is higher than the other 0.05 between the number of tasks completed by all
conditions (Static, Adaptive, and Adaptable), users using the Mixed-initiative condition, compared
excluding the users who completed the medium to the Static condition (t(3)=11.3, cv=3.1) but not to
tasks using the Adaptive condition. the adaptive (t(2) = 2.6, cv = 4.3) and adaptable
Overall, t-test was used to find out the difference conditions (t(2)=3.1, cv=4.3). In addition, there was
between the four conditions. t-Test results showed a significant difference between the Adaptive and
that there was a significant difference of 0.05 Static conditions (t(4) = 4.5, cv=2.7). However, the
between the number of users who completed all number of users who completed all learnable tasks
tasks using the Mixed-initiative condition compared by using the adaptable condition was 11, which was
to the adaptable (t(3)=4.38, cv=3.1) and static higher than the other conditions. Following this was
(t(3)=11.3, cv=3.1) conditions, but nothing the mixed-initiative where 9 users completed their
significant was found when compared to adaptive all learnable tasks, and the Static condition (3 users).
(t(3)=2.04, cv=3.1). The users who completed the The users who completed all tasks using the
tasks using the adaptable and adaptive conditions are adaptive condition were lower (2 users) than all
higher than the static condition. Also, it was found other conditions. The percentage of users who
that the adaptable are higher than the adaptive in completed all tasks using the mixed-initiative
easy tasks and lower in medium tasks. Furthermore, condition was higher than the adaptive and static

195
ICE-B 2008 - International Conference on e-Business

conditions but not higher than the adaptable Proceedings of the 9th international conference on
condition. The main reason behind this is that Intelligent user interfaces, 117-124.
sometimes items’ positions in the lists changed Bunt, A., Conati, C. & Mcgrenere, J. (2007) Supporting
without users’ noticing which caused them interface customization using a mixed-initiative
approach. Proceedings of the 12th international
confusion. conference on Intelligent user interfaces, 92-101.
Debevc, M., Meyer, B., Donlagic, D. & Svecko, R. (1996)
Design and evaluation of an adaptive icon toolbar.
User Modeling and User-Adapted Interaction, 6, 1-21.
Findlater, L. & Mcgrenere, J. (2004) A comparison of
static, adaptive, and adaptable menus. Proceedings of
the 2004 conference on Human factors in computing
systems, 89-96.
Fink, J., Kobsa, A. & Nill, A. (1998) Adaptable and
adaptive information provision for all users, including
disabled and elderly people. The New Review of
Hypermedia and Multimedia, 4, 163-188.
Gajos, K., Christianson, D., Hoffmann, R., Shaked, T.,
Henning, K., Long, J. J. & Weld, D. S. (2005) Fast
and robust interface generation for ubiquitous
applications. Proc. of Ubicomp, 37-55.
Figure 4: Learnable Tasks. Greenberg, S. & Witten, I. H. (1985) Adaptive
personalized interfaces—A question of viability.
Behaviour & Information Technology, 4, 31-45.
Holzinger, A., Kickmeier-rust, M., & Albert, D. (2008)
6 CONCLUSIONS Dynamic Media in Computer Science Education;
Content Complexity and Learning Performance: Is
This paper described an empirical study that was Less More? Educational Technology & Society,, 11,
performed to investigate the effectiveness of the 279-290.
Adaptive, Static, Adaptable and Mixed-initiative Mcgrenere, J., Baecker, R. M. & Booth, K. S. (2002) An
conditions. In this investigation, the aim was to evaluation of a multiple interface design solution for
bloated software. Proceedings of the SIGCHI
assess the effectiveness of these four conditions.
conference on Human factors in computing systems:
One of the more significant findings to emerge from Changing our world, changing ourselves, 164-170.
this study is that Mixed-Initiative approach was the Mcgrenere, J., Baecker, R. M. & Booth, K. S. (2007) A
best in terms of effectiveness in the searching tasks field evaluation of an adaptable two-interface design
but not with the learnable tasks. In the learnable for feature-rich software.
tasks the adaptable was better than all other Mitchell, J. & Shneiderman, B. (1989) Dynamic versus
approaches. In addition, the Static and adaptive static menus: an exploratory comparison. ACM
conditions were found to be less effective than the SIGCHI Bulletin, 20, 33-37.
other conditions in terms of number of tasks Robert, T. & Dermot, P. B. (1987) A self-regulating
adaptive system. SIGCHI Bull., 18, 103-107.
completed by all users and number of users who
TE’ENI, D. & FELDMAN, R. (2001) Performance and
completed all tasks. Further work needs to be done satisfaction in adaptive websites: An experiment on
to establish whether the presence and absence of searches within a task-adapted website. Journal of the
multimodal metaphors on the mixed-initiative Association for Information Systems, 2, 1-30.
approach will help to make the most of the adaptive Thomas, C. G. & Krogsaeter, M. (1993) An adaptive
and adaptable advantages, at the same time as environment for the user interface of Excel.
reducing their disadvantages. Proceedings of the 1st international conference on
Intelligent user interfaces, 123-130.

REFERENCES
Alotabibi, M. & Alzahrani, R. (2004) Evaluating e-
Business Adoption: Opportunities and Threats. Journal
of King Saud University, 17, 29-41.
Bunt, A., Conati, C. & Mcgrenere, J. (2004) What role can
adaptive support play in an adaptable system?

196
 AN ENHANCED SERVICE PROVIDER COMMUNICATION
INTERFACE WITH CLIENT PRIORITIZATION
Case Study on Fast-food Chain Restaurants

Slobodan Lukovic
ALaRI, University of Lugano, Lugano, Switzerland
[email protected]

Nikola Puzovic
Department of Information Engineering, University of Siena, Siena, Italy
[email protected]

Milos Stanisavljevic
Microelectronic Systems Laboratory, EPFL, Lausanne, Switzerland
[email protected]

Keywords: e-Commerce, client prioritization, mobile devices, workload scheduling.

Abstract: With the increased dynamics of modern life, the efficiency and reliability of everyday services is emerging to
be a fundamental concern. On the other hand, modern telecommunication technologies, like wireless Internet
access, are penetrating all segments of our life. However, many every day activities and services still do not
fully exploit new technologies. We propose an approach that enables increased deployment of E-commerce
concepts in the fields where their usage was either small or negligible. Moreover, in the scope of the same
concept, we introduce prioritization of clients in services where it was not commonly present to date.
A solution for enhanced communication interface between service provider and customers is developed. As
a case study, the system is designed and optimized for an implementation in a fast-food chain. The proposed
solution is aiming at increasing of quality of service for customers, and at the same time increasing the op-
erational efficiency of the provider. The main idea behind this approach is to enable customers to use their
mobile devices, such as cell phones or PDAs, for browsing offered services or goods, viewing current service
conditions and placing orders. We will detail theoretical concepts underneath and describe the implementation
on both server and client side.

1 INTRODUCTION vices. Other broadband wireless services like UMTS

and WiMax are becoming more widely available and
Millions of people daily face different kinds of prob- affordable and the number of mobile users that will
lems in common situations. These issues range use these services will increase. Moreover, the num-
from unpredictable traffic and parking problems to ber of online payments is constantly increasing and
big waiting times in restaurants due to inefficient this way of payments is expected to be widely ac-
order/payment service. The rapidly growing urban cepted in very near future. Putting all together, inte-
population additionally increases pressure on service gration of all aforementioned services is logical con-
providers. We look for a response for these problems sequence of technology development and of the evo-
by incorporating e-Commerce concepts into service lution of users’ habits.
provider - client chain. In this paper we detail, based on case study on
The basic idea of service order/payment process fast food chain, a possible solution for incorpora-
automation relies on the rapid increase of number of tion and automation of the service scanning, or-
portable devices that are able to access Internet, and dering, payment and execution that increases effi-
on the growth of the popularity of e-payment meth- ciency, cutting the costs and bringing many other
ods. It is estimated that in the next two years all manu- benefits for all sides involved in the process. Our
factured cell-phones will be equipped with WiFi mod- aims are the improving of existing Quality of Ser-
ules which will boost the availability of Internet ser- vices (QoS) in fast food restaurants, and at the same

197
ICE-B 2008 - International Conference on e-Business

time providing greater flexibility for both the ser- in some cases even free of charge (Smithers, 2007).
vice provider and consumers. We propose a fully At the same time online payment methods are getting
integrated system that incorporates many different widespread across wide range of activities (Weiner,
technological aspects, ranging from Internet brows- 2000),(Ghosh and Li, 2007). Coffees and restaurants
ing and e-payments to service performance evalua- have made steps towards exploitation of WiFi at local
tion and workflow scheduling. The final result is a service points (Friedman and Parkes, 2003) and some
novel communication interface that brings many new fast-food chains started using touch-screens deployed
features and benefits to the client, like better service at tables (eTable) for offer browsing and order placing
overview, waiting time prediction (service availabil- (VanLeeuwen, 2005).
ity overview), prioritizing of clients, multilingual sup- We propose a novel solution for prioritization of
port, etc. On the other side, it cuts costs for service clients using and enhancing different experiences and
providers in terms of staff reducing, better insight in implementing them in fields where until now these
demand and market overview coupled with market technologies have not been commonly used. In order
profiling and targeted marketing. For now no experi- to provide better quality of service we model priori-
mental results are provided. tized orders execution and service provider capacity
The paper is organized as follows: In Section 2 with a well know operating systems task scheduling.
we summarize current state of technology and differ- The theoretical concepts adopted for purposes of our
ent types of its usage that are in scope of interest of work will be discussed in detail in Section 4.
this work, in Section 3 we present an overview of the
overall architecture of the system. Algorithms and
techniques used in realization of proposed concepts
are detailed in Section 4. Finally, Section 5 presents 3 SYSTEM OVERVIEW AND
conclusions and future work. IMPLEMENTATION
CONCERNS
2 STATE OF THE ART We envision a system that integrates WLAN access,
priority order scheduling based on demand prediction
In parallel to development of communication tech- and delivery automation that provides user-friendly
nologies many different service providers have been interface to the client. For the purpose of the client-
adopting them to facilitate the interaction with clients, server communication we propose WiFi wireless In-
to provide new services or increase operational effi- ternet access technology since it is widely used and
ciency. On the other side mobile technologies such already deployed at many fast food points around the
as WiFi, UMTS or WiMax as well as web access world. WiFI access represent an optimal compro-
standards and protocols are more and more oriented mise between simplicity and efficiency. The cover-
toward better support for increased need for mo- age, throughput and level of security are considerably
bile Internet availability. In that sense many mo- greater than in the case of Bluetooth. On the other
bile web standards such as Wireless Markup Lan- side it is very easy to deploy and cheap to use (from
guage (WML), Extended Modeling Language (XML) clients’ point of view it is free) in comparison to other
or XHTML etc. have been developed. Moreover, the broadband services as UMTS and WiMax. The range
development of AJAX (AJAX, 2008) has enabled ex- of the WiFi AP also makes it perfectly suitable for use
change of small amounts of data between client and in a local service point (a restaurant in this case). Nev-
server, hence increasing the interactivity, speed and ertheless, the system is conceived in such way that it
usability without the need to reload the entire contents can be easily ported to other communication medi-
of web page. This is especially valuable when de- ums. By using secure communication and already es-
vices have scarce computational and communication tablished methods of electronic payment, we will also
capabilities. By having more and more mobile de- provide high level of security that is necessary.
vices online many new concepts got enabled. This es- Figure 1 shows the overall organization of the sys-
pecially concerns social networking (Eagle and Pent- tem. The main components are servers in service
land, 2005), mobile commerce (Varshney et al., 2000) points, central data warehouse and a connection to
or intelligent wireless web (Alesso and Smith, 2001). e-payment servers. The local servers are responsi-
The most recent world-wide trend regarding wire- ble for processing clients’ orders and updating cen-
less Internet access is deploying of WiFi Internet ac- tral data warehouse. Payments are performed using
cess at variety of locations such as airports, hotels, secure connection to e-payment servers or by using
restaurants and so on (Friedman and Parkes, 2003); in-house e-payment system. In the following subsec-

198
 AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT PRIORITIZATION - Case
Study on Fast-food Chain Restaurants

Service Point 1
Local server User Server e-payment server

Enter Network
Send initial web page Prepare data

DB Browse and
Central data select services Send order
. warehouse
. Internet Send e-payment request Process request
. Redirect
Service Point n Send e-payment credentials
Local server Payment processed Process
payment
Order processed Confirm
order

e-payment
DB
servers Figure 2: Interaction between the client and the system.

Figure 1: Proposed implementation of the system. delity card, and can be used exclusively for payments
inside the system, and they function in form of deposit
tions and we will also discuss the system from the or credit.
participants’ point of view (both clients’ and service Once the payment is completed and the delivery is
providers’), we will discuss the technology involved ready, it is placed in a delivery slot and the client re-
together with different implementation aspects. ceives a code associated with the order. The client
types the code using keyboard on the delivery slot
3.1 Client’s View of the System that is labeled with appropriate order number or with
clients name. These slots can be implemented as
Client devices are PDAs or smart phones that are ca- parts of rotating table, where each slot has a protec-
pable of running a web browser. User interface is im- tive cover. Once the correct code is inserted, the ser-
plemented in HTML, and it is accessible from each vice is considered to be completed. The time between
web browser, so there is no need to install a special order placement and code insertion is considered as
purpose application. Figure 2 shows the UML se- ’service time’. This information is taken and used as
quence diagram of the users’ interaction with the sys- correction factor for statistical processing and for cal-
tem. At the beginning, the client chooses appropri- culation of the expected time of servicing.
ate WiFi Access Point (AP) that will consent him to
access the services provided in the restaurant. Upon 3.2 Service Provider’s View of the
the entry to the network, clients are redirected to the System
start page from where they can browse the available
services and current conditions of the system. There The system in a service point processes e-Commerce
is a possibility to supply the client with additional orders in parallel with traditional ones (Figure 3). Re-
information that traditional ordering methods cannot quests from both sides go to the same server and their
support (i.e. expected waiting time, multilingual in- execution is scheduled in order of submission. The lo-
terface etc.). The calculated expected waiting time cal server that processes orders is responsible for han-
coupled with delivery automation brings a possibility dling client requests and for scheduling the delivery of
of introducing prioritization in the fast food service. orders. Scheduling and delivery are performed taking
For more details see Section 4. into account also the priority of the clients. The al-
From the start page clients will continue to the gorithm for calculating waiting times is applied each
menu with products that are available in the restau- time a new client enters the restaurant and this infor-
rant. Trough simple web forms clients can make a mation is communicated to the client.
choice, and communicate the selection to the server The technology used in the service point is rather
in the service point when the order is completed. The simple and cheap, and it involves a server computer,
system sends back the information about the order wireless AP and Internet connection that will be used
and order number. The payment can be done through for communication with the central server. The algo-
SSL secured connection using credit-card payment, rithms that are used are described in subsequent sec-
pay-pal and other methods of e-payment. In this case, tions.
a request for payment is communicated to the pay-
ment gateway (paypal server or bank server in case
of credit card payment) that processes it and gives the
confirmation. Another option is to use vendor pro-
vided vouchers that can be issued in the form of fi-

199
ICE-B 2008 - International Conference on e-Business

related to times spent in different phases of the order

processing process:
• Queuing time - time spent in the line/queue while
waiting for the order placing.
• Service execution time - time required to execute
the requested service by the server.
• Service waiting time - the time that passes from
the moment of performing the order to the mo-
ment in which the order is completed.
• Total service time - the time from the moment
Figure 3: Overview of the service points’ organization. the client has entered service point to the moment
when the required services are obtained.

4 SERVICE DEMAND AND 4.2 Classification of Services by

SERVICE TIME PREDICTION Processing Power Requirements
MODEL
Almost all orders differ from each other in type of
In this section we present mathematical concepts pro-
services requested, its quantity and quality which re-
posed for calculation of expected average service time
sults in different service processing power require-
considering prioritized clients. The main contribution
ments. Any order can be further decomposed in set of
reflects in the exploration of scheduling mechanisms
’atomic services’. We assume ’atomic service’ to be
in order to meet the deadlines - guarantee service. We
the simplest possible single order (i.e. an ice cream
briefly expose possible service classification concepts
or coffee). In that sense every order can be seen as
and compare them. The chosen concept is described
a composition of various ’atomic orders’. The most
in detail.
precise way of modeling the orders would be repre-
senting them in form of number of atomic services
4.1 Definition of the Server and Client requests. Unfortunately this method would introduce
huge computational overhead and the entire system
We consider the server - service execution unit that would be very complex to implement. For the sake
takes the order, executes it and performs the deliv- of the simplicity and better efficiency but without los-
ery of the requested service or product. The server ing generality we classify orders according to number
is characterized as multiple parallel process execution of atomic services requested into five classes ranging
unit with a given total capacity - service power. The from small to huge ones. Each of those order types
clients could be clasified in three groups with two lev- are assigned certain evaluated average service time.
els of priority: This evaluation is constantly updated by newly ob-
1. Traditional clients served at the service line in tained data from the system.
a traditional way (worker at the counter table is
serving one client after another from the FIFO -
4.3 Service Time Prediction Model and
first in, first out, queue), taking low priority (pri-
ority equal 0). Scheduling for a Guaranteed
2. Regular e-Commerce clients, performing orders Execution
through wireless system, taking low priority.
3. Prioritized e-Commerce clients, performing prior- Here we describe empirical model for run-time calcu-
ity service orders through wireless system, taking lation of service expected time. This time is provided
high priority (priority equal 1). to clients as additional information and this property
of the system is one more original feature that en-
Initially a certain server capacity is given to the prior-
hances existing services. The service time is modeled
itized clients. This capacity is dynamically adjusted
as a statistical variable with two components:
during service execution in order to guarantee pro-
jected waiting time for prioritized clients and there- 1. Non-stochastic component measuring the neces-
fore quality of service (QoS). We have adopted fol- sary service time for non-prioritized and priori-
lowing naming convention to describe different issues tized services with a given scheduling scheme.

200
 AN ENHANCED SERVICE PROVIDER COMMUNICATION INTERFACE WITH CLIENT PRIORITIZATION - Case
Study on Fast-food Chain Restaurants

2. Stochastic component that incorporates the addi-

tional service time related to the average num- Priority
ber of prioritized clients that are arriving during
Total
the time slot defined with non-stochastic compo-

Subtraction
Service
nent. The additional service time is added into Time Run
the scheduling scheme as a prioritized process and Laxity Interface Engine Time
Priority
Execution
a total service time is acquired after the probe Time
scheduling execution. This component is acquired Server
by processing empirical data related to number of Capacity
Time
prioritized clients and their orders that are dynam-
ically updated every day. During the phase of ini-
tial system deployment these data are unavailable Figure 4: Inference model
and therefore a worst case estimation is taken in-
stead of statistical component.
being executed; laxity = total service time - service
The order execution inside the server is scheduled ac- execution time) among all the ready ones whenever
cording to non-preemptive task execution scheduling. a processing server becomes idle, and executes it to
We consider this scheme to be the most suitable in completion.
case of fast food restaurants as preemptive service Lee et al. (Lee et al., 1994) presents a fuzzy
would cause disorder in execution of currently exe- scheduling algorithm. Their proposed algorithm uses
cuted services and also would require different work- task laxity and task criticality as system parameters
ing concept inside the server (i.e. specialization to and doesn’t consider fairness. Their simulation model
atomic tasks execution and work division that could contains small number of tasks on a uni-processing
support this would result in inefficient use of working unit system and they did not consider system over-
power). loads.
In case that prioritized tasks are not schedulable Chen et al. (Chen et al., 2005) proposed a schedul-
within dedicated server capacity for prioritized tasks, ing model and a related algorithm that is suitable
the amount of dedicated server capacity is increased for both uni-processing and multiprocessing servers.
to the minimal value that guarantees schedulability of They provide a method to detect work overloading
prioritized tasks. This has a consequence of increase and try to balance load with task dispatching. We pro-
in waiting time for non-prioritized clients. However, pose to use model presented in (Hamzeh et al., 2007)
QoS for prioritized clients is guaranteed. using a fuzzy interface engine. The model we pro-
Any practical scheduling algorithm in multipro- pose has a slight modification considering that there
cessing server systems presents a trade-off between are only two levels of priority defined as ”high” and
performance and computational complexity. How- ”low”. As shown in Figure 4, the major factors con-
ever, in our case scheduling computation time is not sidered in used approach to determine the scheduling
an issue (because it can be in the range of seconds) are task priority, total service time, service execution
and we can explore the more complex scheduling time, and used server capacity time. The notion of
schemes. Scheduling could be regarded as soft real- laxity is used in the proposed approach to facilitate
time or even non real-time problem. The Earliest the computation.
Deadline First (EDF) algorithm is the most widely In proposed algorithm as shown in Figure 5, a
studied scheduling algorithm for real-time systems newly arrived task will be added to the input queue.
(Balarin et al., 1998). This queue contains the remaining tasks from last cy-
EDF is more efficient than many other schedul- cle that has not yet been assigned.
ing algorithms, including the static Rate-Monotonic Fuzzy scheduler processes each task separately,
scheduling algorithm. However, when the process- computes its run-time priority and sends it to task dis-
ing server is overloaded (i.e., the combined require- patcher’s priority queue. In a multiprocessing system,
ments of pending tasks exceed the capabilities of the this queue offers tasks to dispatcher by their run-time
system) EDF performs poorly. Researchers have pro- priority order (as shown in Figure 6). Dispatcher of-
posed several adaptive techniques for handling heav- fers a new task whenever one of the processing units
ily loaded situations, but they require the detection of of the system finishes its current task.
the overload condition. Least Laxity First (LLF) al-
gorithm (Ramamritham and Stankovic, 1994) is non-
preemptive and selects the task that has the lowest
laxity (the maximum time that a task can wait before

201
ICE-B 2008 - International Conference on e-Business

1. For each task in input queue will achieve precise insight in offer and service con-
(a) Feeds task’s run-time priority using fuzzy infer- dition, the queuing time will be eliminated and they
ence engine will have the possibility to get delivery in short time
2. While a server has a free processing unit as prioritized users. They will also have more pay-
ment options, and multilingual interface.
(a) assign the task with highest run-time priority to Our future work will focus on collecting and
the processing unit thoroughly analyzing statistical data. In the early
3. Loop forever phase of system deployment, realistic assumptions for
worst case scenario needs to be made. Also, tun-
(a) If a processing unit event occurs endenumerate
ing of scheduling algorithm needs to be performed
i. Go to 2 with a detailed testing with realistic data versus other
(b) If scheduling event occurs scheduling schemes.
i. Update tasks parameters
ii. Go to 1

Figure 5: Proposed algorithm

REFERENCES
Process Unit AJAX, A. (2008). http://www.asp.net/ajax/.
Task Dispatcher
Task Scheduler

Task Queue Priority Queue Process Unit Alesso, P. and Smith, C. (2001). The intelligent wireless
Process Unit
Web. Addison-Wesley Prof., London, 2nd edition.
Balarin, F., Lavagno, L., Murthy, P., and Sangiovanni-
Process Unit
vincentelli, A. (1998). Scheduling for embedded real-
time systems. IEEE Design and Test of Computers,
15(1):71–82.
Figure 6: System view of real-time fuzzy scheduler Chen, G., Chen, G., Ozturk, O., and Kandemir, M. (2005).
An adaptive locality-conscious process scheduler for
embedded systems. In Proc. 11th IEEE Real-Time and
Embedded Technology and Applications Symposium.
5 CONCLUSIONS AND FUTURE
Eagle, N. and Pentland, A. (2005). Social serendipity: Mo-
WORK bilizing social software. IEEE Pervasive Computing,
4(2):28–34.
In this work we have proposed a solution for deploy- Friedman, E. and Parkes, D. (2003). Pricing wifi at star-
ment of e-Commerce concept in the fast food restau- bucks – issues in online mechanism design. In Proc.
Fourth ACM Conf. on Elec. Commerce (EC’03).
rant chain that brings more convenience for both ser-
vice provider and clients. The novelty of the work Ghosh, S. and Li, X. (2007). Critical factors affecting
the acceptance of e-payment services in emerging
lays in possible implementation of client prioritiza- economies. In Proc. Decision Sciences Institute Con-
tion based on a well-known computer science concept ference.
of operating system task scheduling, that will perform Hamzeh, M., Fakhraie, S. M., , and Lucas, C. (2007).
the best under the assumption that we have enough Soft real-time fuzzy task scheduling for multiproces-
information considering stochastic component of ser- sor systems. International Journal of Intelligent Tech-
vice time. For such a credible statistics we need to nology, 2(4):211–216.
have the insight of a system in a long run. The pro- Lee, J., Tiao, A., and Yen, J. (1994). A fuzzy rule-based
posed system brings numerous benefits to both parties approach to real-time scheduling. In Proc. 3rd IEEE
involved in the service process. Conf. Fuzzy Systems.
The system will boost the efficiency of the ser- Ramamritham, K. and Stankovic, J. (1994). Scheduling al-
vice by eliminating ordering waiting time and will cut gorithms and operating systems support for real-time
costs by decreasing staff needed for order acceptance systems.
and delivery performing. It will also increase the vis- Smithers, R. (2007). Mcdonald’s to offer free wi-fi in
ibility (if coupled with Internet - browsing and posi- restaurants. Guardian.
tioning) of the services, and enables better demand in- VanLeeuwen, S. (2005). Taking food service into 21th cen-
tury. In eTable Inc.
sight that brings more flexibility. Moreover, it will en-
able market profiling and targeted profiling, and gives Varshney, U., Vetter, R. J., and Kalakota, R. (2000). Mobile
commerce: A new frontier. Computer, 33(10):32–38.
the possibility to offer more e-services according to
clients’ need. Weiner, S. E. (2000). Electronic payments in the u.s. econ-
omy: An overview. Computer.
There are many benefits for the clients also. They

202
 MOBILE BUSINESS EXPERT ADVISOR

Danco Davcev, Marjan Arsic and Dalibor Ilievski

Faculty of Electrical Engineering and Information Technologies
University “Ss Cyril and Methodius”, Skopje, Macedonia
[email protected], [email protected], [email protected]

Keywords: Mobile Expert, on-demand knowledge, Web services, XML Agents, Fuzzy logic, Pocket PC.

Abstract: In this position paper, we introduce the service which enables consultations by the Experts and/or web
services via mobile devices. The advanced development of wireless networks and mobile devices with
various connection features made a great substrate for the development of services which are based on
immediate response. One such service is a service for on-demand knowledge. A Helpdesk operator can post
a request for consultation to the Experts and/or web services. The goal is to make knowledge available on
demand at any time and any place. This approach gives faster problem solutions, more productive expert
and/or web services, and high availability of the knowledge.

1 INTRODUCTION demands are imposed to these devices: greater

processing power to support real time multimedia
The benefit of mobile devices combined with stable transfer (video stream, voice stream, and file
and cheap wireless networks made new kinds of exchange), greater memory space. Modern feature
services to appear. Experts are becoming more rich applications demand space on the screen for
available than ever. Their services can be easy displaying all available features, which could be an
utilized and customers will have direct answer to the issue when handheld devices are used.
required question. In some cases (if appropriate) one A great consideration should be given to the
of the XML agents communicates with the necessary readability and utilization of small mobile user
Web services to find the solution of the problem. interfaces. In this paper, we present our approach in
Exchange of knowledge and consultation process creating a Mobile Business Expert Advisor (MBEA)
among customers and available expert authority via for demanding and exchanging of various content
helpdesk operators are very important aspects of types and knowledge on mobile devices. The
quick problem solution using the mobile devices. interface of our MBEA adapts the dimensions and
The goal of the mobile technology is to make visibility of the user controls according to the user’s
information instantly available to customers. The preferences and utilization of the controls and also,
new business processes can benefit a lot by using the current usage of the specific media contents
mobile connection between customers and experts. within the business process. Our knowledge – based
The exchange of the various content types of media interface is managed by fuzzy logic and many XML
data (text, drawing, and sound) will enhance the based agents.
efficiency of consultations between the customer The related work in the second section gives
and the expert advisor. It gives benefits in reducing some comments to similar MBEA-s. The
of costs, avoiding future problems or possible architecture of our Mobile Business Expert Advisor
claims, saving time, increasing profit and customer is elaborated in the third section, which is the main
satisfaction. contribution of this paper. The fourth section
There are many desktop applications that describes the design of MBEA. Implementation and
support multimedia communication among Evaluation of MBEA are presented in fifth section.
participants. Communication among several Finally, in the sixth section we conclude the paper.
participants that includes multimedia transfer is at
the beginning of the development and utilization in
the sphere of mobile handheld devices. Hence, new

203
ICE-B 2008 - International Conference on e-Business

2 RELATED WORK appropriate), in our approach we use more flexible

protocol for communication among agents which is
The agent-based approach that uses fuzzy logic to based on fuzzy-based knowledge.
determine importance of certain information is In (Pasquale De Meo, 2007), an XML-based
elaborated in (David Camacho, 2001). The agents multiagent recommender system for supporting
described in this paper are based on access to the online recruitment services is proposed. Although
Web. We use similar approach to determine the main purpose of this system is to provide Online
importance of user interface features in regards to Recruitment Services, it is also agent and XML
user preferences and media contents used in current based and as a consequence, it can easily cooperate
session. However, in our approach is used wireless with company information systems. The so called
network access for communication between ontology of their user agent stores the profile of a
handheld devices. given user concerning the job search. It is different
In (Vlado Glavinic, 2007), an intelligent tutoring from our MBEA because we plan to build a general
system was presented. They use agents to recognize purpose advisor system (not only for recruitment
the device and the way user device connects to the services). For the time being, our objective is to have
global network and according to that information, a a self - adaptive system from the point of view of an
learning content is adjusted to the particular device efficient communication between user and the
and sent to the learner’s device. It differs from our advisor according to the user's preferences and the
system because we introduced agents whose purpose current possibilities of the communication system.
is to intelligently adapt the user interface to the In (Zhiyong Weng, 2007) a feasible framework
preferences of each user. that combines agent mobility and intelligence for
Intelligent Mobile Answering Service is given in consumer-oriented e-business applications is
(Business Wire, 2006). Customers can use free-form proposed. This framework complements the current
questions on any subject with natural language Web-based systems by adding the wireless channel
queries. They can send text messages containing of mobile agents. In our work the mobile agents use
questions in any form just as easy as they send text an adaptive communication protocol based on fuzzy
message to their friends and receive a specific logic. In addition, our framework includes web
answer to their mobile phones. The natural language services.
queries allow customers to ask questions in the way
they normally express themselves. Unlike this, our
approach is based on resolving business issues in 3 MOBILE BUSINESS EXPERT
various format types by the experts. ADVISOR (MBEA)
In (Lu, L. Kitagata, 2003) an agent based
adaptive user interface control for desktop ARCHITECTURE
applications is presented. Although we share similar
ideas, our approach is multi-agent based adaptive There are three kinds of users like customers,
interface which uses fuzzy logic for handheld helpdesk operator and experts. Helpdesk operator is
devices. the link between experts and customers.
In (Weichang Du, 2006), collaborative
applications can be built using two types of agent
collaborations, agent communications through
XACL and agent visiting. Agents are represented as
XML entities, not programming language entities.
Secondly, agent hosting services are implemented as
web services with published WSDL, not programs in
certain programming languages with published
APIs. Thirdly, although XML agents’ behaviors
have to be coded in some supported programming
languages, the interactions between agents and hosts
in agents’ behavior code are through invoking local
hosts’ web services, which is neutral to Figure 1: Architecture of the MBEA.
programming languages and host operating systems.
Although we also use XML entities as agents and The goal of the helpdesk operator is to response
we share similar ideas of using web services (when any kind of requests on demand by the customers in

204
 MOBILE BUSINESS EXPERT ADVISOR

appropriate formats (text, draw, VoIP) and to send and/or Web services and exchanges reformatted
the complex requests to the experts and/or web media types; Passive helpdesk Agent calculates the
services which should be resolved by them. Client helpdesk interface; Media Agent monitors the
applications are implemented on Pocket PCs. media contents (text, image or voice); Experts and
There is only one helpdesk operator and many helpdesk preferences are stored in XML data
experts in one consultation session. storage; VoIP Server is responsible for voice
The network communication is realized in a way transfer and UDP server application manages the
that clients send UDP datagrams to the server, and login, chat, draw and control. For the purpose of
the server resends the UDP datagrams, according to saving the memory and processing power of the
the contents of the received message. clients handheld devices, Passive Helpdesk Agents
The communication protocol between the agents and Media Agent reside on the server side. The
and the information flow rely on fuzzy logic. In this agents are XML based.
case, the fuzzy logic is used for enabling or
disabling (adaptation) some controls depending on
the quality of the service (QoS), i.e. the signal 4 MBEA DESIGN
strength. So, if the signal strength is good, the
features of chat, draw and VoIP will be enabled. In this paper, we describe our approach in
However, for medium signal the VoIP feature will development of a MBEA by using multi XML
be disabled. For bad signal, beside the VoIP, the agents and fuzzy logic. Helpdesk agent resides on
draw feature should be also disabled, and the only the client side, and monitors the helpdesk operator’s
active communication feature will be chat. interaction by using the features of the application
Streams are used for the voice transfer. Desktop and communicates with interoperability agent (e.g.
based clients (for both helpdesk operator and sends the request to the available competent experts
experts) can be also connected to the server. and/or web services, receives the response in
The Web services are used to find the solution of corresponding format and regulates the customers
the problem using the wireless Internet access. If the billing). The triggers that demand helpdesk agent’s
Web services are unable to find the solution, the actions are: drawing, sending chat messages, and
Experts are activated. Some types of Web services communicating using VoIP. This agent is named
are used for the responses. These Web services Helpdesk because it helps the customers to simply
support and convert various types of media data. resolve their requests. At the start of the
The experts interface has controls for chat, draw, communication session, helpdesk agent
VoIP, file upload/download, authority control (VoIP communicates with the Passive helpdesk agent and
channel). interoperability agent that reside on the server. The
The helpdesk operator can give the control over Passive agent demands the information from the
the interface (or by giving the speaking possibility) Media agent about the type of the files to be used for
and reclaim the control later. Helpdesk operator has the particular communication session. This agent is
the role of a moderator in the consultation process. named passive helpdesk since it waits to be invoked;
Figure 1 shows the architecture of the MBEA. it recalculates the interface structure and sends the
Client applications reside on pocket PCs and they data to the helpdesk agent which finally draws the
have three main parts: Helpdesk agent, which interface. The Expert Agent receive the requests in
manages operator interaction within the application some format (text, file, voice) from the
and communicates with interoperability agent. VoIP interoperability agent, concerns about the solution of
client serves for the voice transfer, and the client the requests, sends the solutions to interoperability
application represents all the features mentioned agent, modifies the expert’s profile, checks the QoS
above (chat, draw, voice, file up/download, and billing status.
communication among clients). Expert Agent The main tasks of the interoperability agent are:
receives and sends the resolved requests using receiving the requests from helpdesk agent and
various file types (like text, image, and voice). forwarding them to the expert agents and/or web
Collaboration Agent calculates the cost of the services. Also it exchanges reformatted data files
corresponding expert’s service. The communication with helpdesk agent. Media agent monitors the
is realized through wireless area networks folder with the data files for the communication
connected to the Internet. session, and sends this information to the Passive
The server contains Interoperability Agent helpdesk agent. The Passive helpdesk agent
which redirects the complex request to the Experts recalculates the helpdesk interface according to the

205
ICE-B 2008 - International Conference on e-Business

information from the Media Agent, the history of The GetAdapters function can be used to query
interface affinities (chat_aff, draw_aff, sound_aff, names of network adapters. It calls the built-in NDIS
QoS) for the particular user and the helpdesk agent’s (not NDISUIO) driver. The function fills a buffer
message about the use of the features. Then, it sends with adapter names separated by commas. The
the corresponding information (for the type of the RefreshBSSIDs function requests that the driver
helpdesk interface) to the Helpdesk agent which initiate an AP survey. It takes one parameter: an
adapts the helpdesk interface according to this adapter name. The GetBBSIDs function returns the
information. The collaboration Agent calculates the list of available stations, i.e. peers and Access
time session for realization of the expert task and the Points. The function getPreferredMACAddress
cost of the service returns the MAC address of the connected
Helpdesk and Expert profiles with personal (associated) Access Point.
information about the qualification and the
knowledge for all of them are stored in the XML
database. Additionally a history of affinities is stored
for the helpdesk operator and the experts.
According to the history, Passive helpdesk agent
adapts the helpdesk interface on the beginning of the
communication session.
There are currently three predefined user
interfaces, which can be preloaded on the client’s
device, according to his preferences and the media
contents for the current communication session. The
dimensions and the positions of the controls would
be stored on the helpdesk’s affinity table on the Figure 2: WiFi signal strength – class diagram.
server. One kind of an interface displays chat on the
larger part of the screen, the second interface has In order to create the MBEA for calculation of the
larger drawing view on the screen, and the third one user interface visual features, the following fuzzy
has equal space for the chat and the drawing space. variables are defined for this expert system: U_A
VoIP feature doesn’t occupy much space on the (user affinity) which presents the most used feature
screen, so the buttons for voice are always displayed by the user, i.e. it has the greatest affinity for the
and they will be active if the Signal Strength is feature; M_P (media profile) presents the profile of
satisfied, i.e. QoS is strong. the interface that should be used according to the
Since there is no universal conclusion for making media contents; QoS (Quality of Service) presents
decisions of which user interface should be the strength of the wireless signal. Output variable
preloaded, fuzzy logic approach is used to model named I (action for interface adaptation) presents the
that kind of the imprecise information. necessary interface to be preloaded onto the user’s
The interaction made by the helpdesk operator device screen, according to the input variables.
using the features of the application (chat, draw and The linguistic variable Quality of Service (QoS)
voice feature), is represented as a vector (chat_aff, accepts values from the set of terms {Strong,
draw_aff, sound_aff). A linguistic variable named Medium, Weak}.
RESULT_PREFERENCE is introduced, and it The linguistic variable U_A accepts values from
accepts values from the set of terms {increase_chat, the set of terms {Chat, Draw, VoIP} and it has
increase_draw, the_same}. This variable represents normalized values of utilization between 0 and 1
the visualization of the helpdesk interface, produced (0% to 100%). It is equivalent to the number of
by (1) the helpdesk operator interaction with the times of usage of the features chat, draw or VoIP
interface; (2) the contents of media folder and (3) the respectively.
quality of service of the signal strength that decides The linguistic variable M_P accepts values from
more accurately which interface to be preloaded. the set of terms {Text, Drawing, Sound}. It has
The class diagram for WiFi signal strength is normalized values of utilization between 0 and 1
shown on Figure 2. Two classes are developed, (0% to 100%), which is equivalent to the number of
WiFiPeek and SignalStrength. The WifiPeek class text, drawing and sound files in the media profile
implements all the Wi-Fi query related elements. respectively.
The class uses the NDIS User mode I/O driver In the table 1 is presented the simulation for
(NDISUIO) to perform Access Point (AP) scanning. fuzzy linguistic values for U_A, M_P and QoS and

206
 MOBILE BUSINESS EXPERT ADVISOR

Figure 3: Agents’ collaboration diagram.

the value of I is calculated, based on the fuzzy rules. The interaction among agents is shown on the Fig. 3.
The action needed to adapt the interface is described
with the following XML code like the example 1 of
the Table 1. 5 MBEA IMPLEMENTATION
<IF U_A="DRAW" M_P="DRAWING" QoS="WEAK"> AND EVALUATION
<ASSIGN I="THE_SAME" />
</IF> Applications for Wireless MBEA are developed in
C++ Visual Studio .NET 2008 development
In the first example from the Table 1, the draw framework with using of the MFC (Microsoft
affinity is more utilized than chat and VoIP and the Foundation Classes) library. Operating systems used
media profile has more drawing objects than text or on the pocket PCs are Windows CE, Windows
sound. It is expected to be increased the drawing Mobile. The clients which reside on desktop PCs
interface, but because the signal is weak and the and the server use the standard .NET Framework.
drawing features are not supported on such signal, Operating system for desktop machines on which the
the interface stay unchanged (the_same). The total application is practically deployed, is Windows XP
numbers of fuzzy rules are 27, the combination of SP2. The interface adaptation is realized on different
the values of all linguistic variables. The next screen resolutions.
examples are similar like the first one. The two instances of MBEA interface are shown
on Figure 4 for two different screen sizes of Pocket
Table 1: Some examples for adapting interface using fuzzy PC-s.
linguistic variables. The initial feedback of MBEA was given by 20
colleagues in ICT. They were asked to use the
system for stock exchange analysis in last week.
General opinion among participants for the interface

Figure 4: Two instances of MBEA.

207
ICE-B 2008 - International Conference on e-Business

usability, functionality and visibility is average Based Agents”, Systems, Man, and Cybernetics, IEEE
(80%). Interface is functional and suggestions for International Conference; pp. 2575-2580.
interface rearrangment are mostly done according to Vlado Glavinic, Marko Rosic, Marija Zelic, 2007.
“Agents in m-Learning Systems Based on Intelligent
the user needs.
Tutoring”, Proc. of the HCI Universal Access in
The questions that examine interface usability, Human-Computer Interaction, Beijing, China, pp.
functionality, visibility and provide information for 578-587.
future upgrades are: Business Wire, “AskMeNow Launches Intelligent Mobile
1) Are you satisfied with the way of adaptive Answer Service 'ASKME'”, CNET Networks, 2006.
presentation? http://findarticles.com/p/articles/mi_m0EIN/is_2006_
2) Can you easily select the parameters of MBEA? Sept_25/ai_n16838742 [2008]
3) Were the results of the MBEA solutions are Lu, L. Kitagata, G. Suganuma, T. Kinoshita, T., 2003.
clearly displayed? “Adaptive user interface for multimedia
4) How much the MBEA helped in the business communication system based on multiagent” Proc. of
process? the 17th International Conference on Advanced
5) Does the MBEA satisfy the quality of the Information Networking and Applications, IEEE
service? Computer Society, pp. 53- 58.
Weichang Du, Hui Li, 2006. “XML Agents Technology
The questions are answered with “Yes”, “No” or “I
for Building Collaborative Applications”,
Don’t Know”. The answers are given in the Table 2. Collaborative Technologies and Systems, CTS
International Symposium, Pages: 289 – 297.
Table 2: Results of the questionnaire. Pasquale De Meo, Giovanni Quattrone, Giorgio Terracina,
Average Domenico Ursino, 2007. “An XML-Based Multiagent
Question Q1 Q2 Q3 Q4 Q5
Percent System for Supporting Online Recruitment Services”,
Yes 17 15 16 14 18 80% Systems, Man and Cybernetics, Part A, IEEE
Transactions, Volume 37, Issue 4, Pages: 464 – 480.
No 2 3 1 4 1 11%
Zhiyong Weng, Thomas Tran, 2007. “An Intelligent
I don’t know 1 2 3 2 1 9%
Agent-Based Framework for Mobile Business”,
Management of Mobile Business, 2007. ICMB 2007.
International Conference, Page: 30
6 CONCLUSIONS
This paper presents our approach that uses XML
agents and fuzzy logic in order to achieve adaptive
MBEA. We have developed such mobile system
which determines the features of the interface
according to the contents for the specific session.
Fuzzy logic approach is used for the communication
protocol between XML agents and for interface
adaptation. It can be also used to dynamically
reconfigure the interface according to the
preferences and the type of request. In this way, we
proposed an adaptable MBEA for handheld devices
which brings benefits to the developers of
applications for this software environment, to users
of mobile business systems and at the end, to all
mobile device users. General opinion among
participants for the MBEA usability is positive. In
the future work, we plan to provide an detailed
analysis of the user’s feedback.

REFERENCES
David Camacho, Cesar Hernandez, Jose M. Molina, 2001.
“Information Classification Using Fuzzy Knowledge

208
A MOBILE BUSINESS PROCESS DEPLOYMENT FRAMEWORK
FOR DEVICE INDEPENDENCE AND CONTEXT-AWARE
ENVIRONMENTS

Torab Torabi, Saqib Ali and Hassan Ali

Department of Computer Science and Computer Engineering
La Trobe University, Melbourne 3086, Australia
[email protected], [email protected], [email protected]

Keywords: Business processes, Location aware, device independence, context-aware.

Abstract: Deployment of a business process to mobile devices in different context and in a device independence
environment is a challenging task. The desirable is that a Business Process could be deployed to any mobile
device in any location or environment the user operates. One obstacle in business process deployment on
mobile devices is the rapid changes in mobile technologies. In this research we have developed and
implemented a mobile business process deployment framework that caters for both device independence
and context-aware environments. In this paper the focus is on how device independence and context-
awareness can be integrated for mobile business processes deployment.

1 INTRODUCTION identification (Kaasinen 2003) (Siau and Shen

2003). Mobility is the central distinguishing feature
The companies recently have been paying a great of mobile technology. As mobile devices constantly
deal of attention to the potential of mobile accompany their users, people can receive and send
communication technologies to redefine and extend data regardless of place and time. They can also be
the world of traditional E-Business by making its reached by people at all times. This feature is
applications more available to the mobile users. especially useful to logistic and supply chain
According to (Swaminathan and S.R.Tayur 2003), companies that regularly need to reach their mobile
E-business is defined as “a business process that workforce to allocate various tasks to them.
uses internet or other electronic medium as a The success of the recent applications is based
channel to complete business transactions”. not only on new technology, but rather on its proper
Mobile Business is often described as the use (Liang and Wei 2004). Wireless applications
successor of electronic business and defined as the would be advantageous only when they can be
subset of it (Adam, Chikova et al. 2005). It is useful to their end users (Kalakota and Robinson
assumed that any business operations performed by 2001). Mobile business value proposition originate
desktop computer can also be performed via from the fusion of the wireless technology with
wireless network. However mobile technology already available electronic business applications.
offers additional possibilities that are unique to the On the other hand, the unique features of mobile
wireless world and cannot be performed via fixed business – mobility, localization give rise to
network. For example, providing the mobile users emergence of completely new applications and
with the services of location aware and context- business models.
aware applications which cannot be performed with
a fixed internet connection (Zhang, Archer et al. 1.1 Mobile Business Processes
2003). Mobile business can be defined as the
exchange of goods, services, and information using according to Davenport (Davenport 1992) a
mobile devices (Paavilainen 2002) . business process can be defined as “a specific
Mobile Services are usually characterized by ordering of work activities across time and place,
mobility; reach-ability, localization, and with a beginning, an end, and clearly identified

209
ICE-B 2008 - International Conference on e-Business

inputs and outputs: a structure for action.” The independence is to develop ways for future web
business profit depends on efficient delivery of content and applications to be authored, generated,
goods and services controlled by business process or adapted for a better user experience when
(Ali, Torabi et al. 2006) (Ali, Soh et al. 2005). So delivered via many device types.
there is a need for the companies to make use of the Currently many companies typically design web
technologies to make their product more profitable applications for desktop browsers. And adaptation
and their services more efficient. of this kind of web site for small display is
Many companies have been able to make their effectively impossible and the companies creating a
traditional business processes into mobile business parallel site for these devices is also impractical. To
processes. “Mobile Business Process” is a business solve this problem, the device-independent approach
process, when a place of execution of an activity can is used to support different devices without the high
be different in different instances of the business cost.
process or places can change during the execution of Device Independence technologies can be
an activity (Ritz and Stender 2003) (Berger, Bouzid divided into three different categories namely: -
et al. 2003). Mobile Business Processes can be intermediate, client-side and server-side (Butler,
based on these three assumptions: Giannetti et al. 2002).
Intermediate approach can offer limited
• “Uncertainty of Location”.
adaptation to the content delivery. This approach
• “Uncertainty of Location” is externally gives data-enabled phones access to web sites either
determined, and omitting server’s full resolution colour images or
• A corporation with external resources is changing it to the low resolution depending on the
needed in the execution of the process. device display capabilities (Hwang, Kim et al. 2003)
(Butler, Giannetti et al. 2002).
1.2 Context-Aware Mobile In the client-side approach the content adaptation
Applications can occur on the device itself. The advantage for this
approach is that the adaptation code has direct
Context means situational information, or as access to the device capabilities (Butler, Giannetti et
(Barkhuss and Dey 2003) states: “Context is any al. 2002).
information that can be used to characterize the Server-side content adaptation approach offers
situation of an entity. An entity is a person, place, or maximum control over the delivered content
object that is considered relevant to the interaction including the ability to change content, navigation
between a user and an application, including the and style. In this approach the server is assumed to
user and the application themselves.” have sufficient information about the delivery
Context-aware is a concept where the context, including the delivery device capabilities.
applications can discover and take advantage of Using device independence approach the content
contextual information such as user location, time of integrators do not need to develop content for every
day, nearby people, devices, and user activity. single device, where the content can be delivered to
An example of context-aware applications is the different devices without accumulating more
scenario in which a consumer makes use of the resources and effort.
mobile technologies in retail grocery supply chain In the next sub-section we will discuss different
through mobile shopping of electronically deployment methodologies and we will discuss our
referenced grocery products. This kind of approach in mobile business process deployment.
application is also beneficial to retailers, who by
knowing the exact location of the consumer can 1.4 Deployment
allocate the workforce more accordingly and
efficiently, and can avoid out-of-stocks (Hakkila and In terms of technology, the word “deployment”
Mantyiarvi 2005) (Koolwaaij and Strating 2003). means “Installing, setting up, testing, and execution”
(Ali, Torabi et al. 2006). Therefore, deployment can
1.3 Device Independence be interpreted as a general process that must be
customized according to specific requirements or
In today’s world of rapidly changing technologies characteristics.
and the rise in the use of mobile devices has created Deployment of software, applications or a
a need for content adaptation. According to one of process is a complex task which covers all the
the authors (Mikhalenki 2004), the goal of device

210
 A MOBILE BUSINESS PROCESS DEPLOYMENT FRAMEWORK FOR DEVICE INDEPENDENCE AND
CONTEXT-AWARE ENVIRONMENTS

activities from the end of the development itself to Another kind of deployment technique is “static
installation and maintenance of the application on deployment” where the user connects to the site of
the consumer devices. In (Richman 2001) the the application server through its mobile and then
authors have compared different types of subscribes to a download operation. Then the user
deployment techniques in terms of their scale, receives an SMS containing instructions for
complexity, expressiveness, and barriers to first use. downloading and installing the application.
The deployment solutions being handled by these The framework known as Smart Deployment
implementation techniques: manual, script, Infrastructure (SDI) is designed to facilitate the
language, and model-based deployment. installation of large distributed applications for any
The automation of application or service kind of user terminal. This framework is also
deployment improves correctness, speed and presented for the context-aware deployment of
documentation but, as different companies have applications to the mobile users (Taconet, Putrycz et
experienced, it comes at an increased cost in al. 2003). The framework is implemented with
development time and a steeper administrators’ middleware technologies like CORBA and SOAP
learning curve. which facilitates the development of large scale
According to recent research, business processes distributed applications. SDI offers automatic
are deployed through XML web services. Web deployment of multi-component applications and
services can be considered as the emerging provides a deployment solution to customize
distributed middleware technology that uses a installation and to adapt to device capabilities
simple XML-based protocol to allow applications to (Courtney 2000; Taconet, Putrycz et al. 2003).
exchange data across the web (Zhang, Zhang et al. In the next section we will present a framework for
2004; Hammadi, Ali et al. 2006). At the core of the mobile business process deployment. The
Web Service is the Simple Object Access Protocol framework caters for deployment of a process in a
(SOAP) an XML-based communication protocol for device independence and context-aware
interacting with Web Services. The SOAP environment.
specification includes syntax to define messages,
encode or serialize rules for data exchange and
conventions for representing RPCs (Austaller,
Kangsharju et al. 2004) . 2 THE FRAMEWORK
WSDL is used for describing the services
available. It describes where the service is located, In this section we present a novel framework for
what operations are supported and the format of the business process deployment in a mobile
messages to be exchanged based on how the service environment taking into account the context of the
is invoked (Gokhale, Kumar et al. 2002; Austaller, user and specification of the device to be used. The
Kangsharju et al. 2004). On the other hand Business overview of the architecture is shown in Figure 1.
Process Execution Language (BPEL) is the language The framework proposed consists of five
that is used to implement business processes in Web different components:
Services. It defines a notation for specifying • Deployment component
business process behavior based on Web Services • Process specification component
(IBM, Systems et al. 2002). • Device Specification component
Certain technologies and models have been
• Context-Aware component
presented in recent years for the deployment of
• Mobile Interface component
services and applications using mobile technology.
One of the technologies being used recently by the The “deployment component” is the main
companies for the deployment of services is over- component of the framework. This framework
the-air (OTA) deployment. OTA is becoming consists of two layers namely Composition Layer
increasingly important to support. OTA delivery and Deployment Layer. The work for the
enables easy deployment and upgrades to the composition layer is to compose a selected Business
applications, thereby reducing the disrupting effect Process, setting the context of the deployment and
which installation of new applications and upgrades customizing the process according to the device
may have on mobile users (Taconet, Putrycz et al. being used in the “Mobile Environment”. Where the
2003) (Fjellheim 2006). deployment layer activates the customized business
process to the Mobile Environment.

211
ICE-B 2008 - International Conference on e-Business

2.1 The Framework Design

This section will focus on the Object-Oriented
Design for the proposed Mobile Business Process
Deployment Framework.
We will start our design with the "High Level
Component Diagram" as presented in Figure 2.
Our Object Oriented design is generic, flexible,
scalable, reusable, and it provides the flexibility to
coordinate, coordinate, and synchronize between
components and process automation.
This component diagram shows the system
functionality through a high level of abstraction. For
our design there are nine active components.
In this section we will be present a brief
description of each of the components. We will
briefly discuss the functionalities and properties that
each of the components possesses then in the next
Figure 1: Device Independence & Context-aware mobile section we will discuss each component in detail
process deployment framework. with the help of a class diagram. The brief
explanation of each of the components is as follows:
The second component in this framework is Mapping Engine: - This component is to take
“process specification component”. This is a care of mapping between different protocols. All the
traditional Business Process component consisting functionalities like displaying simple map, Geo-
of activities, resources and user interfaces. User coding, Reverse Geo-coding and routing would be
Interfaces are stored as XML documents in the UI done by this component.
Database associated with process specification Process Specification: - This is a traditional
component. Business Process component consisting of activities,
“Device Specification component” is a typical resources and related user interfaces. All the
component. It consists of different hardware specifications would take place in this component.
configurations, screen resolution settings and other The specification for the Business Process along
features for the mobile devices being currently used with its activities, resources and user interfaces are
in the Mobile Environment. stored in process database.
“Context-Aware component” consists of many
parameters but in our Context-Aware component,
we only consider two parameters “Location and
Environment”. The location of the Mobile User is
determined using different technologies like GPS,
GSM, or MPS as shown in location server defined in
our architecture. The “Location Database” is used to
store all the locations of the users and later can be
used in process deployment.
“Mobile Interface component” consists of the
N number of users using N number of mobile
devices. The Mobile devices being used can be
pocket pc’s, smart phone’s or PDA’s etc.
In the rest of this section we will present the
Framework Design and the System Architecture.
The three-tier system architecture best shows the
interaction between the components and
communication between the three tiers.
Figure 2: High level component diagram.

212
 A MOBILE BUSINESS PROCESS DEPLOYMENT FRAMEWORK FOR DEVICE INDEPENDENCE AND
CONTEXT-AWARE ENVIRONMENTS

Device Specification: - It consists of different SDK Interface that links the application with the
hardware configurations, screen resolution settings mapping engine
and other features for the mobile devices being Through this engine our framework would be
currently used in the Mobile Environment. In the able to do the following functionalities:
framework these would be stored as an XML • Display simple map with centerOnAddress or
documents and would be used at the time of CenterOnPositon. Where the map would be
deployment. Device database is used to store all the centered either by latitude longitude position or
device specification for different mobile devices. by address.
Context-Aware: - Consists of many parameters • Show Geocoded address on the Map.
including "Location and Environment" parameters. • Do Reverse Geocoding and show the position
In this component we show how the location and of the points on the map.
environment parameter can be determined using • Display the route between two points on the
different positioning and sensor technologies. map. This also includes displaying the Text
Context database is using to store the context Instructions required in navigating from start to
information of the user in the mobile environment. end position.
Deployment: - This is the major component of • Do the map customization in terms of
the framework. This component is where the actual orientation, color, mapView and size.
deployment would take place. All the components
would be integrated and coordinated before 2.1.2 Process Specification
deploying to the mobile environment. All the
Protocols and Web Services technologies would also This component takes care of the Business Process
reside in this component. Specification along with its activities and resources.
XML Parsing Component: - All the XML A Business Process can have many activities and
parsing of the incoming responses, XML documents each can have different resources.
stored for Process Specification, Context and Device User interfaces are attached to each subsequent
Specification is carried out by this component. activity. Not every activity may have a user
interface. User interfaces are associated to activities
2.1.1 Mapping Engine during business process specification. The class
diagram for Process Specification is shown in
The purpose of the mapping engine is to provide the Figure 4.
framework with the maps if required during
deployment. The maps are really important if one
wants to show the user its location that in our case
one of the parameters of the context information.
The class diagram for the component is shown in
Figure 3.

Figure 4: Process specification.

Figure 3: Mapping Engine.
A Business Process may have number of
Mapping Engine is adopted from a Location activities and each activity has some resources at its
Based Services Company, iGoPlus, this exercise. As disposal. Among others other characteristics of
can be seen from the class diagram above, there is a resources includes whether or not they are

213
ICE-B 2008 - International Conference on e-Business

consumable or sharable resources. User interface software specification, hardware specification and
specification is an important part of this component. colour capability.
In our framework we are assuming that user
interfaces would not be specified but only 2.1.4 Context-Aware
referenced. Its attributes include id and path where
user interfaces are stored. It should be noted that not In this section we present the design for context-
all activities would have user interfaces attached to aware component. There are many parameters as
it. part of context-aware information, including the
The Process Specification is stored in the location and environment. The contextual
database. It can be seen from the diagram that there information for this component would be stored as
is one main super class "Database Management" XML documents in the database.
which has all the subclasses from it, which are used The design for this component is based on the
in the system. The class "Database Management" working of commercial applications like on-board
contains all the attributes and the operations navigation systems, gadgets or sensors for light or
common to all the classes. noise. The context parameters, location and
environment will be determined as shown in Figure
2.1.3 Device Specification 6.
Having discussed the design for Mapping Engine
and Process Specification, in this section we will
present the design for Device Specification. Device
Specification is stored as an XML document in the
database. Device Specification consists of software
and hardware specification. Other components
interact with this component through Device
Interface. The design of the device specification of
the framework is shown in Figure 5.

Figure 6: Context-aware component.

The framework will be determining location

through two positioning technologies that are MPS
and GPS. The position from these two technologies
is in the form of Latitude and Longitude. Whereas
Environment parameters is determined through
sensor interfaces as shown in Figure 6.

2.1.5 Deployment
Having specified Business Process, activities and
resources, the deployment component brings context
information and customized user interfaces together
before deploying it to mobile environment. The
composition layer coordinates and integrates all
three components and the mapping engine
component before the protocol layer can deploy it.
Figure 5: Device specification.
The design for the deployment component is shown
in Figure 7.
As can be seen from the diagram, the framework
interacts with the device specification component
through an interface. One device would have one
specification and the specification would have
attributes such as screen resolution, screen size,

214
 A MOBILE BUSINESS PROCESS DEPLOYMENT FRAMEWORK FOR DEVICE INDEPENDENCE AND
CONTEXT-AWARE ENVIRONMENTS

tier. ASP .Net local Web Server is used to support

XML Web Services and Simple Object Access
Protocol (SOAP). These technologies set up the
communication between the server and the client.
Contextual Information and specifications for
mobile devices are also stored as XML documents
in the Mysql Database.

Figure 7: Deployment component.

This component is the backbone for the

deployment process. The Composition layer
coordinates and integrates all the three components
together towards the deployment process. Examples
of functions being performed by this layer are as
follows:
• getLocation():- Gets the location of the
particular user from the Context-Aware
Figure 8: Three tier architecture.
Component. The response returned would be in
terms of latitude and longitude.
Deployment Engine Component consists of two
• getDeviceSpecification():-Gets the Device
layers Composition Layer and Protocol Layer. The
Specification for a particular device from the
responsibility for the composition layer is to
Device Specification Component.
compose a selected Business Process, setting the
• setDeviceSpecification():- Sets the Device
context of the deployment and customizing the
Specification for a particular mobile device in
business process according to the device being used.
the Device Specification Component.
The protocol layer would set protocols for the
A Business Process can be deployed either by deployment like setting up parameter for SOAP
HTTP connection or by SOAP commands. The commands and setting up HTTP connections.
Framework caters for both of these protocols.
The database layer consists of database and
2.2 System Architecture Mapping Engine. For communication between
Business Logic Tier and Mysql, there is MyODBC
The system architecture for the Mobile Business Interface that takes the request from the server and
Process Deployment Framework is divided into sends it to database.
three main parts client layer, business logic and
database layer. The System Architecture is shown in
Figure 8. 3 CONCLUSIONS
The client layer or tier interfaces between the
system and the user. It consists of a mobile device In this paper we have discussed deployment of a
running on operating system windows CE 5.0. The business process to a mobile device in a chosen
Business Process is deployed to a mobile device in context and in a device independence environment.
form of XML document. The communication A Business Process can be deployed to any mobile
between client tier and business logic tier is done device in any location or environment the user may
through XML web services hosted on ASP. Net be. This framework helps companies and developers
Web Server. to deploy same business processes to different
Apache Cocoon Server is used in the framework mobile devices without any software or user
to determine the software capabilities of a mobile interface configuration. Much research has been
device from the responses received from the client done in the field of context-aware and device

215
ICE-B 2008 - International Conference on e-Business

independence, our contribution to this research is to Hammadi, A. M., S. Ali, et al. (2006). A new modelling
bring these aspects together and design a framework approach to enhance reliability of transactional
that can bring more flexibility to the provider and oriented web services. The 8th International
Conference on Information Integration and Web-based
the consumer.
Application & Services (iiWAS 2006), Yogyakarta
Indonesia.
Hwang, Y., J. Kim, et al. (2003). "Structure Aware Web
REFERENCES Transcoding for Mobile Devices." IEEE Internet
Computing 7(5): 214 - 221.
Adam, O., P. Chikova, et al. (2005). Managing Inter- IBM, B. Systems, et al. (2002) "Business Process
organizational business processes using an Execution Language for Web Services."
architecture for M-Business scenarios. International 2006,http://www-
Conference on Mobile Business (ICMB'05). 128.ibm.com/developerworks/library/specification/ws-
Ali, S., B. Soh, et al. (2005). A New Adaptive Business bpel/, Accessed on: 5/5/2005
Model for E-Commerce. Third International Kaasinen, E. (2003). "User needs for location-aware
Conference on Information Technology and mobile services." Personal and Ubiquitous Computing
Applications (ICITA), Sydney, Australia. 7(1): 70-79.
Ali, S., T. Torabi, et al. (2006). "A Case for Business Kalakota, R. and M. Robinson (2001). Trends Shaping the
Process Deployment for Location Aware Mobile Economy. M-Business: The Race to Mobility,
Applications." International Journal of Computer Mc Graw- Hill.
Science and Network Security, (IJCSNS) 6(8): 118 - Koolwaaij, J. and P. Strating (2003). Service Frameworks
127. for Mobile Context-Aware Applications. Conference
Ali, S., T. Torabi, et al. (2006). Location Aware Business on e-Challenges, Bologna, Italy.
Process Deployment. The 2006 International Liang, T. P. and C. P. Wei (2004). "Introduction to the
Conference on Computational Science and Its special issue: Mobile commerce applications."
Application, ICCSA, Glasgow, UK., (LNCS 3983). International Journal of Electronic Commerce 8: 7-17.
Austaller, G., J. Kangsharju, et al. (2004). Using Web Mikhalenki, P. (2004) "Introduction to Device
Services to Build Context-Aware Applications in Independence."http://www.xml.com/pub/a/2004/09/22
Ubiquitous Computing. Web Engineering 4th. /di.html, Accessed on: 8/4/2006
International Conference, ICWE. Paavilainen, J. (2002). "Mobile Business Strategies:
Barkhuss, L. and A. Dey (2003). Is context-aware Understanding the Technologies and Opportunities "
computing taking control away from the user? Three International Journal of Electronic Commerce,
levels of Interactivity Examined. The fifth annual Addison Wesley Professional 1(1).
conference on Ubiquitous Computing Richman, D. M. (2001). Model based process deployment.
(UBICOMP'2003). The 20th. Conference on Digital Avionics Systems:
Berger, M., M. Bouzid, et al. (2003). "An approach to DASC:IEEE.
agent-based service composition and its application to Ritz, T. and M. Stender (2003). B2B mobile business
mobile business processes." IEEE Transactions on processes: scenarios and technologies. 14th.
Mobile Computing 2: 197 - 206. International Workshop on Database and Expert
Butler, M., F. Giannetti, et al. (2002). "Device Systems Applications (DEXA'03).
Independence and the Web." IEEE Internet Siau, K. and Z. Shen (2003). "Mobile Communications
Computing 6(5): 81 - 86. and Mobile Services." International Journal of Mobile
Courtney, P. E. (2000). "Sterling Commerce's e-Business Communications 1: 3 - 14.
Process Integration." EAI Journal First Impression: Swaminathan, J. M. and S.R.Tayur (2003). "Models for
75. Supply Chains in E-Business." Management Science
Davenport, T. H. (1992). Process Innovation: 49: 1387 - 1406.
Reengineering Work Through Information Taconet, C., E. Putrycz, et al. (2003). Context aware
Technology. deployment for mobile users. 27th. Annual
Fjellheim, T. (2006). Over-the-air deployment of International computer software and applications
applications in multi-platform environments. The conference (COMPSAC).
Australian Software Engineering Conference Zhang, J., L. J. Zhang, et al. (2004). WS-trustworthy: a
(ASWEC). framework for Web services centered trustworthy
Gokhale, A., B. Kumar, et al. (2002). "Reinventing the computing. 2004 IEEE International Conference on
Wheel? COBRA vs. Web Services." World Wide Web Services Computing (SCC 2004), Shanghai, China.
(WWW). Zhang, J. J., N. P. Archer, et al. (2003). "Driving Forces
Hakkila, J. and J. Mantyiarvi (2005). Collaboration in for M-Commerce Success." E-Business Management:
Context-Aware Mobile Phone Applications. The 38th Integration of Web Technologies with Business
Annual Hawaii International Conference on System Models 1(1): 51 - 76.
Sciences (HICSS'05).

216
 GIS-BASED MAP GENERATION USING NEW SURVEY
TECHNIQUES

Balqies Sadoun and Omar Al-Bayari

Surveying and Geomatics Engineering Department, Al-Balqa’ Appplied University, Al-Salt, Jordan
[email protected], [email protected]

Keywords: Geographic information System (GIS), Digital Elevation Model (DEM), Ground Control Point (GCP),
Three Dimensional (3D), Global Positioning System (GPS), Real Time Kinematics (RTK),
Orthophotography.

Abstract: Conservation of historical sites depends on acquiring all data related to their life history and their shape. We
are using modern survey technologies such as GPS, satellite images, photogrammetry and classical
instrumentation to build a Geographic Information System (GIS), digital library and precise base map with
all its embedded benefits at a low cost.
GIS offers digital documentation of the site and its surrounding environment, which is essential for the
conservation and protection process of any heritage. In this work we are presenting GIS, digital base map
and an orthophoto for a newly established museum and its surrounding to be used in GIS modeling.
Global Positioning System (GPS) precise measurements and a high resolution satellite image were used to
produce a Digital Terrain Model (DTM) and an orthophoto for the site in order to create a three
Dimensional model (3D). Finally, GIS and (3D) texture model for the museum were produced to conserve,
protect, monitor and manage the facility. It offered as well, the scheme to create a web site to post all
acquired information to publicize the new museum and the ability to employ any modern location based
service (LBS) technologies (emergency, security, guiding etc.).

1 INTRODUCTION satellite images, digital photogrammetry and GIS,

will offer endless of working possibilities (Bayari,
It is very important to use modern technologies such 2005).
as satellite images, remote sensing, photogrammetry The site we intended to conserve is a museum
and GIS (Grosman, 2000) to conserve cultural and located to the west of Amman, Jordan which was the
historical sites. New survey technologies are very house of a national hero. It is beauty and
valuable in the mapping and the management of the significance is due to its location and the Islamic
archeological data (Kvamme, 1999), (Bewley and arcade architecture of its quarters. It is the first of its
Raczkowski, 2002). kind in the country and need to be digitally
The objective of this work is the development of conserved to allow the usage of modern LBS
a base-map: to display a detailed map of the site and technologies in its operation. Satellite images were
its general components, in order to portray an in- used to extract features with real dimensions and a
depth data of any selected part and to create 3-D base-map was extracted from a topographic map and
model with the possibility of a walk through visit so othorectified Image for the site. Moreover, a three
as to conserve and manage the site efficiently. dimensional texture model for the museum was built
To achieve our objectives we had produced a using field survey and GIS starting from topographic
Digital Elevation Model (DEM) and orthophoto planner. Real measurements were translated to a
using some satellite images and a photogrammetry database in a GIS environment to build the 3D solid
software. Then, the DTM produced is used to create model. Then, sketch up texture mapping software
3D model (Baltsavias et al., 2001). Using digital was used for building texture based on capturing
photogrammetry and computing capabilities we field photos. Using Arc GIS environment we
were able to digitally document and 3D model the provided a real 3D model with true virtual reality
cultural site. The produced digital map by means of model to allow all 3D GIS spatial analysis. We

217
ICE-B 2008 - International Conference on e-Business

designed a website that offers all the needed model of the building was produced in GIS. The
information (Location, base-map, photos, videos of sketch up texture mapping software was used to
the 3D model etc.). GIS has a great Capability to build texture based on capturing field photos, then
integrate and update graphic and non-graphic data converting the 3D building texture to Arc scene so
according to user inputs to produce all kind of maps. as to produce 3D GIS Texture Model. Finally, the
beautiful arcade house with the Umayyad Arabesque
style (Fig.1) was turned into a digital data base
2 METHODOLOGY museum.
AND SOFTWARE USED A Topographic Survey of selected natural and built
up features was done to produce the needed maps
A Geographic Information System for the site was and plans with true features to scale. GPS Real-Time
built to house all the digital data to allow it's Kinematic (RTK) and total station provide identical
updating, posting on the web and the usage of output coordinate data to perform topographic
modern LBS technologies. The digital data allows surveys of the terrain, facilities, and infrastructure.
better analysis, presentations, updating in addition to
the production of 3D GIS texture model and a web
site.

Several Software packages were used in the process

including:
- Auto Desk Land Desk Top 2004: is used to create
DTM, feature layer and data base, in order to
produce a topographic plan for the site and to export Figure 1: Main Building.
AutoCAD layers to shape files to use in Arc Map.
- ENVI 4.3, (the Environment for Visualizing
Images) is used to create a high-resolution color
image; PC sharpening fuses high-resolution
panchromatic imagery with multispectral imagery
and to correct the images from the distortion by
Orthorectification algorithm.
- ArcGIS (version 9.1), is used to Produce Base-Map
from digitizing the ortho and classic map. 3D model
was built in ArcScene depending on the attributes
tables (extrusion the building according to height).
Finally it Generates Triangulation Irregular
Networks (TIN) for the site and compose all the data
(images, 3D features, TIN, Sketch Up file, ENVI)
together.
- Sketch up Pro 5: is used in exporting and importing
3D building from and to ArcGIS. It is a powerful
tool for creating, viewing, and modifying 3D ideas
quickly and easily. Although it had been developed
for the conceptual stages of design, Sketch UP is a
very powerful 3D tool for creating presentations.
- Photoshop: is used to eliminate any appearing
effect on photos (shadow on facades, trees, humans
and cars, etc.).
By providing measurements from the field and
processing them we produced the topographic plane.
Then, we used digital photogrammetry technique to Figure 2: Topographic Plane for the museum site scale 1:
orthorectify Quick Bird image to construct 500.
orthophoto. We used the topographic map and the
orthophoto to produce a base-map. A solid 3D Both systems employ similar (or identical) data
collector devices, feature coding, attribution of

218
 GIS-BASED MAP GENERATION USING NEW SURVEY TECHNIQUES

features, stakeout methods, COGO functions, etc. information that is used as a framework upon which
COGO is a suite of programs aimed at coordinate additional data of a specialized nature may be
geometry problems in Civil and Gemomatics compiled. It is as well, a map from which other
Engineering. maps are prepared by the addition of information. To
Land Desk Top 2004, field observations (1500 produce the base-map we had to do a topographic
points) were loaded to produce the final topographic survey, and then to use a Quick Bird Orthorectified
plane. Processing starts by adding the data, building image to digitize roads network, agricultural
the surface and the TIN. Finally, a topographic plan features, buildings and finally to use a topographic
was created using land Desk Top 2004 as shown in map of Swieleh; a suburb in Amman. The collected
Figure 2. data produced an orthorectified image and
Digital photogrammetry technique is used to topographic plane. The flow chart of map generation
Orthorectify Quick Bird image which was available is presented in Figure 5.
as black and white multispectral, color, or PC-
sharpened product. We chose 8 GPS points that can
be used as GCP’s, and 5 points by Total Station
from topographic plane. In order to orthorectify the
image in GeoTIFF format, and files containing the
RPCs for the Quick Bird Image, Digital Elevation
Model (DEM) was used (Certer, 1988). All the data
had been used as input to the ENVI software to
produce the orthophoto high resolution 0.6m and
multispectral RGB (Fig.3) and 3D surface (Fig.4).

Figure 5: Map generation flowchart.

Map projection is a systematic transformation that

Figure 3: Orthophoto. allows the orderly representation of the earth's
spherical graticule on flat map. Registration is based
on a family of mathematical tools that are used to
modify the spatial arrangement of objects in a
dataset into some other spatial arrangement. The
purpose is to modify these geometric relationships
without substantively changing the contents of the
data itself. The registration process involves
changing one of the views of the surface spatial
relationships to agree with the other, without
concern about any particular geodetic referencing
system.
Figure 4: 3D surface view. Digitizers are the most common device for
extracting spatial information from maps,
photograph or other documents. In the Swieleh
3 BASE MAP GENERATION topographic map we digitized the contour lines and
names of places. In the Orthorectified image we
A base map is the graphic representation at a digitized the Agricultural lands (Fig. 6), Buildings
specified scale of selected fundamental map (Fig.7), drainage pattern and roads network (Fig. 8).

219
ICE-B 2008 - International Conference on e-Business

Figure 6: Agricultural lands.

Figure 9: Final base map layout scale 1:5000.

Figure 7: Layer Buildings.

Figure 10: GIS data for Wasfi Atal memorial.

Figure 8: Main and sub street Layer.

As for the final base map layout (Fig. 9), we

collected all Layers and added north arrow, legend,
scale, etc...
Architectural Heritage Conservation is better
achieved through building 3D photo-texture model
integrated with GIS. The modernization of our world
needs technical document production. Digital
documentation of cultural heritage is not an end in
itself but serves as a tool to make accessible
information and better presentations. Figure 11: Three solid model after extraction in GIS

220
 GIS-BASED MAP GENERATION USING NEW SURVEY TECHNIQUES

In the process of 3D modeling and building the 4 CONCLUSIONS

height database we used the digital maps to
implement the field height data through GIS (Fig. GIS is an effective tool in managing, sorting,
10). analyzing and presenting any architectural or other
Then, we applied the heights to their corresponding attribute data to serve in the field of conservation,
features to produce 3D solid model (Fig. 11) in GIS management and the employment of modern
environment. technologies (LBS for emergency, guided tours,
To build real texture 3D model under GIS planning etc.).
environment (to conserve the real texture of all the - The production of a digital base-map by means of
facades), digital photos for the sides of the buildings satellite images and digital photogrammetry is a
were captured using a high resolution digital camera. new and fast technique, which allows the updating
Then, Photoshop software was used to eliminate any and the production of endless needed maps.
extra features captured in the photos such as trees, - The production of 3-D texture model helps in
cars and people as it is presented in Figure 12. performing advanced analysis and studies.
Processing topographic surveys and high-
resolution satellite images is an excellent
technology to conserve important sites accurately
with real features.
- Finally, the creation of an important web site rich
in information to allow global visitors, researchers
and concerned groups is the optimal benefit of the
work.

REFERENCES
Figure 12: Photo processing.
Grosman, D., 2000. "Two example for using combined
prospecting techniques, in: M. Pasquinucci and F.
Finally, Sketch Up software was used to export and Trement (eds.), Non-destructive Techniques Applied
import to and from ArcGIS (Using the SketchUp in Landscape Archaeology". The Archaeology of
ArcGIS Plugin). The SketchUp ArcGIS Plugin was Mediterannean Landscape 4, Universita di Piusa, 245-
installed in the GIS environment to enable the 255.
transformation of 2D GIS data to sketchUp, Kvamme, K. L., 1999. "Recent directions and
seamlessly and transfer 3D texture model to an development in geographical information system",
ArcGIS geodatabase. Figure 13 presents the real and Journal of Archaeological Research 7 (2): 153-201
Bewley, R., and Raczkowski, W., 2002. Past achievements
the 3D texture model for comparison purposes. and prospects for the future development of aerial
archaeology: an introduction, in: R. Bewley and W.
Raczkowski, (eds.), Aerial Archaeology, Developing
Future Practice, IOS Press, Amsterdam, 1-8.
Baltsavias, E.P, Pateraki, M, Li Zhang, 2001.
"Radiometric and geometric evaluation of Ikonos
GEO images and their use for 3D building modeling.
Proc". Joint ISPRS Workshop High Resolution
Mapping from Space 2001, Hannover, 19-21
September.
Bayari, O. (2005). "New Survey Technologies For
Production OF GIS Model Of The Ancient Roman
Jerash City In Jordan". CIPA International
Symposium, Torino, 2005, Italy.
Certer, J.R.1988, "Digital Representations of Topographic
Surface", Photogrammetric Engi-neering and Remote
sensing 54 (11), 1577-1580.

Figure 13: Actual and 3D texture model.

221
POSTERS
 eCT: THE B2C E-COMMERCE TOOLKIT FOR THE
WEBCOMFORT PLATFORM

Frederico de Carvalho Baptista, João de Sousa Saraiva and Alberto Rodrigues da Silva
INESC-ID/Instituto Superior Técnico, Rua Alves Redol 9, 1000-029, Lisboa, Portugal
[email protected], [email protected], [email protected]

Keywords: e-Commerce, WebComfort, content management systems (CMS), e-commerce extensions for CMS.

Abstract: Electronic commerce (e-Commerce) involves a complex set of business workflows regarding the buying and
selling of products or services by means of electronic systems such as the Internet. e-Commerce applica-
tions can be supported, amongst other alternatives, by content management systems (CMS) through specific
extensions. This paper presents the eCT extension, an e-Commerce extension developed on the top of the Web-
Comfort CMS platform, whose main goals are: (1) adaptability to different contexts, such as different types of
stores, countries, cultures or tax systems; (2) extensibility, due to the ease of development and addition of new
functionalities; and (3) ease of installation and configuration, to fulfill different business requirements. This
paper discusses the major technical details of the eCT, such as: (1) the product facet model; (2) the support to
different e-commerce models; and (3) the support for different payment, shipping and taxing methods.

1 INTRODUCTION and CMS concepts. Section 2 introduces the eCT

(the B2C e-commerce Toolkit for the WebComfort
Electronic commerce (e-Commerce) involves busi- platform). Section 3 overviews of the eCT archi-
ness processes related to the buying and selling of tecture and its extensibility features. Section 4 dis-
products over electronic systems, such as the Internet. cusses some of the architectural decisions taken dur-
Electronic commerce is expected to continue growing ing the development of this toolkit. Finally, Section 5
at a rapid pace, and organizations are using it to reach presents the conclusions for this project so far, as well
a wider audience and create new business opportuni- as the future work.
ties, through ”electronic shops”, that offer products
from a single firm, or from multiple individual stores
(electronic shopping malls (Khosrow-Pour, 2006). 2 THE ECT TOOLKIT
On the other hand, in the last years, a number
of web-oriented content management systems (CMS) The ”e-Commerce Toolkit” (eCT for short) is an e-
(Suh et al., 2003) has appeared, aiming to facilitate commerce extension for the WebComfort platform
the management and publication of digital contents. (WebComfort.org, 2008). WebComfort is an Enter-
These systems allow the collaboration of the various prise CMS, promoted by SIQuant (SIQuant, 2008),
users involved in the process of creating and orga- developed on top of Microsoft ASP.NET technology.
nizing digital content, and typically provide extension In the WebComfort platform, a Module provides
mechanisms that support the addition of new types of mechanisms to manage and layout a certain kind of
content as well as new languages and visual themes. content (e.g., images, text, links), and a Toolkit is an
These systems are providing organizations a way to integrated collection of modules that provide specific
support their needs for constant evolution and com- functionalities. For example, the WebComfort Stan-
plex business requirements. dard Toolkit defines a set of modules allowing the
This paper is structured in five sections. Sec- management and visualization of typical Web con-
tion 1 introduces the broad context of e-Commerce tents (e.g., images, text).

225
ICE-B 2008 - International Conference on e-Business

The eCT is a Toolkit for e-commerce support, more 3.1 Overview

specifically for Business-to-Consumer (B2C) busi-
ness model. From a quick perspective, the eCT allows The eCT aggregates a set of WebComfort modules
the configuration of one or more electronic stores, that implement B2C e-commerce functionalities. The
that allow customers to browse the product catalog, eCT has a multi-tier (three-tier) architecture, with the
add products to the shopping cart and proceed to the typical presentation, logic and data layers, with a We-
check-out. On the other hand, store administrators bComfort module being transversal to this tiers (see
can configure and manage a number of store details Figure 2). The presentation tier of a module con-
(such as product catalog, inventory, promotions and tains the UI elements (Web User Controls and Web
orders) whereas eCT administrators manage aspects Forms) that present information to the user and inter-
that are available to all the stores (such as languages pret/respond to their actions.
and localization, currency units, shipping/payment The logic tier contains the module’s business
and taxing methods). logic. This level can interact with the WebCom-
fort API, that provides a set of infrastructural fea-
tures available to the WebComfort modules, like au-
thentication or multi-language support. In a similar
way, infrastructural features of the eCT are provided
by the eCT API, that cover areas like: eCT config-
uration, store configuration, store operation manage-
ment, catalog, inventory and promotions. The eCT
configuration involves the configuration of global op-
tions shared by all stores; the store configuration in-
volves the configuration of each store options (e.g.,
accepted shipping/payment/taxing methods, currency
units); the store operation management involves the
management of orders, customers and statistics; the
catalog refers to the storing and management of prod-
Figure 1: A WebComfortPortal with the eCT toolkit.
uct data; the inventory corresponds to the information
of the physical storing of products; and promotions
corresponds to the features that allow the promotion
3 ECT EXTENSIBILITY of certain products. At the eCT level, there are a num-
ber of possible extensions, namely: Product facets;
FEATURES/AREAS Shipping; Payment; Taxing; Logging; Statistics and
Reporting; and Currency Units.
The eCT is an extension to the WebComfort platform, The Data tier isolates the persistence aspects of the
aiming to support electronic stores that can be based module. The eCT toolkit uses two data repositories,
anywhere on the world and available to a global audi- for the eCT data and statistics/logging data.
ence. Due to the variability of possible contexts, the Now we present in more detail some of the more
eCT must provide solutions for the different aspects relevant aspects/features of the presented architecture.
that can vary (e.g., languages, payment systems, ship-
ping systems, taxation, currencies, types of products 3.2 Product Facet Model
to sell). So, the main principles that guide the design
of the eCT are:
One main requirement identified was the support for
1. Adaptability to different contexts (such as differ- different types of products (e.g., books, digital files),
ent countries and cultures, types of stores); which may differ in the way they are presented to the
end-user, as well as in internal behavior. The eCT
2. Extensibility, namely by the ease of development supports the notion of a ”base product”, which can be
and integration of new features; specialized or extended by applying facets according
3. Ease of installation and configuration, allowing a to the context. These facets can be perceived as ex-
quick setup of eCT according to stores’ different tensions to a product, that can add a certain behavior
business requirements. and/or data. For instance, in a fine arts portal we can
allow the sale of fine arts originals by defining a facet
This section discusses how these principles were ”Fine arts original”, that allows the association of this
taken into consideration during the eCT design. information (author, materials, dimensions, etc) to the

226
 eCT: THE B2C E-COMMERCE TOOLKIT FOR THE WEBCOMFORT PLATFORM

Figure 2: Overview of the eCT’s architectural features.

product. This model allows the association of various

facets to a product, allowing the composition of facets
to enrich it.
In addition of the physical products (books,
clothes, etc.) supported by these extensions, the eCT
supports also the providing of services, that differ
from physical products in the sense that providing a
service does not result in ownership, but results in
some benefit to the customer. Examples of this can be
found at tourism, entertainment, or cultural services.
Figure 3: e-Commerce Models Comparison.
3.3 e-Commerce Models
definition of new methods and the easy integration of
An important aspect of the eCT is its support for these methods into the eCT. (Due to space restrictions
different business models, so that it can be applied other issues are not discussed here and the reader can
to different domains and business contexts, namely consult (Baptista, 2008))
(1) Store; (2) Shopping Mall; and (3) Shopping
Mall with integrated shopping cart. According the
”Store” model, the WebComfort portal should sup-
4 DISCUSSION
port only one store; according the ”Shopping mall”
model, the portal should support a set of independent The majority of the aspects presented above are re-
stores, that manage their own specificities (catalog, sponsible for allowing the eCT to be a flexible exten-
orders and payments management); and according sion for the WebComfort platform. Even maintaining
the ”Shopping Mall with integrated shopping cart” the WebComfort approach of trying to keep as few de-
model, stores are not totally independent, and should pendencies between components as possible, namely
share the orders and payment management features dependencies between the eCT and the WebComfort
(see Figure 3). platform and between the different components of the
extension.
To support the identified requirements – adaptabil-
3.4 Payment, Shipping and Taxing ity to different contexts; extensibility; and ease of in-
Methods stallation and configuration – a set of inter-related fea-
tures were defined: the product facet model; the sup-
The adaptation to different contexts mainly concerns port to different e-commerce models; and the adapta-
the support of different payment, shipping and tax- tion to different contexts in terms of payment, ship-
ing methods. This is essential, because it is important ping and taxing methods.
to supply methods that fulfill the needs of different To implement the product facet model we adopted the
stores. To achieve this, it is important to allow the Decorator design pattern, that allows to add respon-

227
ICE-B 2008 - International Conference on e-Business

sibilities to individual objects dynamically and trans- This paper presented and discussed some of the main
parently (Gamma et al., 1995). This way, instead of features of the eCT, namely: the product facet model;
trying to support from scratch all the expected prod- the supported B2C e-commerce models; and the
uct features in a complex class, it is now possible to adaptation to different contexts in terms of payment,
define a relatively simple class and add features in- shipping and taxing methods.
crementally with decorators, being the final product For future work we plan to improve this toolkit,
the sum of all these facets. Consequently, this pattern mainly in the areas of promotions and inventory, that
allows the definition of a vast number of extensions do not yet have all the expected features, and to de-
(facets), that materialize specific product semantics. velop more Product Facets and more providers so that
Regarding the behavior that allows the adapta- we can support a greater number of different con-
tion to different contexts in terms of payment, ship- texts.
ping and taxing methods options we opted for the
Provider Model (Rob Howard, 2004), that is a mix
of the Abstract Factory, Strategy and Singleton pat- REFERENCES
terns (Gamma et al., 1995). In the eCT, this allowed
the definition of an API that exposes the operations Baptista, F. d. C. (2008). eCT - Biblioteca de
relative to the payment, shipping and taxing areas (as Comércio Electrónico para a plataforma We-
illustrated in Figure 4). For instance, the Payment API bComfort. Master’s thesis, Instituto Superior
can have a method responsible for processing a pay- Técnico, Portugal.
ment, that contains no business logic; instead it sim-
ply forwards this call to the configured provider. The Gamma, E., Helm, R., Johnson, R., and Vlissides, J.
providers implement different payment methods, con- (1995). Design Patterns: Elements of Reusable
taining whatever business logic they require. Thus, Object-Oriented Software. Addison Wesley.
custom providers for each of these services can be Khosrow-Pour, M. (2006). Encyclopedia of E-
easily and independently developed and configured. Commerce, E-Government, and Mobile Com-
merce. IDEA Group Publishing, USA.
Rob Howard (2004). Provider model design pat-
tern and specification. Retrieved Monday 14th
April, 2008 from http://msdn2.microsoft.
com/en-us/library/ms972319.aspx.
SIQuant (2008). Retrieved Monday 17th March, 2008
from http://www.siquant.pt.
Suh, P., Addey, D., Thiemecke, D., and Ellis, J.
(2003). Content Management Systems (Tools of
the Trade). Glasshaus.
WebComfort.org (2008). Retrieved Monday 17th
March, 2008 from http://www.webcomfort.
org.
Figure 4: Overview of the provider model for payment,
shipping and taxing.

5 CONCLUSIONS

Nowadays organizations and enterprises are using

electronic channels to manage and publish digital
contents as well as sell their products or services.
In the context of our research, the e-Commerce
Toolkit (eCT) was implemented, to support differ-
ent kinds of electronic shops, either offering products
from a single entity or from multiple individual stores
(electronic shopping mall).

228
SURVEY OF CONSUMERS’ DECISION MAKING PROCESS FOR
ONLINE MUSIC SERVICE

Ayako Hiramatsu, Takahiro Yamasaki and Kazuo Nose

Osaka Sangyo University, 3-1-1 Nakagaito, Daito, Osaka 574-8530, Japan
[email protected], [email protected], [email protected]

Keywords: Consumer behavior, Modeling, Decision making, Online music service.

Abstract: This paper describes a survey of consumer behavior for online shopping, especially online music services,
because they are the most popular online shopping service in Japan. Based on Howard's consumer decision
model, questionnaires about decision making for online music services were given to 282 students. The
questionnaire results show that almost 90% of these students have downloaded from online music services,
and high school students use such services slightly more than university students.

1 INTRODUCTION making processes in online shopping. The target of

this analysis was narrowed to online music services,
Internet penetration in Japan is remarkable. In 2006 which are the most popular online shopping service
its penetration rate was 68.5%, and the number of in Japan. Questionnaires about decision making for
Internet users was estimated at 87.54 million (White online music services were done with Japanese
paper, 2007). With the Internet’s diffusion, the consumers.
popularity of e-shopping continues to increase.
63.7% of all Internet users have been using e-
shopping for more than two years (Internet 2 DECISION MAKING PROCESS
Association Japan, 2005, 2006), and these numbers OF CONSUMERS
are still growing. For marketing, understanding the
behavior of e-shopping users is crucial. The Howard-Sheth buying behavior model presented
To understand consumer behavior, user models in 1969 is one important model that explains
are often constructed by analyzing the behavior data consumer decision making. It explains the
of users. Concerning Internet users, there have been complexity of the consumer decision making
some researches about online game models (Hsu and process for incomplete information with an S-O-R
Lu, 2004) and e-shopping users (Shih, 2004) based paradigm that consists of three components:
on the Technology Acceptance Model (TAM) (Liu, stimulus-organism-response. In this model,
2007). Farag et al. also researched a model that consumers are motivated by perceiving a stimulus,
analyzed relationships between e-shops and real and then they shift to purchase after composing
shops (Farag et al, 2007). Moreover, Watanabe and concepts to learn purchases. Furthermore, it is
Iwasaki analyzed why consumers buy PCs through hypothesized that the problem is simplified as
the Internet (Watanabe and Iwasaki, 2007). However, learning continues. As a result of experimental
decision making processes based on various factors researches based on this model, Howard proposed
and the relationship between them have not been the simple consumer decision model shown in
analyzed much, due to such difficulties as direct Figure 1.
observation and analysis of subjective condition of This simple consumer decision model is
users. comprised of six interrelated components, as shown
Our research aims to understand how consumers in Figure 1: information (F), brand recognition (B),
make e-shopping decisions. Consumer behaviors for attitude toward the brand (A), confidence in judging
real shops (Evans et al. 2006) were researched. the brand (C), intention (I), and purchase (P). The
Based on Howard's famous consumer decision three central components (B, A, and C) comprise the
model (Howard, 1989), we examined decision

229
ICE-B 2008 - International Conference on e-Business

buyer's brand image and the ABCs of consumer and the rate of downloading songs is 14.0%, which
behavior. is second. Based on this present condition, we chose
“Information” (F) is what is received and stored. online music services as our research subject for
The perception is stored, not the stimulus. “Brand online shopping through Internet.
recognition” (B) reflects the extent to which the
consumer knows enough about the brand to 3.2 Survey Items
distinguish it from others. “Confidence” (C) is the
consumer's degree of certainty that his/her To adapt Howard's model to online music services,
evaluative judgment of a brand, whether favorable we considered the following survey items. Channels
or unfavorable, is correct. “Attitude” (A), the second for acquiring information about online music
part of brand image, is the extent to which services are considered communication among
consumers expect the brand to satisfy their particular friends and families, information from such media as
needs. “Intention” (I) is a mental state that reflects television, radio, and magazines, and such Internet
the consumer's plan to buy a specified number of sources as e-mails and web pages. “Brand
units of a particular brand in a specified time period. recognition” corresponds to musicians and online
“Purchase” (P) shows whether the consumer really music service providers. For “Confidence”, we
does buy the brand. The influence from considered confidence in artists and providers.
“information” (F) to “intention” (I) is connected by a “Attitude” items include sound quality, price,
dotted line. This arrow is added in routine problem payment methods, data size, and outtakes or live
solving cases when consumers repeatedly buy a low- versions. “Intention” included consumer's income,
price or a daily product. popularity, season, and advertising campaign.
“Purchase” shows whether consumers really buy the
music. This experiment examined the degree of
C experience with online music services.
Confidence

F B I P 4 QUESTIONNAIRE SURVEY
Information Brand Intention Purchase
recognition
4.1 Survey Condition
A
Attitude The questionnaires asked participants to indicate
their degree of agreement with the above items
Figure 1: Howard’s consumer decision model modified for based on a five-point scale: 5-strongly agree, 4-agree,
routine problem solving. 3-neutral, 2-disagree, and 1-strongly disagree. The
question details are shown in the appendix. 282
consumers (164 university students, 106 high school
3 ANALYSIS TARGET students, and 12 others) answered the questionnaires.

Table 1: Results of download medium.

3.1 Online Music Service
High school University
All
With annual growth of about 2% since 2003, the size students students
of the visual contents market was estimated in 2005 Mobile phone 44.0% 64.2% 29.3%
at approximately 11.3 trillion yen (about $ 0.1 PC 22.0% 9.4% 29.9%
trillion), broken down as follows: visual contents/5.3 Both 23.8% 19.8% 27.4%
trillion yen, audio contents/1.0 trillion, and text
Others 10.3% 6.6% 13.4%
contents/5.0 trillion (White paper, 2007). Moreover,
the music or game contents market has especially Table 1 shows the ratios of download media.
expanded. The usage rate of musical contents is Many respondents use mobile phones. From Table 1,
10.9%, which is the highest among such various more than 80% of high school students and over
contents as visual contents, audio contents, and text 50% of university students use mobile phones,
contents, of contents by Internet with PCs. Even including respondents who use both PCs and mobile
using contents through mobile Internet, the usage phones. Table 2 shows the numbers of download
rate of musical ring tones is the highest with 15.6%

230
 SURVEY OF CONSUMERS’ DECISION MAKING PROCESS FOR ONLINE MUSIC SERVICE

experiences from online music services. Almost opportunity of income does not equal the
90% of students have downloaded music. opportunity of downloading.

Table 2: Number of download experiences. Table 3: Questionnaire results.

Strongly Strongly
High school University Agree Neutral Disagree
All agree disagree
students students
F101 17.4% 34.0% 27.0% 9.9% 11.7%
51～ 37.9% 38.7% 37.8% F102 7.1% 28.0% 35.8% 12.4% 16.7%
11 ~ 50 25.2% 28.3% 23.8% F201 8.2% 24.1% 31.2% 17.4% 19.1%
2～10 23.8% 25.5% 23.2%
F202 2.5% 15.6% 36.2% 19.1% 26.6%
1 2.5% 0.9% 1.2％
F301 4.6% 5.3% 24.1% 18.1% 47.9%
0 10.6% 6.6% 14.0%
F302 2.8% 13.5% 27.3% 20.9% 35.5%
F303 6.4% 14.2% 24.5% 17.0% 37.9%
4.2 Questionnaire Results
F401 14.9% 35.1% 24.1% 13.5% 12.4%
The questionnaire results are shown in Table 3. The F402 24.5% 25.2% 20.9% 14.2% 15.2%
symbols on the left row correspond to the number of F403 24.8% 29.8% 23.4% 8.2% 13.8%
the questions shown in the appendix. F404 17.4% 32.6% 26.6% 12.4% 11.0%
The results of Information questions reveal that F501 8.5% 27.7% 33.3% 13.1% 17.4%
many students don’t think information by e-mail is F502 15.2% 20.9% 34.0% 14.2% 15.6%
very important. Furthermore, information by radio is F601 9.2% 14.5% 27.0% 17.4% 31.9%
not important because students do not usually listen F602 11.7% 17.4% 26.2% 14.2% 30.5%
to radio. Such information from Internet as web logs F603 11.3% 13.1% 29.1% 14.2% 32.3%
and web pages is relatively important. Direct word- F701 17.0% 29.4% 22.7% 13.5% 17.4%
of-mouth is more important than word-of-mouth by F702 11.0% 27.3% 31.9% 16.0% 13.8%
Internet. The most important information comes F703 18.1% 22.7% 26.6% 13.1% 19.5%
from TV; clearly, students are greatly influenced by
B101 6.0% 16.0% 24.5% 21.3% 32.3%
TV.
B102 27.7% 29.4% 23.4% 9.2% 10.3%
As the results of “Brand recognition” questions,
C101 4.3% 13.1% 42.6% 21.3% 18.8%
students pay much more attention to the musicians
than delivery providers. Since consumers cannot C102 13.5% 22.3% 41.1% 13.8% 9.2%
directly examine the products of online shopping, it A101 48.6% 23.0% 14.2% 5.7% 8.5%
is often asserted that the reliability of the online A102 44.3% 17.7% 14.9% 8.9% 14.2%
stores is very important. However, with online A201 58.5% 22.0% 11.7% 4.3% 3.5%
music services, consumers can listen to trial songs as A202 21.6% 19.1% 27.0% 17.4% 14.9%
products, providing them a change to examine A203 12.1% 13.1% 18.4% 23.8% 32.6%
products almost directly. Therefore, consumers care A301 56.7% 19.5% 13.5% 4.3% 6.0%
about songs as products, but they have little concern A302 16.0% 12.4% 30.9% 16.0% 24.8%
about the reliability of providers. A303 19.9% 21.3% 27.0% 11.3% 20.6%
The “Confidence” results for indirect purchases A401 41.5% 19.1% 20.6% 8.5% 10.3%
suggest that consumers do not have confidence that I101 47.2% 25.5% 14.5% 7.1% 5.7%
they will be satisfied with products and providers.
I102 39.4% 23.0% 25.5% 5.3% 6.7%
Therefore, many respondents answered “Neutral.”
I201 7.1% 6.4% 18.8% 23.4% 44.3%
Regarding questions about “Attitude,”
consumers are naturally interested in price. Trial I202 5.3% 5.3% 21.3% 19.5% 48.6%
services, sound quality, and data size are clearly I301 19.9% 18.8% 24.1% 14.2% 23.0%
important. Opinions are divided about accounting I302 31.2% 25.5% 19.1% 9.9% 14.2%
systems. I303 27.7% 22.7% 22.3% 10.3% 17.0%
For questions about “Intention,” consumers want I304 17.0% 18.1% 28.7% 14.9% 21.3%
to repeatedly download and buy songs. Many I401 14.5% 14.9% 24.5% 18.4% 27.7%
consumers also want to download new releases or I402 31.9% 19.9% 20.6% 9.6% 18.1%
hit songs. Because the price of one song is cheap or
because they can be billed for their purchases, the

231
ICE-B 2008 - International Conference on e-Business

4.3 Result Comparisons by Attributes opinions about radio. Regarding questions about
motivation in “Intention,” high school students gave
The questionnaire results were analyzed and many positive answers about downloading new
compared based on respondent attributes. releases or hit songs.
(a)Download medium Table 5: Comparison between high school and university
We compared the results by three types of students.
download medium attributes: mobile phones, PCs,
and both. Significance of the compared results was Positive Negative
found in “Information,” as shown in Table 4. Here, High
University
High
University
“positive” means the rate of respondents who school school
answered “strongly agree” or “agree,” and F402 60.4% 41.5% 16.0% 38.4%
“negative” means the rate of respondents who F403 70.8% 43.3% 10.4% 29.3%
answered “disagree” or “strongly disagree.” F501 50.0% 28.7% 15.1% 40.9%
Consumers who use both mobile phones and PCs F502 50.0% 28.7% 16.0% 38.4%
tend to consider word-of-mouth information from F601 30.2% 19.5% 33.0% 60.4%
friends important. PC users tend to consider such F602 36.8% 25.0% 34.0% 52.4%
word-of-mouth information from Internet as web F603 37.7% 15.9% 32.1% 56.1%
logs important. However, although information by
I301 55.7% 25.6% 25.5% 47.0%
Internet is identical to web logs, such differences
I302 73.6% 45.1% 11.3% 32.9%
between mediums do not consist in information from
web pages provided by corporations etc. Mobile I303 65.1% 38.4% 16.0% 36.0%
phone users regard information from magazines and I304 55.7% 21.3% 19.8% 49.4%
TV as important. Furthermore, no users obtained
significant information from radio. Particularly, PC Table 6: Comparison by experience.
users regarded information from radio as Positive Negative
unimportant.
Heavy Middle Light Heavy Middle Light
F101 58.9% 57.7% 39.4% 17.8% 15.5% 29.8%
Table 4: Comparison by download medium.
F201 37.4% 36.6% 24.0% 29.0% 33.8% 46.2%
Positive Negative F202 21.5% 16.9% 15.4% 39.3% 43.7% 53.8%
Mobile Mobile A101 64.5% 81.7% 72.1% 19.6% 5.6% 14.4%
PC Both PC Both
phone phone A102 57.9% 67.6% 62.5% 29.0% 18.3% 20.2%
F101 53.2% 40.3% 61.2% 21.0% 27.4% 14.9% A401 69.2% 59.2% 52.9% 14.0% 18.3% 24.0%
F201 25.0% 43.5% 38.8% 37.9% 33.9% 31.3% I301 48.6% 40.8% 26.9% 30.8% 33.8% 46.2%
F402 53.2% 46.8% 50.7% 25.8% 32.3% 28.4%
F403 62.9% 43.5% 53.7% 15.3% 32.3% 16.4% (c) Experiences
The compared results by experiences are shown
F501 46.8% 21.0% 31.3% 22.6% 50.0% 19.4%
in Table 6. “Heavy” shows the rate of users who
F502 41.1% 25.8% 34.3% 22.6% 40.3% 29.9% have downloaded more than 51 times. In the same
F601 26.6% 14.5% 26.9% 41.1% 62.9% 49.3% way, “Middle” is between 11 and 50 times and
F602 32.3% 16.1% 37.3% 38.7% 58.1% 41.8% “Light” is less than 10 times. Experienced users
F603 29.0% 16.1% 26.9% 38.7% 59.7% 46.3% believe that information from web logs, BBS, and
F701 48.4% 40.3% 53.7% 25.8% 32.3% 28.4% word-of-mouth from friends are important. “Trial
F702 33.9% 41.9% 50.7% 30.6% 17.7% 26.9% service” is regarded as important by all category
users. But middle experienced users only consider it
(b) Generation relatively important. The more experienced users
Table 5 shows the compared results with high also care more about data size. For questions about
school and university students. For “Information” motivation in “Intention,” only seasonable songs are
questions, high school students are clearly more important for heavy users, though the other
influenced by TV. Many university students gave questions show no differences by experiences.
negative opinions about magazines. On questions
about radio, although answers from high school
students equally included both positive and negative
opinions, most university students had negative

232
 SURVEY OF CONSUMERS’ DECISION MAKING PROCESS FOR ONLINE MUSIC SERVICE

5 CONCLUSIONS F102: Do you think word-of-mouth information

from your family is important?
This paper described a survey of consumer decision - Word-of mouth by Internet
making process for online shopping, especially F201: Do you think the information from web
online music services, the most popular service in logs or BBS without registration systems is
Japan. Based on Howard's consumer decision model, important?
questionnaires about decision making for online F202: Do you think the information from BBS
music services were done with Japanese 282 with registration systems is important?
students. Differences between generations download - E-mail
mediums, and experiences were analyzed. As future F301: Do you think direct mail is important?
work, we need more examinations with improving F302: Do you think the information from e-mails
questionnaires etc. for members only is important?
F303: Do you think the information from fan
clubs is important?
REFERENCES - TV
F401: Do you think the information from TV
White paper, (2007). Information and Communications in commercials is important?
Japan F402: Do you want to get TV programs theme
http://www.johotsusintokei.soumu.go.jp/whitepaper/ja songs?
/h19/index.html F403: Do you think the information in musical
Internet Association Japan, (2006). Internet white paper TV programs is important?
2006, Impress. F404: Do you want to get movie theme songs?
Internet Association Japan, (2005). Internet white paper - Magazines
2005, Impress.
Chin-Lung Hsu and His-Peng Lu, (2004). “Why do people
F501: Do you think the advertisements in
play on-line games? An extended TAM with social magazines are important?
influences and flow experience,” Information & F502: Do you think articles about musicians in
Management, 41, pp.853-868. magazines are important?
Hung-Pin Shih, (2004). ”An empirical study on predicting - Radio
user acceptance of e-shopping on the Web,” F601: Do you think the information from radio
Information & Management, 41, pp.351-368. commercials is important?
Chuanlan Liu, (2007). Modeling Consumer Adoption of F602: Do you think the information from radio
The Internet as a Shopping Medium, Cambria Press.
music programs is important?
Sendy Farag, Tim Schwanen, Martin Dijst, and Jan Faber
(2007). “Shopping online and/or in-store? A structural F603: Do you think information about musicians
equation model of the relationships between e- from special radio programs is important?
shopping and in-store shopping,” Transportation - Web pages
Research Part A, 41, pp.125-141. F701: Do you think the information of hit charts
Kazuo Watanabe and Kunihiko Iwasaki, (2007). “Factors on web pages is important?
Affecting Consumer Decisions about Purchases at F702: Do you think the information from video
Online Shops and Stores,” Proceedings of IEEE delivery web sites is important?
CEC/EEE 2007, pp80-87.
F703: Do you think the information in musician
Martin Evans, Ahmad Jamal, and Gordon Foxall, (2006)
Consumer Behavior, John Wiley & Sons. web sites is important?
John A. Howard, (1989). Buyer Behavior in Marketing “Brand recognition”
Strategy, Prentice Hall. B101: Do you try to learn a lot about music
delivery providers?
B102: Do you try to learn a lot about the
APPENDIX musicians whose songs you are downloading?
“Confidence”
[Questions] C101: Do you have confidence in the reliability of
“Information” delivery providers before purchasing?
- Direct word-of-mouth C102: Do you have confidence that you will be
F101: Do you think word-of-mouth information satisfied with a particular musician’s songs before
from your friends is important? downloading?

233
ICE-B 2008 - International Conference on e-Business

“Attitude”
- Trial
A101: Do you think a trial service in delivery sites
is important?
A102: Even if you know the song, do you listen to
a trial of it?
- Sound quality
A201: Do you pay attention to sound quality?
A202: Do you think such alternative versions as
live recordings and outtakes are important?
A203: Are you satisfied with the sound quality by
which you can judge the song?
- Accounting system
A301: Do you think price is important?
A302: Do you think a monthly fee is a better
accounting system?
A303: Do you prefer payment per song as an
accounting system?
- Data size
A401: Do you consider the data size of songs?
“Intention”
- Service
I101: Do you think the possibility of re-download
is important?
I102: Do you think the payment method is
important?
- Money
I201: When you have extra money, do you want
to download a song soon?
I202: On payday, do you want to download a
song soon?
- Motivation
I301: Do you want to download seasonable
songs?
I302: Do you want to download new releases?
I303: Do you want to download hit songs?
I304: Do you want to download award-winning
songs?
- Campaign
I401: If you can get one free song by purchasing
of ten as part of a special campaign, do you want
to download?
I402: If you can get unlimited songs much for 300
yen (about $2.7) for a limited time, do you want
to download?

234
 TYPICAL PROBLEMS WITH DEVELOPING MOBILE
APPLICATIONS FOR HEALTH CARE
Some Lessons Learned from Developing User-centered Mobile Applications
in a Hospital Environment

Andreas Holzinger, Martin Höller, Marcus Bloice

Institute of Medical Informatics, Statistics and Documentation (IMI), Research Unit HCI4MED
Graz University Hospital, Auenbruggerplatz 2/V, Graz, Austria
[email protected], [email protected], [email protected]

Berndt Urlesberger
Department of Neonatalogy, Graz University Hospital, LKH-Universitätsklinikum, Austria
[email protected]

Keywords: Mobile applications, touch screens, human-computer interaction, information systems.

Abstract: This paper provides an overview of the experiences gained during the design, development and
implementation of mobile applications for use within the clinical domain. Current problems and issues that
arose during the development of the software are documented and discussed. Medical professionals'
opinions, both medical doctors and nurses, and their input were coupled with front end development (user
interface design) and back end development (software engineering) to decide on the most optimum
development path and to select the most appropriate environments. Most of all, this project can be seen as a
further example that User-Centered Development (UCD) is necessary, however it is not sufficient when
developing mobile, cross-platform, and future-proof applications for medicine and health care.

1 INTRODUCTION extremely hectic (Brekka, 1995), (Holzinger &

Errath, 2007). Mobile computer user satisfaction is
During the past few years, improvements in the certainly an issue and this has not been researched
technology of touch screens, further miniaturization, extensively (Ozok et al., 2008). Obviously, all
lower power consumption, and longer battery life aspects of Human–Computer Interaction (HCI) and
has made it possible to design and produce better Usability Engineering (UE) are of growing
mobile computers, tablet PCs, and small mobile relevance, and must be especially considered when
devices such as PDAs or smart phones. The market in the process of developing software for medical or
for such mobile computing devices is rapidly health care purposes (Holzinger, 2007). This has
expanding whilst at the same time the technological resulted in a set of commonly accepted development
performance of these devices is steadily increasing, practices, such as User-Centered Development
(Antinisca Di & Cecilia, 2007). In this paper we (UCD) (Holzinger, Searle & Nischelwitzer, 2007),
concentrate on discussing the lessons learned during (Holzinger, Sammer & Hofmann-Wellenhof, 2006).
the development of specific user-centered software Although such considerations are important, we
for tablet PCs and define the tablet PC as our mobile must accept that for an application to work well, all
computing device (Prey & Weaver, 2007). However, aspects must be taken into account, and there is the
although mobile computers have been available for a potential danger that HCI professionals and usability
relatively long time in hospitals (Forman & engineers tend to discount the underlying aspects of
Zahorjan, 1994), different studies show that health software engineering (SE) (Thimbleby, 2007).
care professionals are reluctant to use poorly
designed mobile systems, as the patient care
workload is very time constrained and can be

235
ICE-B 2008 - International Conference on e-Business

2 MOBILITY IN HEALTHCARE persons within group) and, furthermore, the situation

is conducive to a communicative environment. An
There are several areas of healthcare where mobile optimum solution to this problem would be to use
computers are necessary. On top of this, each area mobile devices, which would aid communication
has its own unique requirements. In order to and ease data visualization.
differentiate between them, we form two distinct PDMSs are often located within intensive care
sections: a) Round Assistance, which consists of units; units with standard care patients do not have
help, or assistance, that a doctor might require when PDMSs (this is due to the fact that in this case
doing routine “rounds” of the wards to check the monitoring data does not have to be stored
patients’ progress, etc.; and b) On-call Assistance, continuously). In such circumstances mobile
which consists of help or assistance that a doctor applications could be used as information servers
may need while on call-duty. These are not routine during a round (displaying laboratory data, etc.).
check-ups on patients; rather they are performed as Here mobile applications are essential.
and when deemed necessary by medical In both the above scenarios the mobile device
professionals who are on call-duty. has to have a screen size and display resolution
Two main clinical situations exist which suitable for graphical data presentation. Therefore
determine whether or not a Patient Data only tablet PCs or laptops are appropriate. Due to
Management System (PDMS) is used in the unit, hospital budget shortages, duties-on-call have
Again, this depends on a number of factors. Wards become more common. In such a situation a senior
that use such PDMSs face specific problems (Junger doctor is available by phone. Data presentation using
et al., 2001). mobile applications eases decision finding and is
However, any pros of automatic data storage are beneficial to the senior doctor. In such cases, mobile
also faced with cons: a) overloading the medical devices such as PDAs or smart phones may also be
professional with data b) communication difficulties appropriate. However, this is true only in very
that occur with patient details. In both of the above specific circumstances.
cases, these problems will ensure that the future
worth of mobile applications is secured.
All PDMSs have basic operation centers, which 3 METHODS AND MATERIALS
are generally PCs that are within a distance of two to
three meters from the patient. There, the majority of Within our project the prototype of a mobile system
the medical operations are performed (data for visualizing a patient’s overall status during ward
observation, decision finding, drug prescription). rounds was developed for the intensive care unit of
Due to the huge amount of data available to the the department of neonatology at Graz University
medical professional using the PDMS, it is difficult Hospital, which is amongst the largest in Europe.
to gain a quick overview of the patient’s situation. An automatic patient monitoring system stores a
Details can be found within the PDMS, but huge amount of various data at fixed intervals of 15
overviews are difficult to get. Switching between minutes for each patient. The measured data consists
different patient details on the same computer/screen of vital signs, administered medications, expulsions,
requires a lot of time and is actually not especially and so on.
helpful. The problem for the medical professionals was
One solution to this problem may be a set of two twofold: 1) The system’s user interface for viewing
to three screens, on which data is displayed. Another and analyzing the data is, however, very cluttered
solution would, of course, be the use of a mobile and containing extreme large amount of data in a
device, which communicates with the PDMS (via unstructured way. Although a graphical timeline plot
WLAN, for example). is provided, doctors often have to analyse the raw
A problem which is often encountered when numeric data. The average time spent on the analysis
doing rounds on the medical wards is that any of one patient is between 5 and 15 minutes,
discussion of a patient’s medical needs is often depending on the doctor’s experience. 2) The system
centered on the PC display belonging to the patient. is non mobile, requiring the medical professionals to
Most often, one person sits while others stand proceed to the stationary PC’s.
around in a circle in order to view the display. This The aim of our project was not only to provide
situation presents several problems, not least the mobility but also to significantly reduce the time
difficulty in seeing smaller details on the screen spent on information perception, so the raw data
(which does, of course, depend on the number of only has to be analyzed if the patient is not in a good

236
 TYPICAL PROBLEMS WITH DEVELOPING MOBILE APPLICATIONS FOR HEALTH CARE - Some Lessons
Learned from Developing User-centered Mobile Applications in a Hospital Environment

condition. This is accomplished by providing instant If, for example, value A is bad, the overall alarm
visual feedback about the patient’s status using a state of the module becomes bad too. In the
combination of star plots and traffic light metaphor. visualization the alarm state is represented using the
traffic light metaphor, i.e. the background of the star
3.1 Device plot is filled with green, yellow or red .

The device used for the prototype was an LE1600 3.2.2 Database Interface
tablet PC by Motion Computing, which supports
stylus and finger input. Extensive experiments about The database in question was an Oracle 8i database,
the differences between finger versus stylus input for which there are JDBC drivers available. Java
have already been undertaken (Holzinger et al., Database Connectivity is an API (Application
2008b). Table 1 contains the technical specifications Programming Interface) which allows database-
of this device. independent connectivity between the Java
programming language and a wide range of
Table 1: Technical specifications of the LE1600. databases, including Oracle 8i (Oracle, 2008).
Oracle have available a JDBC compliant driver
CPU Intel Pentium M at 1.6 GHz for this database which allowed for relatively quick
Memory 1 GB development of the interface between the Java
Display size 12.1” XGA LCD
program and the patient data.
Display dimensions 247 mm x186 mm
Of course, speed was also an issue – without
Display resolution 1024 px × 768 px
good response times the software would not be
Hard disk size 60 GB
Weight 1.4 kg
useful as a way of accessing data quickly and easily.
Physical 296 mm x 240 mm x 18.7 mm However, in this regard JDBC and Oracle perform
dimensions extremely well. Database and driver support,
therefore, was not a technological issue when
3.2 Software developing the application; however it was a factor
which played a role in deciding which programming
The software was written using Java 1.5 and the language to chose. It is also worth pointing out that
Swing user interface toolkit, where we had quite were the JDBC drivers not available, this would
positive experiences from former projects have led to an extremely long development cycle
(Holzinger et al., 2008a). Basically, services should and may not have been possible at all.
be adapted at runtime to the features of the device. This is an absolutely crucial aspect that must be
Also, end users should at any time specify that considered when developing mobile applications
services are delivered to match certain parameters. that access an external database: your programming
For example, end users may request that an image be language/platform/database combination must have:
printed while specifying a particular resolution, a) the ability to perform the required task and,
format and/or number of colours (Stefano, Claudia less obviously,
& Luigi, 2007). b) should have available quality drivers and
libraries to ease development.
3.2.1 Visualization Front-end
The front end is based on so called “visualization
modules”. Each module is created and configured by 4 LESSONS LEARNED
the medical doctor. It reflects a subset of the
available data, containing interrelated values. The 4.1 General
doctor chooses which values are contained within
the module. The module “Circulation” for example As already mentioned in the introduction, mobile
would contain heart rate, blood pressure, oxygen devices have increased in capability by many factors
saturation, etc. For each of these values the medical over the past few years, in terms of both features,
doctor sets three intervals for defining the alarm such as integrated cameras, and raw processing
states of good, mediocre and bad. Each module has power. Coupled with the fact that mobile devices
an overall alarm state, which can too be good, now contain many of the attributes that constitutes a
mediocre or bad. This overall alarm state is defined PC, several software development platforms have
by the individual values’ alarm states. become available by various vendors, most notably

237
ICE-B 2008 - International Conference on e-Business

Google, Microsoft, and Sun Microsystems, among slightly more abstract yet equally capable
others. However, their usefulness in the area of alternatives (such as Adobe’s Air, and Microsoft’s
medicine must be discussed. Silverlight).
Interestingly, the proliferation of mobile phones,
and jointly, the popularity of games for mobile 4.3 Java SDK
phones have cemented Java’s position as the
environment of choice for developing mobile Creating game applications using the Java SDK
applications for small devices. seems to be the de facto way of producing
Having said this, developing applications in the applications for small devices, but here we must
area of medicine requires that a platform inhibits far judge its worth as a platform for more serious
different attributes than those platforms used to application development.
develop mobile games. In this section, we discuss A number of aspects of the Java SDK were
the problems, and currently available solutions for analysed. For example, Java has been known to
developing real-world, cross-platform solutions for render fonts very poorly on screen, and has been a
mobile devices. topic of discussion for quite some time, with various
workarounds and techniques available to cure the
4.2 Currently Available Cross problem. Another severe restriction is the missing
Platform Development support for floating point numbers in MIDP (Mobile
Environments Information Device Profile) versions prior to 2.0,
though there exist third party workarounds for this
In the classical development world, if you wish to problem. So long as the mobile device is MIDP 2
develop a piece of software you must usually first compliant, this is no longer an issue, however this is
decide on the device on which you will develop your something which must be ascertained before
application. Palm, for example, offer an SDK for development begins and could potentially be an
their devices which eases the development of issue in the future lifecycle of your software.
applications. Therefore, software developed for
Palm devices can only be run on Palm OS. 4.4 Alternatives to Java - AJAX
Limiting yourself to one single platform
seriously diminishes your potential market when AJAX is a term used to describe a number of
selling your software, or, in the case of the medical currently available technologies that when combined
profession, may demand redevelopment (if, for form a framework with which you can build
example, a device or platform is no longer available desktop-like applications for the web (Turner &
when inventory is recycled). Cross-platform Wang, 2007).
development, in many respects, eases this as you can The often touted examples are Google’s AJAX
develop software that runs on any device where the applications such as Gmail, Calendar or Maps – they
runtime environment is available (Bishop & allow drag & drop, ‘refreshless’ updating of
Horspool, 2006). information, and offer a desktop like and feel.
Since Java’s philosophy of “write once, run Of course, being AJAX applications, they can
anywhere” was incepted, several vendors have run on any supported browser such as Firefox,
delivered cross platform languages. Safari, or Internet Explorer, eliminating the need for
Microsoft has developed .NET, and its Micro the developer to worry about which operating
Edition competes with Java’s Mobile Edition in the system the user is running, as long as the operating
mobile domain. A new player in this area is Google, system itself supports the browser. This has the
who, along with 30 other technology companies, is added advantage that most of the high-end
currently touting the Android platform (Android, technology can reside on a server rather than on the
2008). If a doctor or medical professional were to user’s device. An AJAX application can therefore
carry a device with them at all times, it would access an Oracle database, without the programmer
suffice to say that the device must just be portable. having to worry about Oracle drivers being available
Tablet PCs have the advantage of having high for their framework/operating system combination
system resources, large screens, and most run the One of AJAX’s advantages stems from the very
ubiquitous Windows operating system. fact that it was conceived as an internet platform – it
On Windows, there are any number of cross- is geared towards users implementing a point and
platform development environments that one could click device rather than a keyboard. This bodes well
choose from, varying from the obvious, to the for small devices, as input is generally carried out

238
 TYPICAL PROBLEMS WITH DEVELOPING MOBILE APPLICATIONS FOR HEALTH CARE - Some Lessons
Learned from Developing User-centered Mobile Applications in a Hospital Environment

using a stylus rather than a keyboard. Using Google One more thing to consider, however, is that Java
Maps on a traditional PC, for example, it is possible is not available for Windows Mobile or Palm
to find your street and house without having to input (discontinued since the 12th of January, 2008). It is
a single character into the keyboard (assuming some therefore the author’s opinion that AJAX could still
knowledge of geography). Creating applications, be considered the most optimal solution in creating
therefore, using AJAX geared towards mobile cross platform applications. However, it is unlikely
devices that utilize styli should demand no extra that PDAs have the required resolution and screen
effort of the part of the programmer; in fact the size required to view patient information effectively.
inverse is true – it should be more instinctive to Of course, that is not to say AJAX cannot run on
generate applications that do not require keyboard tablet PCs, this is certainly what AJAX was
input. designed to do. Consider also the quickly changing
medical field – AJAX applications when run, by
4.5 Java Limitations definition, are always up to date (this, however, is
also true of Java’s Webstart). Therefore, ruling out
Java’s Runtime Environment (i.e. the Java Virtual AJAX altogether would be foolish, it certainly has
Machine) is not platform independent. It is simply a its niche, but perhaps not in the medical domain, or
runtime which is available on a (diminishing) rather not in our specific area of patient care in the
number of platforms. Its source code, however, is medical field.
platform independent, but this is also true of C/C++
code, often referenced as being platform specific. 4.6 NET
Consider the following example in Java:
for (int i = 1; i <= 10; i++) Microsoft’s .NET framework is as platform
{ independent as Java, in the sense that the runtime
System.out.println("Number: " + i); can be ported to any platform. Currently Microsoft
} only supplies a runtime for the Windows line of
And the following code written in C++. operating systems, but because Microsoft submits
the specification for the Common Language
for(int i = 1; i <= 10; i++) Infrastructure to both ECMA and ISO, it is an open
{ standard. Therefore, it is possible for third parties to
cout << "Number: " << i << endl; create implementations of the framework on other
}
platforms. This is currently the goal of the Mono
Both these source codes are transferable between project, which aims to port the framework to Linux.
platforms. The C++ code requires that a compiler is However, for our requirements in the medical
available for the platform that you want to run your domain, we required a far more concrete
code on, but the same is true of Java – it requires implementation, and Java officially releases several
that a Java VM is available on the machine you wish versions of its framework, something that ruled out
to run your code on. .NET at this time for our purposes.
The main difference is that the compiled output Again, inter-platform operability was an absolute
is not transferable, but the code is. Of course, requirement for us, as potentially many different
programs are far more complicated than this, and machines and platforms would be using the software
even primitive types, such as integers, vary in size across the university hospital, and because we have
from C++ compiler to C++ compiler. And Java’s no control over what devices the hospital purchases
compiled output is transferable from one platform to for its medical professionals.
another, without even the need to recompile,
something impossible to achieve with a C++ 4.7 Chosen Platform
compiled application. However, this still begs the
question: is Java useful as a platform in the medical The decision was made to opt for Java as the
domain where usage on mobile or small devices is a language of choice. The medical domain demands
must? The authors thinks so. While Java’s Virtual unique considerations that eventually ruled out most
Machine is available only for Windows, Mac OS platforms that are currently available. Several years
and Linux\UNIX, these are only operating systems from now, .NET may be a contender, and AJAX was
available for tablet PCs as of the time of writing. It certainly a consideration. However, AJAX lacks the
also seems very unlikely that another operating maturity and robustness required for the purposes of
system will appear in the foreseeable future. this project, and Java’s large library meant

239
ICE-B 2008 - International Conference on e-Business

development could be performed as rapidly as Holzinger, A. (Ed.) (2007) HCI and Usability for
possible. It can be seen that the field of medicine Medicine and Health Care: Third Symposium of the
demands far more considerations when developing a Workgroup HCI&UE of the Austrian Computer
piece of software. Everything must be considered, Society, USAB 2007, LNCS 4799, Berlin, Heidelberg,
New York, Springer.
from screen resolution, platform availability, speed, Holzinger, A., Emberger, W., Wassertheurer, S. & Neal,
portability, library capability, and supported L. (2008a) Design, Development and Evaluation of
technologies. Java’ maturity, concrete standards Online Interactive Simulation Software for Learning
(such as the JDBC API) and wide ranging third party Human Genetics. Elektrotechnik &
support makes it the choice for medical software Informationstechnik (e&i), 125, 5, 190-196.
development. Holzinger, A., Errath, M. (2007) Mobile computer Web-
application design in medicine: some research based
guidelines. Univ Access in the Information Society
International Journal, 6, 1, 31-41.
5 CONCLUSIONS Holzinger, A., Höller, M., Schedlbauer, M., Urlesberger,
B. (2008b). An Investigation of Finger versus Stylus
It is clear that special considerations must be made Input in Medical Scenarios. ITI 2008: 30th
when developing applications in the medical International Conference on Information Technology
domain, especially if these applications should be Interfaces, June, 23-26, 2008, Cavtat, Dubrovnik,
IEEE, in print.
platform independent, future proof, and mobile.
Holzinger, A., Sammer, P., Hofmann-Wellenhof, R.
There are a plethora of frameworks, environments, (2006) Mobile Computing in Medicine: Designing
and programming languages available, each with Mobile Questionnaires for Elderly and Partially
their own specific advantages and disadvantages but Sighted People. Springer LNCS 4061. Berlin, New
only some are suitable for the medical domain. By York, Springer, 732-739.
reading this paper, it should be possible to save Holzinger, A., Searle, G., Nischelwitzer, A. (2007) On
anyone a lot of research and work if you they are some Aspects of Improving Mobile Applications for
considering writing a cross platform, portable the Elderly. In: Stephanidis, C. (Ed.) Coping with
application in the medical domain. Almost all Diversity in Universal Access, Research and
Development Methods in Universal Access, LNCS
considerations were taken into account, from screen
4554. Berlin, Heidelberg, New York, Springer, 923-
resolution, doctors’ wishes, language suitability, and 932.
operating system capability. By working close to Junger, A., Michel, A., Benson, M., Quinzio, L. A., Hafer,
medical professionals, UI experts, and software J., Hartmann, B., Brandenstein, P., Marquardt, K.,
engineers, it was possible to ascertain what special Hempelmann, G. (2001) Evaluation of the suitability
considerations must be taken into account when of a patient data management system for ICUs on a
working in this field. By analysing these general ward. International Journal of Medical
considerations, a number of concrete factors could Informatics, 64, 1, 57-66.
be defined which eventually led to development path Oracle (2008), Oracle 8i. Online available:
http://java.sun.com/javase/technologies/database last
and programming environment that was chosen.
access: 2008-06-10
Ozok, A. A., Benson, D., Chakraborty, J., Norcio, A. F.
(2008) A comparative study between tablet and laptop
REFERENCES PCs: User satisfaction and preferences. International
Journal of Human-Computer Interaction, 24, 3, 329-
Android (2008), Open Handset Alliance, Online available: 352.
www.openhandsetalliance.com/android_overview.htm Prey, J., Weaver, A. (2007) Tablet PC technology: The
l, last access: 2008-06-10 next generation. Computer, 40, 9, 32-33.
Antinisca Di, M., Cecilia, M. (2007) Performance analysis Stefano, C., Claudia, R., Luigi, U. (2007) A Java mobile-
and prediction of physically mobile systems. 6th enabled environment to access adaptive services. 5th
international workshop on Software and performance. international symposium on Principles and practice of
Buenos Aires, ACM, 129-132. programming in Java. Lisboa, Portugal, ACM, 249-
Bishop, J., Horspool, N. (2006) Cross-platform 254.
development: Software that lasts. Computer, 39, 10, Thimbleby, H. (2007) User-Centered Methods Are
26-35. Insufficient for Safety Critical Systems. In: Holzinger,
Brekka, T. (1995) Select mobile computers tailored to A. (Ed.) USAB 200, LNCS 4799. Berlin, Heidelberg,
healthcare environment. Health Management New York, Springer, 1-20.
Technology, 16, 13, 48, 50. Turner, A., Wang, C. (2007) AJAX: Selecting the
Forman, G., Zahorjan, J. (1994) The Challenges of Mobile framework that fits - The right tools make the
Computing. IEEE Computer, 27, 4, 38-47. difference. Dr Dobbs Journal, 32, 6, 40-42.

240
 A STUDY OF FACTOR AFFECTING CUSTOMER SWITCHING
BEHAVIOR OF MOBILE TELECOMMUNICATION 3.5G
SERVICES

Jungwoo Lee, Eok Baek

Yonsei University, Republic of Korea
[email protected], [email protected]

Sujung Sung
Autoeversystems, Republic of Korea
[email protected]

Keywords: Switching Behavior, Attractive of Alternatives, Network Externality, Switching Cost, Moderated Effect,
3.5G Mobile Telecommunication.

Abstract: Recently, Mobile Telecommunication businesses contend with each other to expand their customer base by
using an aggressive marketing strategy. In order to determine if this strategy is effective, customer's and
their switching behavior needs to be studied. This study identifies and analyzes direct, indirect factors
affecting customer switching behavior such as attractiveness of alternatives, network externality, switching
cost. The research results show that attractiveness of alternatives, network externalities have a direct impact
on customer switching behavior. These factors also have a moderated effect on customer switching behavior
but the switching cost does not influence customers directly.

1 INTRODUCTION that affect switching behavior based on the two

aforementioned theories.
As the service industry becomes diverse and
enhanced, competition increases among providers
and brand preference is reduced (Fornell, 1992), 2 LITERATURE REVIEW AND
prompting companies to practice defensive HYPOTHESES
marketing in an attempt to maintain existing
customers and attract new ones. In the mobile
communication service industry, the cost for signing
2.1 Switching Behavior
up a new customer is ten times greater than
Carpenter & Lehmann(1985) defined switching
maintaining an existing client, thus customer
behavior as 'customer shift to other companies'
retention has become an essential element of
products' and Reichheld & Sasser(1990) as
survival in the industry (Bolton, 1998). Accordingly,
'customer defection'. Switching behavior can be
there is a need to examine customer behavior
explained as a concept that counters customer
associated with switching to the competition. There
retention and intent to repurchase, and customer
are three main objectives of this study.
retention refers to the overall notion that
The first is to look into the dynamics of the
encompasses the behavior of actual usage as well as
mobile communication market based on previous
the attitude toward the intent of continued usage
studies conducted on the switching barrier.
(Oliver, 1999; Allen & Rao, 2000). Studies of
The second is to explain how network externality
customer retention in the service industry have
is expressed in 3.5G services and exchanges
focused on customer satisfaction and switching
influences with customer switching behavior.
behavior as major factors that affect how customers
Finally, this study attempts to conduct empirical
can be retained. According to past studies, although
analyses base on the correlations among the factors
the intensity varies in the relationship between

241
ICE-B 2008 - International Conference on e-Business

customer satisfaction and customer retention, the alternative lacks attractiveness, consumers will
former is generally regarded as a contingent factor continue to use current services albeit not
for the latter (Anderson and Sullivan, 1993; Jones, completely satisfactory (Porter, 1980).
1998; Oliver et al., 1993; Kim, Sang-Hyeon, 2002).
H 2: Switching barrier has effects in the
This implies that high customer satisfaction has a
relationship between customer satisfaction toward
negative effect on switching behavior, an opposing
the current service and switching to 3.5G services.
concept of customer retention and intent to
H 2-1: Switching cost has a positive(+) effect
repurchase. Accordingly, it can be concluded that
between customer satisfaction and switching
the higher the degree of satisfaction about the
behavior.
current mobile communication service, the lower the
H 2-2: Alternative attractiveness has a negative(-)
probability of switching to 3.5G services. This study
effect between customer satisfaction and switching
focuses on the switching behavior prompted by 3.5G
behaviour
services, and the definition of switching behavior
shall include both churning to other service H 3: Switching barrier has effects on switching
providers and upgrading to 3.5G services within the behavior to 3.5G services.
same provider. H 3-1: Switching cost has a negative(-) effect on
switching behavior to 3.5G services.
H 1: Customer satisfaction toward current mobile
H 3-1a; Continuity cost has a negative(-) effect on
communication service has a negative(-) effect on
switching behavior to 3.5G services.
switching behavior to 3.5G services.
H 3-1b: Contractual cost has a negative(-) effect on
switching behavior to 3.5G services.
2.2 Switching Barrier H 3-1c: Initial cost has a negative(-) effect on
switching behavior to 3.5G services.
Jones(2000) defined switching barrier as the
H 3-2: Alternative attractiveness has a positive(+)
psychological and financial difficulties that a
effect on switching behavior to 3.5G services.
customer faces in the process of changing service
providers. Switching barrier consists of the
switching cost and alternative attractiveness, and
2.3 Network Externalities
these factors play moderating roles between
First studied by Katz and Shapiro(1985), they
customer satisfaction and switching behavior
defined network externalities as the concept that "the
(Anderson et al., 1994; Jones et al., 2002).
value consumer gains from a product or service is
Jones(1998) categorized the switching cost into
proportional to the size of the network that the
continuity cost, contractual cost, searching cost,
product or service belongs to". In other words,
learning cost, initial cost and dissociation cost.
network externalities come into consideration when
However, since mobile communication providers
the value of a product or service depends on the
offer similar services and usage of 3.5G services is
number of its users.
not significantly different from conventional mobile
Consumers generally wish to be linked to large-
phone, the searching and learning costs can be
scale networks because they expect products or
disregarded. Furthermore, since consumers are not
services with large user bases provide higher quality
involved in face-to-face interactions with their
and value. Accordingly, network externalities are
service providers, the dissociation cost of severing
referred to as consumption externalities of quantity,
personal relationships does not have to be
which does not provide much value on its own but
considered. Accordingly, this study only deals with
signifies network characteristics of a product or
continuity, contractual and initial costs.
service whose value increases with expansion of the
Alternative attractiveness is defined as
user base (Shapiro, 1999).
consumers' awareness of service providers that can
The value of 3.5G services would be marginal
replace current providers (Jones et al, 1998) and
without a significant user base of video telephony,
related to the concept of service augmentation in
which is the core of 3.5G services. Once the number
related literatures (Porter, 1980). Service
of video telephony users increases, the value of 3.5G
augmentation refers to providing additional services
services will also increase due to the bandwagon
to be distinguishable from other service providers. In
effect. Thus the following hypotheses were
turn, when there is an alternative that provides better
services, consumers will disconnect the relationship established.
with current providers and opt for the better option
(Bradford, 2000). On the other hand, if the

242
A STUDY OF FACTOR AFFECTING CUSTOMER SWITCHING BEHAVIOR OF MOBILE TELECOMMUNICATION
3.5G SERVICES

Table 1: Manipulative definition of variable.

Questionnaire
Category Research Variables Manipulative Definition Items Evaluated
Structure
Customer recognition and •Overall level of satisfaction
Customer Overall Customer
sentiment after service regarding fundamental services and I-1, 2
Satisfaction Satisfaction
experience provider
• Whether 3.5G services are offered
Degree of awareness •Awareness of characteristics and
Alternative
regarding providers that quality of 3.5G services(High-speed II-1, 2, 3, 4
Attractiveness
offer alternative services data service, video telephony, USIM
card, roaming)
•Psychological/financial cost
associated with notification of new
Cost associated with trouble
Continuity phone number
in switching service II-5, 6, 7, 8
Cost • Cost associated with lost contacts
providers
Switching • Reluctance to use video telephony
Barrier due to compromised anonymity
Cost of losing financial •Cost of losing current service
Switching Contractual
benefits by staying with the provider's membership services II-9, 10
Cost Cost
current service provider • Cost of losing discount benefits
• Sign-up fee for the new service
provider
Additional costs
• Cumbersome sign-up procedure
Initial Cost associated with switching to II-11, 12, 13
• Cost of signing up for new value-
a new service provider
added services and loss of current
contents
•Change in behavior according to
Customer awareness of the network externalities
Network Externalities III-1, 2, 3
user base for new services •Degree of awareness regarding the
number of users for new services
•Intent of continued use of services
from the current provider
Customer's switching
•Possibility of switching to new
Switching Behavior behavior from current IV-1, 2, 3, 4
service
service
•Conviction of switching to new
service

H 4: Network externalities have a negative(-) effect questionnaires retrieved, those with incomplete or
between customer satisfaction toward current insincere responses were discarded and the
service and switching behavior to 3.5G services. remaining 158 were used as the survey data for the
study. Likert 7-point scale was used for every
H 5: Network externalities have a positive(+) effect
question except for the items regarding
in switching behavior to 3.5G services.
demographics, and SPSS 12.0 was used to perform
statistics analysis, reliability analysis, factor analysis
and regression analysis.
3 RESEARCH METHODS
In order to verify the hypotheses of this study, 4 DATA ANALYSIS
manipulative definition was established for each
variable based on the previous studies described in
Chapter 2 to elicit evaluation items for this study. 4.1 Characteristics of Sampe Space
Online surveys were conducted from November 12
to 16, 2007 for male and female mobile telephone The statistics related to the 158 respondents for the
users between the ages 10 and 60. Among 180 survey. The male-female ratio was somewhat

243
ICE-B 2008 - International Conference on e-Business

balanced at 53.2% and 46.8%, but 83.8% of the Table 2: Result of Analyses.
respondents were in the 20~40 age group. Hypotheses Result
Hypotheis 1 Adoption
4.2 Analyses of Variable Scale Hypotheis 2 Partialy adoption
Reliability and Propriety Hypotheis 2-1 Discard
Hypotheis 2-2 Adoption
The internal consistency method was used for
Hypotheis 3 Partialy adoption
verifying reliability of this study. The survey
questions were used without modification because Hypotheis 3-1 Discard
there was no indication of reliability being Hypotheis 3-2 Adoption
undermined, and the Cronbach's Alpha coefficient Hypotheis 4 Adoption
for the factor variables used in the study were Hypotheis 5 Adoption
between 0.6 and 0.9, displaying a high level of
internal consistency. 4.4 Additional Analysis: Effect of
In addition, factor analysis was performed for Intensifying the Switching Cost
each item that constitutes a variable to analyze
propriety and find out whether each variable was
Research Model
isolated and assessing a unique property.
Additional analysis was performed for this study to
VARIMAX rotation based on major element
assess how switching costs customer satisfaction and
analysis was used to elicit factors, and factor loading
switching behavior based on the fact that the initial
was set at 0.4 for those with Eigen value of 1.0 or
switching cost in particular has influence on the two
higher, which yielded 5 factors. The analytical result
factors. The respondents were divided into two
indicated that the two variables that explain
groups based on whether they felt the initial
switching cost - contractual cost and initial cost -
switching cost was high or low to analyze the effects
should be grouped as a single factor. It was
between degree of satisfaction and switching
determined that the respondents regarded both
behavior. The result indicated that there is
contractual and initial costs as being associated with
significant influence between level of satisfaction
switching to or signing up for a new service
and switching behavior for the group that felt the
provider, and the two variables were redefined as the
initial switching cost was high, but the influence
initial switching cost.
insignificant for the other group.
4.3 Verification and Analysis of
Hypotheses
Verification for the 5 research hypotheses were
conducted in two stages. First, multiple regression
analysis was used to verify the factors that directly
affect switching behavior. H1, H3 and H5 are
classified as direct factors. The second stage
involved analyzing the moderating effect that
influence the relationship between current mobile
service satisfaction and switching behavior, for
which H2 and H4 correspond with. Hierarchical
multiple regression analysis was used to analyze the
moderating effect. In addition, the average value of
each factor was calculated for regression analysis. Figure 1: Research Model for Study Results.
Adoption criteria of hypothesis were set at 95%
reliability and a significance level of 0.05 or lower. The result corresponds with the basic research
Summarizing the analytical result, 5 of the 7 model demonstrating that customers who feel that
hypotheses established for the study were adopted there is a high cost of switching mobile service
and the remaining 2 were discarded. providers recognize the cost as a switching barrier
and will not display switching behavior. Therefore,
it can be learned from the additional analysis of this

244
A STUDY OF FACTOR AFFECTING CUSTOMER SWITCHING BEHAVIOR OF MOBILE TELECOMMUNICATION
3.5G SERVICES

section that the initial switching cost reinforces the the number of 3.5G service subscribers and changes
research model of this study. in the actual profit structure.

5 CONCLUSIONS REFERENCES
This study analyzed a comprehensive model of Sang Hyeon Kim, Sang Hyun Oh, 2002, The Determinants
switching behavior to examine consumer awareness of Repurchase Intentions in the Service Industry :
of the new 3.5G mobile communication services and Customer Value , Customer Satisfaction , Switching
Costs , and Attractiveness of Alternatives, Korean
the factors that affect the switching behavior to 3.5G
Marketing Asscociation, 17(2), pp. 25-55
services. The study was conducted in two stages. Allen, D. R., Rao, T. R., 2000, Analysis of Customer
First, the study assessed the factors that directly Satisfaction Data: A comprehensive guide multivariate
affect the level of satisfaction as well as network statistical analysis in customer satisfaction, loyalty and
externalities, perceived switching cost and service quality research, Milwaukee, Wisconsin:
alternative attractiveness of 3.5G services. Second, ASQC Quality Press
the interaction effects of network externalities, Anderson, E. W., M. W. Sullivan, 1993, The Antecedents
alternative attractiveness and switching cost were and Consequences of Customer Satisfaction for Firms,
examined as moderating variables between Marketing Science, 12, pp. 125-143
satisfaction and switching behavior. Anderson, E. W., Fornell, C., Lehmann, D.R., 1994,
Customer Satisfaction, market share, and profitability:
The expectation of 3.5G services providing various Finding from Sweden, Journal of Marketing, 58, pp.
distinguishing features from conventional mobile 53-66
communication to offer new sources of satisfaction Bolton, R. N., 1998, A Dynamic Model of the Duration of
has been reflected in this study. In the case of video the Customer's Relationship with a Continuous
telephony, the flagship 3.5G service, network Service Provider: The Role of Satisfaction, Marketing
externalities significantly affect consumers' Science, 17, pp.45-65
switching behavior in terms of the perception of how Bradford, Neeru Sharma, Patterson G., 2000, Switching
many others were using the service. costs, alternativeness and experience as moderators of
relationship commitment in professional, consumer
As for the switching cost, number portability and
services, International Journal of Service Industry
sign-up fee waiver offered by the service providers Management, 11
allow consumers to alleviate the burden of notifying Carpenter, Gregory S., Donald R. Lehmann, 1985, A
their new numbers to their contacts or signing up for Model of Marketing Mix, Brand Switching and
new services. Competition, Journal of Marketing Research,
Whereas previous studies on the subject have Vol.22(August), p.318-329
focused solely on the switching barrier, this study Fornell, C., 1992, A national customer satisfaction
bears significance in that network externalities were barometer: the Swedish experience, Journal of
analyzed from the consumer behavior perspective. Marketing, Vol. 55.
Jones, M. A., 1998, Satisfaction and Repurchase
Moreover, the study provides a foundation for
Intentions in the Service Industry: The Moderating
establishing a practical strategy for securing a 3.5G Influence of Switching Barriers, Unpublished
service customer base to gain a competitive edge. dissertation, University of Alabama
However, this study failed to cover factors such as Jones, M. A., Mothersbaugh, D. L., Beatty, S. E., 2002,
consumer usage characteristics and more than 80% Why customer stay: measuring the underlying
of the survey respondents were in the age group dimensions of services switching costs and managing
between 20 and 40. It is therefore necessary to their differential strategic outcomes, Journal of
diversify the age groups of the sample space to Business Research, 55, pp. 441-450
increase the accuracy of the study. Questions have Katz, M., Shapiro, C., 1985, Network Externalities,
Competition and Compatibility, American Economic
been raised regarding the effectiveness of the
Review, 75(3), pp. 424-440
strategies implemented by the mobile Oliver , R. L., 1993, Cognitive, Affective, and Attribute
communication providers to maintain and acquire Bases of the Satisfaction Response, Journal of
customers for 3.5G services, but this study did not Consumer Research, 20(December), pp. 418-430
examine whether customers display switching Oliver, R. L., 1999, Whence Consumer Loyalty?, Journal
behavior according to the strategies. In turn, it would of Marketing, 63(special issue), pp. 33-44
be necessary in future studies to assess whether Porter, M. R., 1980, Competitive Strategy: Techniques for
service providers' strategies are effective based on Analyzing Industries and Competitor, New York, The
the mechanisms explained in this study as well as Free Press

245
ICE-B 2008 - International Conference on e-Business

Reichheld, F. F., Sasser, W. E., 1990, Zero Defections:

Quality comes to services, Harvard Business Review,
68(September-October), p.105-111
Shapiro, C., Varian, H. R., Information Rules: A Strategic
Guide to the Network Economy, Harvard Business
School Press, Boston, MA

246
 ADOPTION OF NEAR FIELD COMMUNICATION
TECHNOLOGY IN BUSINESS TO CONSUMER SERVICES

Arto Wallin, Juha Häikiö and Jaana Määttä

VTT Technical Research Center of Finland, My, Kaitoväylä 1, Oulu, Finland
Arto. [email protected], [email protected], [email protected]

Keywords: Adoption of mobile services, Near Field Communication, NFC, NFC-based mobile services.

Abstract: Touch based interaction has been found out to provide fast and easy way to use mobile services. Near Field
Communication technology (NFC) is one of the most promising ways to conduct touch based interaction
with mobile phones. In this paper our aim is to identify challenges of the adoption of NFC related business
to consumer services. The results of this study are based on three pilot cases conducted in Finland in 2007.
Our research presents findings based on qualitative company interviews and both quantitative and
qualitative user questionnaires. We conclude that there are several challenges in the adoption of NFC-based
applications. However, expect for the shortage of NFC enabled mobile phones, challenges of the adoption
may be reduced with careful planning and implementation of NFC-based applications.

1 INTRODUCTION Field Communication (NFC). One of the main

advantage of the NFC technology is that it can be
Although the diffusion of mobile phones has been used to communicate with other mobile devices or
outstanding during last years, use of mobile services tags without a configuration (Calvet, 2005).
has not gained comparable success. One reason Therefore, the communication with the NFC enabled
behind the slow diffusion of mobile services might mobile phone should be as simple as touching
be that the usability of mobile services has not been objects.
at a sufficient level. This is partly caused by the In addition to the advantages for consumers in
poor user interfaces of today’s mobile phones, which form of usability, use of touch based interaction
can be experienced when, for example, trying to technology may provide business benefits for
type text or navigate in menus using mobile phone’s service providers. In the case studies reported in this
keypad. Difficult user interfaces may even totally paper, NFC technology was used by the service
prevent some users, such as elderly people, from providers with an aim to improve the effectiveness
using mobile services. of their service processes. Hence, besides increasing
Difficult user interaction of traditional mobile customer satisfaction, the service providers expected
services has created a need to develop easier and that the NFC technology might also provide indirect
more intuitive user interfaces. A promising solution cost savings to them.
for this problem is based on a physical selection Despite the clear benefits of a touch based user
method, in which users interact with the digital interaction, the adoption of NFC has not been as fast
environment by touching Radio Frequency and wide as some have expected (Balaban, 2007).
Identification (RFID) tags with the mobile devices After the fade of initial hype, there still exist some
(Välkkynen, Niemelä & Tuomisto, 2006). challenges that have to be overcome before the wide
According to the previous study by Riekki et al., scale adoption of NFC is realized. This paper aims
services used by touching were experienced as ease to identify and analyse these challenges using both
to use and learn. Touching was also appreciated, quantitative and qualitative data obtained from three
because “it gives clear feeling of control” (Riekki, pilot cases conducted in the city of Oulu during the
Salminen & Alakärppä, 2006). autumn 2007. Data contains both the service
Touch based user interaction can be conducted providers’ and the consumers’ perceptions about the
with a mobile phone by using new short range challenges of adoption.
wireless communication technology called Near

247
ICE-B 2008 - International Conference on e-Business

The paper is organised as follows. Section two User acceptance of services is a crucial condition
introduces a theoretical background and earlier for the success of mobile services. Some existing
studies related to the adoption of NFC technology acceptance models emerge issues, which can be used
and mobile services. In the section three, we present as an evaluation framework for mobile services.
our research methodology and pilot cases. We will Kaasinen (2005) has developed Technology
then present the results of our empirical studies in Acceptance Model for Mobile Services. This model
the section four, which is followed by the discussion has its roots in Technology Acceptance Model
and conclusion. created by Davis (1989). In Kaasinen’s model user
acceptance of mobile services is consisted of three
factors, which affect the users’ intention to use a
2 RELATED RESEARCH mobile service. These factors are perceived value of
service, perceived ease of use and trust. According
Theory and experiences about the adoption and to Kaasinen (2005), the fourth factor, ease of
diffusion of innovations has been studied for years. adoption becomes more central, as a user proceeds
Even today many papers are based on diffusion from intention to the actual use. These fore-
innovation model introduced in 1962 and later mentioned factors of Technology Acceptance Model
refined by Rogers (1995). Innovation adoption for Mobile Services can be seen also as a foundation
theories focus on identifying and describing factors for examination of NFC-based mobile services.
that have some effect to the adoption. Kargin and
Basoglu (2006), for example, introduce framework
for adoption factors of mobile services. They state 3 RESEARCH SETTING
that usefulness of the mobile service and past
experience influence on the consumer’s attitude The purpose of our research was to examine
towards new technology and thus to the adoption of suitability of the NFC technology to selected
technology. business to customer services and to find out general
Although there is plenty of research about the barriers to adoption of NFC technology. Since this
adoption of mobile services (e.g. Nysveen et al. topic is quite new, our research approach is
2005; Bouwman et al., 2007; Carlsson et al., 2005), explanatory and it presents finding of both
the research about the adoption of NFC-based qualitative and quantitative research conducted in
mobile services is almost non-existent. Main reason three individual pilot cases. Table 1 summarises the
for this is the novelty of NFC technology. Despite of characteristics of these pilots including the user
the novelty, touching as an interaction technique in amount, the duration, the user segment and the goal
NFC-based applications has been studied in several of the pilot. More detailed information about the
studies. These studies consider touching with a pilots is presented in the following subsections.
mobile phone as a very promising interaction The objective of the qualitative research was to
technique (Rukzio et al., 2006; Riekki et al., 2006; find out detailed information about business actors’
Välkkynen et al., 2006; Anokwa et al., 2007). expectations, experiences and challenges
Some studies have already evaluated usability of considering the adoption of NFC in their service
the NFC technology in form of case studies. Jaring processes. Interviewed actors included upper level
et al. (2007) survey the results of six small NFC managers from two private companies and two
pilots, which are focused on improving mobile public organisations. At least one interview was
solution workflows and usability. They conclude conducted in each of these four organisations. In
that NFC based solutions are easy to use, but the addition, restaurant pilot included three interviews
small and limited keypad poses challenges to the with operational level employees. All interviews
design of NFC applications. Häikiö et al. (2007) were conducted after the finish of the pilots in the
have studied the use of the NFC technology in beginning of 2008. Interviews followed semi-
elderly care environment. The study provides structured format and lasted from one to two hours
evidence that the adoption of the touch based each. The interviews were recorded and transcribed
interaction does not require a specific knowledge or for the further analysis.
experience with technology. In addition, elderly
people, who had impaired motor skills, were able to
use a touch-based user interface in the pilot test.

248
 ADOPTION OF NEAR FIELD COMMUNICATION TECHNOLOGY IN BUSINESS TO CONSUMER SERVICES

Table 1: Summary of the analysed pilots.

Pilot Pilot name Responses / Duration User segment Goal of the pilot
# Users
1 SmartParking 48/51 + 2 months Frequent Improve effectiveness of parking in
5/5 traffic parkers/traffic city area. Provide fast, easy and
wardens wardens efficient parking solution for
customers.
2 Fast track of 23/27 5 weeks Restaurant’s Provide fast ordering and payment for
restaurant Pannu regular customers. busy lunch time customers.
3 Oulu city theatre 101/141 2 months Employees from Improve customer experience and
(8 plays) several different provide fast and easy ticket checking
companies. and refreshment reservation.

Companies selected to the pilots were expected wardens tested a new NFC based parking control
to benefit from the use of NFC technology in their solution, which was envisaged to improve the
business processes. In order to find out a clearer effectiveness of the parking control. Although the
vision for the NFC technology opportunities, the NFC based parking control was highly appreciated
first section of the interview concentrated on by the traffic wardens, it will not be analysed in
organisation’s background information, a position at detail, since the focus of this paper is on the business
the market and a current technology usage. The to customer services.
second part surveyed the effects of NFC technology, A main pilot group consisted of 51 customers
which were detected during the pilots. The aim was who parked frequently at the city area. Customers
to find out what benefits and challenges are directed were able to use SmartParking for on-street parking
to the business actors’ service processes. The last and also in one parking hall located in the city
part considered pros and cons of the NFC centre. On-street parking transaction was initiated by
technology in a more general level. touching an NFC tag attached to car’s windscreen. A
The objective of the quantitative research was to user had to then select a parking zone either by
find a general overview about the drivers and touching one of the parking zone tags attached to
barriers for the customer adoption of NFC in fare collection machines and lamp posts, or
different services. The data was collected with web alternatively, selecting a zone code manually from
and paper based user questionnaires. Pilot the phones menu. The selection of the parking zone
participants received an e-mail with a link by touching is illustrated in Figure 1. Parking time
forwarding them to the questionnaire or they filled started running after the touch to the parking zone
the paper questionnaire after a pilot period. The tag and it ran until user came back to her/his car and
number of customers participating in the pilots is touched again the tag on the windscreen. Hence,
reported in the table 1. Altogether, there were 224 SmartParking allowed charging from the exact time
pilot participants who used NFC enhanced services the car was parked.
in the pilots. Final response rates of questionnaires
of pilot 1, 2 and 3 were 94%, 85% and 72%,
respectively. The response rates can be considered to
be very high especially in pilots 1 and 2.
All pilot participants received Nokia 6131 NFC
phone for the pilot period. These NFC-enabled
phones were equipped with proper applications for
the pilot use. NFC related services and information
was available without a charge to the individual pilot
customers. Moreover, pilot participants were
allowed to use NFC enabled phones for free of
charge during the pilots, also for their own purposes.

3.1 Pilot 1: SmartParking Figure 1: A user selects a parking zone by touching an

NFC tag on the fare collection machine.
The objective of the pilot was twofold. Firstly, the
aim was to provide flexible and efficient mobile In-house parking was initiated when customers
parking solutions to the citizens. Secondly, traffic arrived at a gate of a parking hall and they touched

249
ICE-B 2008 - International Conference on e-Business

an NFC reader near the gate. The touch to the NFC also for the payment of their lunches. The payment
reader opened the gate and parking time started transaction was conducted by tapping the NFC
running. Customers repeated the same operation enabled phone on the contactless point of sales
when they left the parking hall, which stopped the terminal. The used terminal was a commercially
parking time. In addition to the NFC-based parking available RFID based terminal provided by a cash
application, customers had a possibility to download register provider.
information to their phones by touching info tags,
which were attached on parking meters. 3.3 Pilot 3: Oulu City Theatre
3.2 Pilot 2: Fast Track of Restaurant The target of the pilot was to develop theatre’s
Pannu service processes and deliver more added value to
theatre customers with more flexible and easier
An initiating problem behind the pilot was that busy NFC-based activities. Since the queuing and slow
workers want to have fast service at the restaurant ticket checking activities were seen as bottlenecks,
because of their short lunch break. However, since one of the main activities in this pilot was a mobile
most employees come to the restaurant at the same phone specific ticket stored in the back-end system.
time, traditional service, where a waiter asks for The ticket was checked before entering to the play
orders, may be quite slow. In order to find a solution by ticket checkers’ NFC enabled mobile phone. In
for this problem, a new “fast track” service concept practice, a theatre-goer and a ticket checker touched
was developed. The fast track provides an each others’ mobile phones. Theatre pilot included
opportunity for the busy customers to select their also other NFC related activities, such as, pre-
lunch by touching the RFID tagged menu with their reservation for interval refreshments, reclaiming of a
NFC enabled mobile phones, which is illustrated in programme for the play, downloading information
Figure 2. The lunch order was directed straight to via info tags and downloading video trailers from
the kitchen without intervention of the waiter. In smart posters.
addition to placement of lunch order, customers The pilot participants, 141 persons with average
were also able to pay for their lunch using NFC age of 44.5 years, tested and used NFC based
phones and information tags provided a possibility services before and during the plays. Each
to download information to the mobile phone. participant visited at least in one play. The
participants were informed about the theatre pilot by
sending related information to companies locating in
Oulu region. Selected participants were groups of
employees from diverse industry areas. After the
participants had visited the theatre and watched the
play, they received web questionnaire to give a
feedback about the use of NFC in theatre
environment.

4 RESULTS FROM THE PILOTS

Analysis of the data obtained from the pilots reveals
Figure 2: Touching a tagged lunch menu with an NFC
enabled mobile phone. several common challenges for the adoption of NFC
based services. In the following these challenges are
The objective of the pilot was to test the divided into two categories based on the actor whom
suitability of the NFC technology to the restaurant the challenge primarily concerns.
environment and find out if it could ease the rush
time at the restaurant. Restaurant Pannu invited 27 4.1 Customers’ Barriers to Adoption
of their regular lunch customers to participate in the
pilot. Customers, who were selected to the pilot The use of info tags revealed to be easy regardless of
group, used restaurant frequently during lunch use experience concerning NFC-enabled mobile
hours, but they also hoped to have faster lunch time phone. Use period of NFC-enabled mobile phone
service. 11 customers used the NFC enabled phone was very short in the theatre pilot when compared to
other pilots. However, users experienced touching

250
 ADOPTION OF NEAR FIELD COMMUNICATION TECHNOLOGY IN BUSINESS TO CONSUMER SERVICES

and learning to use the info tags very similar in the very quickly. Therefore, the other services in a
theatre pilot and the restaurant pilot. A following parking meter were easily ignored. Moreover, many
table present average values concerning experience of those users, who used info tags, did not see the
about learning to use info tags and successful content of tags necessary for themselves during
touching. The scales in questionnaires were from 1 parking. For these reasons, the use of info tags was
(“very difficult”) to 5 (“very easy) concerning quite low in the SmartParking pilot.
learning and successful touch and from 1 (“totally
useless”) to 5 (“very useful”) concerning usefulness
of info tags’ content.

Table 2: Average values concerning experiences about

learning, touching and usefulness of tags’ content pilot-
specifically.
Pilot/Feature Theatre Restaurant
Pilot Pilot
Learning 4.59 4.64
Successful touch 4.63 4.60
Usefulness of info tag content 3.00 3.07
Opinions about usefulness of info tags’ content
varied among pilot users. 14 out of 23 users used
info tags in the restaurant pilot. Info tags were
touched 143 times during the pilot. Thus, there were
approximately 5 touches to info tags per user in the
pilot. Those, who did not use info tags in restaurant, Figure 3: Info tag menu in a parking meter.
stated that they did not have need to use those tags
or content of tags was not interesting. Three of those None of the pilot users had ever used an NFC-
users, who used info tags, saw that info tag content enabled mobile phone before the pilot tests. Thus, in
was very useful for them. However, many users did order to use NFC-based mobile services after the
not see available info tag content very useful for pilot period they should invest in a new NFC-
themselves in lunch time. enabled mobile phone. Although many users were
Set of info tags was also available for pilot users willing to use NFC-enabled mobile services also in
in the theatre pilot. Experience about usefulness of the future, a demand for buying a new NFC phone
info tags’ content was very similar with experiences was experienced as a confine to the future use.
in the restaurant pilot as shown in the table 2. Some Furthermore, because of the limited supply of NFC
users saw that info tags were nice extra to the theatre services, many persons were not ready to buy a new
experience. In contrast, some users stated that info NFC-enabled mobile phone just for few services.
tags did not provide interesting information content In the theatre pilot many users saw a price of
for them. Info tags were touched 275 times in the services as an essential criterion for the future use of
theatre pilot. This means approximately 2 touches to services in theatre environment. In the restaurant
info tags per user during the pilot. Smart poster tags, pilot 6 out of 22 users would be ready to pay for the
which offered information about theatre plays, were services available during the pilot. Some users stated
more popular. They were touched 345 times during a that they would use these services also in the future,
pilot period. Most of those users, who did not use if the service would be inexpensive. Overall, many
smart posters, told they did not have time to use users required that services should provide a clear
poster tags or they did not even notice these tags in benefit for them before they would be ready to pay
the theatre. for the services.
The placement of info tag application in
SmartParking pilot differed from the tag placement 4.2 Service Providers’ Barriers to
in two other pilots. The info tags were placed Adoption
outside where those were attached to the parking
meters (Figure 3). Context of use was clearly Service providers’ challenges and barriers to the
different than in other pilots, because car drivers adoption are based on the interviews with business
wanted usually to get through the parking payments actors. Interviewed actors emphasized the fact that

251
ICE-B 2008 - International Conference on e-Business

benefits of the NFC technology are not expected to Proposition 1. Real added value is a required for
be realized immediately. However, they all believed the customers’ adoption of NFC enabled services,
that the NFC technology might somehow benefit especially if customers have to pay for it.
their business in the future. As the results presented in the section 4 point out
Firstly, interviewees experienced a low consumers require meaningful content in the
penetration rate of NFC enabled mobile phones to be information tags, which provides added value for
main obstacles in short term adoption of NFC them. Furthermore, a content of info tags should be
technology. The penetration rate was considered to suitable with environment and the context of use.
be especially important in the cases of SmartParking Significance of the environment was emphasized in
and city theatre, which require large customer bases the theatre pilot, as theatre play related information
to be justifiable from the business perspective. was more interesting than a content, which was not
Secondly, substitutive technological solutions directly related to the theatre visit. In the
were considered as a notable challenge to the NFC SmartParking pilot car drivers tended to use a
adoption in piloted use cases. This challenge was keypad of a mobile phone more frequently than zone
highlighted in the discussions with managers of tags in a parking zone selection. It is rather easy to
restaurant and parking operator. According to the remember the different parking zones in the city
restaurant manager’s opinion, the need of NFC centre of a relatively small city, thus the perceived
based ordering application may be questioned when value of the NFC application can decrease in
waiters’ wireless handheld devices become more specific environment more than in another.
general. However, interviewee considered NFC to More meaningful services and content might also
be very potential complementary payment increase customers’ willingness to pay for the
technology to be used in restaurants. From the services. As restaurant and theatre pilots pointed out,
viewpoint of parking operator, some other willingness to pay for the service and content varies
technological solutions, such as long distance RFID clearly between customers. Many customers were
or licence plate recognition, are competing in the not even willing to pay at all for the service or
same category with NFC when investment decisions content available in the pilots. More personalised
are made. Thus, interviewee from parking operator services might be one solution for this problem,
stressed that the new technology should bring some since those could deliver more focused content to
additional value in comparison with other potential the customers.
technologies. To conclude, before an implementation of an
Lastly, interviewees regarded that the NFC enabled mobile service, it would be essential to
interoperability of NFC with existing information identify information and service contents, which can
systems is of considerable importance. provide added value to the customer in specific
Municipality’s parking supervisor stated that it environments. This is even more important in cases
would be very difficult to start using new parking where service provider requires that larger customer
solution that would be incompatible with existing groups would be ready to pay for the NFC-based
mobile parking systems. Importance of services.
interoperability was also noticed in restaurant
environment. This was stressed in the interview with Proposition 2. Penetration rate of the NFC enabled
the restaurant’s management, since some of the phones has to be substantially higher before a
value that customer perceived from the mobile business to customer service intended for large
payments was eroded because of the integration customer groups can be profitable.
problems of the NFC point-of-sales reader and the Although there are several niche solutions where
cash register. NFC technology can be currently used successfully,
large scale business to customer services are still
waiting to be rolled out. The main reason for the
5 DISCUSSIONS lack of large scale NFC enabled customer services is
the same as the interviews with companies and
In this section we propose a sequence of research public sector organisation pointed out: NFC services
propositions related to the barriers to adoption of intended for the large customer groups require a
NFC technology and suggestions how these barriers wider diffusion of NFC enabled mobile phones.
can be lowered or even totally removed. Following Therefore, the major challenge is how to get NFC
propositions are provided with illustrations and enabled mobile phones to the users who do not even
explanations from previously introduced pilot cases. necessarily want a new phone. In addition, currently

252
 ADOPTION OF NEAR FIELD COMMUNICATION TECHNOLOGY IN BUSINESS TO CONSUMER SERVICES

the shortage of different models of NFC-enabled A field of research is quite unanimous that the
mobile phones limits users’ possibilities to select a measurement of benefits of mobile technology is
phone, which responds to their needs. The NFC difficult and the assessment of potential future
phone model selection is expected to increase in the benefits is even more difficult (e.g. Heijden &
near future as different manufacturers have Valiente, 2002). In spite of that it may be critical
published plans to introduce new NFC enabled from the service provider’s viewpoint to be able to
mobile phone models. So, at the moment, it seems assess the business value of an investment to new
that this challenge might be gradually fading. technology. Image benefits of being in the forefront
of technology developers may be counted as a
Proposition 3. Taking into use of NFC services
benefit by some business actors. However, when
should be made as easy as possible for a customer.
companies from more traditional industries plan and
A basic idea behind the NFC enabled mobile compare the investment on NFC technology to other
services is that they should be easy to use and technologies, pointing out the business value is
adoption of use should not be difficult. Hence, in essential. Therefore, more effort should be put on
order that the adoption of a new NFC-based service the assessing the total benefits of the NFC
would be as fluent as possible, it should not require technology.
too many preliminary operations from a user before As opposed to the benefit assessment, more
the actual use of the service. Different initial effort should be also set to assessment of the total
preparations, such as registration and requirement to costs of the NFC technology. Many indirect costs,
install an application, reduce consumers’ interest on such as management of the tags or protecting the
using the service. At worst, those can completely confidentiality, integrity and availability of the
prevent the adoption of an application. services should be remembered when planning to
In the pilots of this study, users received pre- implement touch based services for customer
installed and configured NFC enabled phones for the markets. These issues may turn out to be of
pilot use. Therefore, users were able start the use of significant importance if NFC solutions are used in
phone immediately without time-consuming public places where those are left without the
preliminary operations. In commercial use the pre- supervision and those may be exposed to vandalism
installation of all NFC related application to the or difficult weather conditions.
mobile phone is not likely to be feasible. One
possibility to solve this problem is to offer settings
and configurations into a mobile phone by touching 6 CONCLUSIONS
an NFC tag. In that way the easiness of taking NFC
services in to use might be ensured.
NFC technology makes it possible to implement
Proposition 4. NFC service should be made mobile services and applications, which are easy to
interoperability with an existing technological learn because of intuitive and natural interaction
infrastructure. technique. Although NFC technology has already
been successfully adopted in some business to
Fourth proposition is very important in the cases
business services, there are still some challenges to
where NFC based service is taken into use alongside
be overcome before the wide scale adoption of NFC
with other information systems. The importance of a
in business to consumer services can be reached.
fluent integration is emphasized, because it may be
Based on the results of three pilot cases, which
very expensive and complex for the service
were analysed in this study, the main barrier to the
providers to maintain two concurrent technological
short term adoption is the shortage of NFC enabled
solutions. Major issues may arise, for example, if the
mobile phones. Also other challenges were
service provider of an existing information system is
identified, however, these challenges can be taken
not willing or capable to integrate NFC based
into account at the level of individual company
services to the existing system. This may result in
when NFC enabled services are planned and
high costs, if the whole information system has to be
implemented. Therefore, these issues should be
redesigned or it may even totally prevent a company
carefully considered before the implementation
from adopting a new NFC based service.
phase of the service.
Proposition 5. Business value of NFC technology Our research has some limitations, which leaves
should be more clearly represented. place to further research. Firstly, NFC services were
provided free of charge to the customers during the
pilots, which may not necessarily be the case in real

253
ICE-B 2008 - International Conference on e-Business

life. Although customers were asked about the Jaring, P., Törmänen, V., Siira, E., and Matinmikko, T.,
willingness to pay for the services, more realistic 2007 Improving Mobile Solution Workflows and
long term pilot should be conducted to find out real Usability Using Near Field Communication
business potential of the services. Secondly, the Technology. In Proceedings of the European
Conference on Ambient Intelligence 2007, Darmstadt,
selection of pilot participants was mainly based on Germany, Lecture Notes in Computer Science, 358-
the customers’ own eagerness. For example in the 373.
SmartParking case, the user segment of pilot Kaasinen, E. 2005. User acceptance of mobile services –
customers was mainly restricted to frequent parkers. value, ease of use, trust and ease of adoption. Espoo:
However, it would be important to know how VTT Publications 566. http://www.vtt.fi/inf/
customers who need parking services less frequently pdf/publications/2005/P566.pdf
experience the service. Hence, future research Kargin, B., Basoglu, N., 2006 Adoption Factors of Mobile
should pay focus on different customer segments Services. In ICMB’06 International Conference on
Mobile Business.
and their willingness to adopt NFC services.
Nysveen, H., Pedersen, P.E., Thorbjornsen, H., 2005
Intentions to Use Mobile Services: Antecedents and
Cross-Service Comparisons. Journal of the Academy
ACKNOWLEDGEMENTS of Marketing Services 33(3).
Riekki, J., Salminen, T., Alakärppä, I., 2006 Requesting
Pervasive Services by Touching RFID Tags. Pervasive
This work was done in the SmartTouch
Computing. 5(1).
(www.smarttouch.org) project (ITEA 05024), which Rogers, E., 1995 Diffusion of Innovations. The Free Press.
is a project within ITEA 2 (Information Technology New York, 5th edition.
for European Advancement), a EUREKA strategic Rukzio, E., Leichtenstern, K., Callaghan, V., Holleis, P.,
cluster programme. The SmartTouch project has Schmidt, A., Chin, J., 2006 An Experimental
been partly funded by Tekes, the Finnish Funding Comparison of Physical Mobile Interaction
Agency for Technology and Innovation. Techniques: Touching, pointing and scanning. In
Dourish, P., Friday, A. (eds.) UbiComp 2006, The 8th
International Conference on Ubiquitous Computing.
LNCS, vol. 4206, Springer, Heidelberg.
REFERENCES Välkkynen, P., Niemelä, M., Tuomisto, T. 2006
Evaluating Touching and Pointing with a Mobile
Anokwa, Y., Borriello, G., Pering T. & Want, R. 2007. A Terminal for Physical Browsing. In NordiCHI 2006:
user interaction model for NFC enabled applications. Changing Roles.
In Fifth Annual IEEE International Conference on
Pervasive Computing and Communications
Workshops, 357–361.
Balaban, D., 2007 NFC: Ready to Roll? Card Technology
Today. 12(3).
Bouwman, H., Carlsson, C., Molina-Castillo, F., Walden,
P. 2007 Barriers and Drivers in the Adoption of
Current and Future Mobile Services in Finland.
Telematics and Informatics 24.
Calvet, J.C., 2005 The Role of RFID in the Mobile Phone.
Teletronik. 3(4).
Carlsson, C., Hyvönen, K., Repo, P., Walden, P., 2005
Asynchronous Adoption Patterns of Mobile Services.
In Proceedings of the 38th Hawaii International
Conference on System Sciences. IEEE Computer
Society Press.
Davis, F.D. 1989. Perceived usefulness, perceived ease of
use, and user acceptance of information technology.
MIS Quartely:13/1989, 319-339.
Heijden, H., Valiente, P., 2002 The Value of Mobility for
Business Process Performance: Evidence from
Sweden and the Netherlands. In ECIS 2002.
Häikiö, J., Isomursu, M., Matinmikko, T., Wallin, A.,
Ailisto, H., Huomo, T., 2007 Touch-based user
interface for elderly users. In Proceedings of
MobileHCI, Singapore.

254
 TAXONOMY FOR MOBILE TERMINALS
A Selective Classification Scheme

Gunther Schiefer and Michael Decker

Institute of Applied Informatics and Formal Description Method, University Karlsruhe (TH), 76128 Karlsruhe, Germany
[email protected], [email protected]

Keywords: Simple Phone, Feature Phone, Smartphone, Handheld, Web-Tablet, Netbook, Mobile PC, Tablet-PC,
classification, mobile Terminal.

Abstract: There is a great number of different types of mobile computing devices like cellular phones, Handhelds and
notebooks. However in literature so far there is no common accepted definition for these terminal types. Es-
pecially the term “Smartphone” is used for very different types of mobile phones. In our article we therefore
propose a comprehensive taxonomy for mobile terminals that clearly distinguishes between different classes
of terminals.

1 MOBILE TERMINALS is that mobile computing is a rather new technology

so a market consolidation like that for personal desk-
Everyday life without mobile terminals is almost top computers didn't take place yet. There are also
unthinkable. With the increasing use of them for a many application scenarios for mobile computers
multiplicity of tasks the equipment variants increase with different requirements so it is necessary to have
likewise. At present the most popular mobile termi- mobile terminals especially developed for individual
nals with the highest penetrations rates are mobile fields of applications. To get an impression of the
phones and Feature Phones. According to Kuhn great variety of mobile terminals one can refer to the
(Kuhn, 2004) a mobile terminal is a portable com- WURFL1 project which assembles a profile database
puter-assisted communication system with inde- with information about capabilities of different mo-
pendent data processing capacity, which can com- bile terminals: the file comprehends over 5.000
municate as endpoint of a wireless connection with different profiles. Some software products for mo-
other IT-systems. “Independent data processing bile terminals are available in customized versions
capacity” means that the mobile terminal is equipped for several hundred types of mobile terminals, e.g.
with random access memory (RAM), a central proc- “Yahoo!Go”2 a mobile applications that provides
essing unit (CPU) (this implies programmability), access to services like e-mail, news, weather fore-
own network-independent power supply (battery) casts and maps.
and the possibility to directly interact with human Nevertheless we can identify distinct classes of
users by providing means for data input (e.g. keys, mobile terminals with common characteristics.
microphone, touchscreen) as well data output (e.g.
display, light emitting diodes, speaker). Thus radio
frequency identity tags (RFID), Smartcard etc. are 2 CLASSES OF TERMINALS
not mobile terminals in the sense of the above defi-
nition because they don’t dispose over an independ- Within the range of the mobile terminals there are a
ent power supply and a user interface. Computers number of different types. They differ in the charac-
integrated into machines (e.g. vending machines, teristics listed above. There are different approaches
machines in factories) are also not considered as to divide the terminals into different classes (see for
mobile terminals because a human user can’t carry example (Noesekabel and Lehner, 2002), (Meier,
them along. 2002) or (Roth, 2005)). The borders between the
Mobile terminals differ particularly with regard
to their size, their use and their performance for the 1
http://wurfl.sourceforge.net/
execution of mobile applications. One reason for this 2
http://mobile.yahoo.com/go

255
ICE-B 2008 - International Conference on e-Business

Mobile terminals

In narrow sense In broad sense

Wireless mobile Board-

Mobile phones Smartcards
Computer computer

Simple Feature Smart Mobile Internet Mobile

Handhelds
Phones Phones Phones Devices Standard PC

Tablet-PC Subnotebooks

Figure 1: Classification tree of mobile terminals.

different terminal classes are seen differently by regarded in our classification (see also Figure 1)
different authors and are often not clear. For exam- have the ability for bidirectional wireless communi-
ple according to Brome and Zeman (Brome and cation and have an independent power in the form of
Zeman, 2008) and Canalys (Canalys.com, 2004) the an accumulator. Hereby the often vague defined
classification into Feature Phone and Smartphone is term Smartphone will be clearly separated from
determined by whether additional software is exe- Feature Phone and Handheld.
cuted directly by the operating system or by a run-
time environment like Java ME. On the one hand 2.1 Mobile Standard PC
this distinction isn’t obvious, on the other hand there
are terminals which can do both. The wireless mobile standard computers include
Mobile terminals can differ in the following cri- laptops/notebooks, subnotebooks (also called net-
teria (list is not complete): books) and Tablet-PC (including ultra mobile PC
(UMPC)). Subnotebook is the name for a particu-
• Size and weight larly small and light notebook, whose equipment is
• Input modes very similar to those in a full notebook, however has
• Output modes no optical drives (CD/DVD-ROM-drive) and a
• Performance smaller keyboard. An UMPC is a Tablet-PC which
• Kind of usage is maximally 8 inches large, weighs maximally
• Communication capabilities 900 grams and has a display with a resolution of at
• Type of operating system least 800x480 pixels; Keyboards can be attached
• Expandability externally. On these terminals the conventional
desktop operating systems (Linux, Windows) can be
Some of these characteristics can be used to dis- used as well as the compatible software packages
tinguish classes of terminals from each other. The like office suites. Substantial distinguishing features
following section gives criteria for the classification are the size of chassis and display, the weight, the
of mobile terminals into distinct classes without the
need to do a deepened technical analysis. All classes

256
 TAXONOMY FOR MOBILE TERMINALS - A Selective Classification Scheme

integrated peripherals (e.g. optical Drive included) sion modules. Handhelds usually have a touch-
and the operation kind of the terminals. sensitive display and can be used with a pen (or the
Notebooks with wireless communication capa- fingers), or they have a text keyboard and navigation
bilities are mobile terminals according to definition keys for it. Increasingly there are terminals which
given above. However in the following considera- can be operated through both of them. Terminals are
tions we won’t focus on them because our article is only classified as Handheld if they cannot communi-
more concerned with smaller and lighter terminals cate in communication networks for mobile teleph-
that can be used by a walking or standing user. ony like GSM or UMTS.
Notebooks usually need a „stand space“ like a table
or someone’s lap. Tablet PCs on the other hand are a 2.4 Smartphones
special subclass of the mobile standard computers,
because they have a touch-sensitive display and The term “Smartphone” designates a combination of
therefore can be used while standing with only one Feature Phone (see below) and Handheld with usu-
hand so they can be operated without a stand space. ally somewhat smaller dimensions, smaller display
and partly reduced function range in comparison to
2.2 Mobile Internet Devices the Handheld. They form their own class. On the one
hand mobile computers which have the characteris-
Beside the mobile standard PC there is a terminal tics of Handhelds and the ability to communication
class “Mobile Internet Devices” (MID) (also called over mobile telephony networks and have almost
Web-Tablet or mobile Thin Client) with similar everywhere and „seamless“3 connection to the Inter-
equipment and display size, however clearly reduced net belong to this class. On the other hand portable
function range and normally no mechanical drives. radio terminals with the characteristics of Feature
Characteristically these terminals are equipped like Phones that have extended input modes belong to
Handhelds (see below) with an operating system in this class, too. The extended input mode can be
the read-only memory (ROM) and are immediately provided by a touch-sensitive display (which can be
ready for use after switching them on. Their function used with the fingers or a pen) or a complete text
range is similar to the Handhelds and they are keyboard.
mainly used for browsing the web and as client for
terminal server sessions. They depend mostly on a 2.5 Feature Phones
constant wireless network connection for the fulfil-
ment of their tasks. The Web-Tablets (like the Tablet Feature Phones are mobile phones (see below) with
PC) can be operated through a touch-sensitive dis- usually somewhat larger display and extended func-
play, the mobile Thin Clients usually through a tion range. The primary use purpose of the Feature
keyboard. Phones lies in mobile communication, which in-
cludes both the voice communication and text-based
2.3 Handhelds communication (SMS, MMS, e-mail etc.). For this
terminal class the ability for wireless data communi-
Handhelds are small computers, which — like the cation over the mobile communications networks
name already suggest — can be held in one hand. (e.g. GPRS or UMTS) and the absence of the ex-
They are called often also Personal Digital Assistant tended input modes are characteristic. These are
(PDA). Historically these were originated from the limited to a number keyboard and a few additional
pure organizers, which could be predominantly used keys. Particularly terminals which support not only
for personal information management (PIM) appli- the GSM standard but also UMTS are equipped with
cations like calendar, address book or personal a Web browser and software to play back multime-
notes. Today's Handhelds have clearly higher per- dia contents. On these terminals the user can install
formance and are usually equipped with multimedia additional software. Since many terminals have
capabilities. As written above, only Handhelds who proprietary operating systems, in most cases addi-
are able to establish a wireless connection are re- tional software is offered on basis of Java Micro
garded here. They are equipped usually with an
operating system in the ROM and are immediately 3
ready for use after switching on (no boot-time Seamless means that these terminals change unno-
needed). Apart from the pre-installed programs in ticed and without further user interaction inde-
the ROM, users can install further applications in the pendently from one radio cell to another (soft
RAM or the flash memory and attach various exten- handover) and give the user the impression of only
one large radio cell.

257
ICE-B 2008 - International Conference on e-Business

Edition (Java ME), which is supported by nearly nected to a mobile terminal using a cable or blue-
every Feature Phone manufactured today. tooth.

2.6 Simple Phones

3 CONCLUSIONS
The classical mobile phones (also termed “cellular
phone”) were developed primarily for mere voice In this article we distinguished the mobile terminals
communication. However nowadays text-based clearly from other mobile devices. With the intro-
communication in the form of the short message duction of the term Feature Phone, which is used
service (SMS) is a standard feature of mobile rarely by other authors, mobile phones can be di-
phones. For Simple Phones it is characteristic that vided selectively into different classes. The term
they are not able to use wireless data communication Smartphone, which was until then fuzzy, become
based on TCP/IP (e.g. browsing the internet or re- consequently clear. With the introduction of the
ceive e-mail). The resolution and colour depth pro- class of special terminals, which comprehends all
vided by displays of these terminals was signifi- terminals for special tasks that cannot and should not
cantly improved in the recent years. The displays be assigned to the standard classes, the classification
became also bigger but they are usually noticeably of mobile terminals is altogether clear and complete.
smaller than those of Feature Phones. The function
range of these terminals can only be extended with
software in a small scale, e.g. by additional ring-
tones, pictures and multimedia animations or games, ACKNOWLEDGEMENTS
whereby these are strongly limited by the memory
free space and the performance of the processor. This work has been funded by the Federal Ministry
of Economics and Technology, Germany (BMWi,
2.7 Special Terminals Contract No. 01MD06012). The responsibility for
the content of this article lies solely with the authors.
Beyond the versatile usable standard terminals there
are a number of specialized mobile terminals which
cannot be arranged into the above classification. REFERENCES
These are for example e-book readers for storage
and reproduction of electronic books, electronic light Brome, R., Zeman, E., 2008. Feature Phone definition.
pens, which can read bar code or printed texts and Phone Factor, LLC.
process electronically or bar code readers with num- Canalys.com, 2004. Emea mobile device market continues
ber keyboard for the stocktaking in supermarkets: A upward trend in Q2. Reading, UK, Canalys.com.
worker scans the product code, enters the number of Kuhn, J., 2004. Commercial use of mobile applications:
available products in shelf and transfers all immedi- Results of the Delphi study "Mobile Business" (in
German), Regensburg, Germany, University Regens-
ately wirelessly to a server. burg.
Navigation terminals for travellers became quite Küpper, A., 2005. Location-based services: fundamentals
popular in the last few years. Those terminals can be and applications, Chichester, UK, Wiley.
used for locating as well as for route guidance. Meier, R., 2002. Generation of customer value by mobile
These terminals can calculate their own position services: Potentials by communication and networking
based on the signals received from several satellites (in German), Wiesbaden, Germany, Deutscher Uni-
(Küpper, 2005). The global positioning system versitäts-Verlag.
(GPS) of the US military or the Russian system Noesekabel, H., Lehner, F., 2002. Integration of web- and
GLONASS are already in use for many years and mobile-based services. IN REICHWALD, R. (Ed.)
Mobile communication: Value creation, technologies,
can be used nearly world-wide without fees. The new services (in German). Wiesbaden, Germany, Ga-
projected Galileo system by the European Union is a bler.
civilian system that will provide some service for Roth, J., 2005. Mobile Computing: Fundamentals, tech-
navigation free of charge; the full operation of this nology, concepts (in German), Heidelberg, Germany,
system is now scheduled to start in the year 2013. dpunkt-Verlag.
GPS (ore others) receivers can be integrated into a
mobile terminal but there are also separate GPS
modules (so called “GPS mouse”) that can be con-

258
 IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND
EVOLUTIVE UI-COTS COMPONENTS ARCHITECTURES

José Andrés Asensio, Luis Iribarne, Nicolás Padilla

Applied Computing Group, University of Almeria, Spain
{jacortes,luis.iribarne,npadilla}@ual.es

Rosa Ayala
Computers and Environmental Group, University of Almeria, Spain
[email protected]

Keywords: Trading agents, COTS User Interfaces, Model Transformation, Model-Driven Engineering, Cooperative Sys-
tems, e-Business Systems Modelling.

Abstract: Most of the complex (e-Business) information systems need to accomplish with the use of open standards.
Environmental Management Systems (EMS), for instance, state the international regulations of the ISO 14000
family, which establish the requirements to be fulfilled by an EMS in order to be accepted as such. One
of these requirements concerns the User Interfaces Development. Because of the variety of final users that
interact in this sort of complex information system (politicians, technicians, administrators, and so on) and
due to a great deal of information (some critical and confidential), it is important to have real and practical
scientific/technical proposals in order to build fast and efficient information exploitation systems. The human-
computer interaction (HCI) of these systems need user interfaces that adapt to the users profiles’ habits, and
with intelligent software agents that mediate by the users in the search processes, exploitation and decision-
making tasks. In this work we present a part of the SOLERES-HCI, a framework of the Soleres Project
for developing COTS user interfaces by using trading agents. Our studies are being applied for developing
advanced EMS and approaching Model-driven engineering techniques to the UI-COTS development.

1 INTRODUCTION half of the users and facilitate the information inter-

pretation tasks, the decision-making tasks and predic-
tion/prevention tasks (which are the most important).
Organizations and administrations that work with en-
In this work, we present a part of the framework
vironmental information for specific actions such as
Soleres-HCI that supports all the human-computer
territory management or planning and organization of
interaction issues of a complex environmental ma-
natural resources (among others), need to have inno-
nagement system (EMS). This portion of the frame-
vative quality information systems in order to guaran-
work concerns the developing of adaptable and evo-
tee the success of their everyday activities. There-
lutive user interfaces of the system by using: (a) tra-
fore, for the experts who operate the system not only
ding agents that intercede between user agents and
it is important to have reliable and updated informa-
the information, (b) developing of traders that fol-
tion that helps to make the most appropriate deci-
low model-driven engineering perspectives, and (c)
sions, as it happens in critical complex systems. Due
specialized Commercial Off-The Self (COTS) com-
to the variety of final users that cooperate with each
ponents for real-time user interface architectures.
other and interact with the system for decision mak-
The rest of the paper is structured as follows. Sec-
ing (for instance, politicians, technicians, administra-
tion 2 describes some issues for modelling user inter-
tors, etc.), it is also important to have: (a) exploita-
faces from UI-COTS. Section 3 continues with our
tion information systems (environmental, in this case)
SOLERES-HCI proposal, a human-computer inter-
to facilitate the human-human and human-computer
action perspective based on trading (ISO/IEC, 1997)
interaction and coordination; (b) intelligent user in-
and software agents. Finally, Section 4 explains some
terfaces that adapt to the users profiles’ habits and;
of our ongoing research in the SOLERES project.
(c) intelligent software agents that intercede on be-

259
 conflictos.

The latter refers to a virtual consultant or supervisor who will cooperate with other agents into
the pre-established multi-agent architecture and will respect an organization and cooperation
model (this project’s aim). The purpose of this cooperation and therefore, each MA’s job, will
ICE-B 2008 - International Conference on e-Business be to facilitate the tasks of information exploitation: they will interact with the searching
information system (next level in the system, later we will offer details of this level) and they
will filter the irrelevant information. The use of this sort of systems is rather useful because of
its natural capacity for representing and implementing organizational and social aspects that
can help to identify and/or solve conflicts.

SOLERES-HCI

2 MODELLING UI-COTS User User

In recent years, different approaches for the design Agent IU Agent IU

of user interfaces have been presented, most of them ...
Agent MA Agent MA
following model-driven UI designs, for instance:
(a) IDEAS (Interface Development Environment Cooperative system for the taking decision in group
(Trading-based multiagent architecture)

within OASIS) (Lozano et al., 2000): A methodo- New Trading Service

logy of UI development based on UML models. information (Trader)

(b) OVID (Object, View and Interaction Design)

Agent Agent
(Roberts et al., 1998): A methodology for UI de-
sign directed towards objects developed by IBM. ...
EPU EPU
(c) TERESA (Paterno, 1999): A tool for the UI gen-
eration by using ConcurTaskTrees (CTT). Searching Information System

(nota para Isa: traduce también el texto dentro de la figura)

(d) WISDOM (Whitewater Interactive System Deve- Figura
Figure 3. Arquitectura
1: Generalgeneral del sistemaof
architecture deSOLERES
información SOLERES.
project.
lopment with Object Models) (Nunes, 1998): A
methodological proposal for UML-based UI. Cooperative system for the decision taking in group
TRADING SERVICE (TRADER)
(e) UMLi (Pinheiro, 2002): An extension of UML Searching information system
notation for the UI design. Environmental
trading service
Environmental
Information
In the SOLERES team there are also current works on Documents
Comes
– 10/15 –
( EID )

UI modelling that use and extend the UML diagrams from the
IU

(Almendros and Iribarne, 2005) (Almendros and Irib- Query Register

Activate agent
arne, 2007). None of these works deal in depth with
trading UI-COTS modelling. Agent

On the other hand, our research aims to study

UI following the approach of Component-based Soft-
Parser
ware Development (CBSD) specialized for COTS Planner Environmental
Info Maps
templates
EIM / EID

components (Commercial Off-the-Shelf), (Meyers and ( EIM )

Oberndorf, 2001). There are really few works show-

ing realistic cases of IS development following the
COTS paradigm or using multi-component UI-COTS. Processes for the
generation of
environmental
In (Iribarne et al., 2004) we developed an experiment information maps

COTS composition in Geographical Information Sys- Original satellite

images ENVIRONMENTAL INFORMATION MAPS (EIM)

tems (GIS). However this approach doesn’t solve a

dynamic and evolutive UI-COTS perspective. ENVIRONMENTAL PROCESS UNIT

Furthermore, the tendency during the last few

years in CBSD is to facilitate the automatic integra- Figure 2: Some details of an EPU object.
tion of commercial components by means of the com- For the process of data exploitation we identify and
position (assembly) of their parts. The advances in structure the type of queries and the sort of informa-
trading services (or traders) have played an impor- tion suitable to be consulted, by using techniques of
tant role for it. Inspired on the trading model for hierarchical decomposition (i.e., trees, cut and prun-
COTS components in open-distributed systems (Irib- ing) and neural networks.
arne et al., 2004), (Iribarne et al., 2005) we develop a Figure 1 shows the (short) architecture of the
trading agent service for UI-COTS in EMS systems. SOLERES information system in our project frame-
work. At the user layer side (top), the system is de-
signed in order to be used for environmental decision-
3 TRADING IN SOLERES-HCI making tasks and in cooperation among different peo-
ple (system’s users) organized following different or-
Soleres HCI is the framework of the SOLERES en- ganized models (i.e. depending on their hierarchy).
vironmental management system, specialized in the This human-computer interaction and human-
human-computer interaction. This level of informa- human interaction is guided by a cooperative system
tion system follows the paradigm of Computer Sup- for the decision-taking tasks in group supported by a
ported Cooperative Work (CSCW) and uses techno- multi-agent architecture (next layer). Each user of the
logy of agents and multiagent architectures. cooperative system has an UI agent whose function is

260
 IMPLEMENTING TRADING AGENTS FOR ADAPTABLE AND EVOLUTIVE UI-COTS COMPONENTS
ARCHITECTURES

Figure 3: MDA steps of the trader views.

to adapt the UI to his/her needs by identifying his/her source pattern matched over source models and of
interaction habits with the UI. This UI agent mediates a target pattern that creates target models for each
between the user and (a) the rest of the system’s users match. Figure 4 shows a piece of the ATL code for
(who have their own UI agent), (b) the search infor- the transformation CIM/PIM of one of the five inter-
mation system (next layer) directly, or (c) an environ- faces of a trader agent: the Lookup interface. In the
mental software agent (environmental agent - MA). code, we can observe how associations, properties or
A virtual consultant cooperates with other agents operations of the Lookup class are created.
within a pre-established multi-agent architecture and
respects a model of organization and cooperation (that rule Lookup {
from f01:in MM!Lookup
is or project’s aim). The purpose of this cooperation
using {
(MA) is to facilitate the tasks of information exploita- O Lookup: out MM!Operation = ’null’;
tion: they interact with the search information system O get lookup if: out MM!Operation = ’null’;
(we call an Environmental Process Unit- EPU) and ...
filter the irrelevant information. The use of this sort }
to t01:out MM!Class( name<-’Lookup’,visibility<-#public )
of systems is quite useful because of its natural ca-
do { ...
pacity to represent and implement organizational and O Lookup<-thisModule.create Operation
social aspects that can help to identify and/or solve (’Lookup’, #public, ”, false, false, 1, 1);
problems. Figure 2 shows internal details of an EPU. O get lookup if<-thisModule.create Operation
(’get lookup if’, #public, ’Lookup’, false, false, 1, );
For implementing the trading agent, we follow a O get register if<-thisModule.create Operation
Model-Driven Engineering (MDE) perspective based (’get register if’, #public, ’Register’, false, false, 1, 1);
on the classical Model-driven Architecture (MDA) of A trader C Lookup C Trader<-thisModule.create Association
the OMG. In this perspective, the trader model stays (’trader C Lookup C Trader’, #public, false);
Pr trader C Trader<-thisModule.create Property
the three stages of the MDA: CIM/PIM/PSM (see Fig-
(’trader’,#private,’Trader’,false, false, 1, 1, false, ”, false, false);
ure 3). A trader meta-model (MM) is defined for each ...
stage, which describe the way to generate a diagram- }
matical model (UML) of the trader. }

To translate a model into views (stages), we use Figure 4: A piece of the ATL transformation
model-transformation techniques. In our case, we
use ATL for implementing the transformations of the Figure 5 shows the PIM trader model: a partial
trader. The language ATL (ATLAS transformation class diagram. In this framework we have also used
language) (Jouault and Kurtev, 2006) provides declar- the GMF/Eclipse to develop a tool for documenting
ative and imperative constructs. The declarative part CIM trader models. We use the OCL language of
of ATL is based on rules. Such rules consist of a UML/OMG for describing the semantical restrictions

261
ICE-B 2008 - International Conference on e-Business

project I+D TIN2007-61497 (SOLERES. A Spatio-

Temporal Environmental Management System based
on Neural-Networks, Agents and Soft. Components).

REFERENCES
Almendros, J. and Iribarne, L. (2005). Designing GUI com-
ponents from UML use cases. In IEEE Int. Conf. on
the Engineering of Computer-Based Systems, pages
210–217. IEEE Computer Society.
Almendros, J. and Iribarne, L. (2007). Visual Languages
for Interactive Computing: Definitions and Formal-
ization, chapter User Interaction and Interface Design
with UML. Idea Group Inc. Hersey, U.S.A.
Figure 5: Informal class diagram model of the trader. Iribarne, L., Troya, J., and Vallecillo, A. (2004). A tra-
of a trader. At the end, java code generation from ding service for cots components. Computer Journal,
4(3):342–357.
PSM trader models can be obtained by using a parser
written in MOFScript. Iribarne, L., Troya, J., and Vallecillo, A. (2005). The De-
velopment of Component-Based Information Systems,
For the implementation of the described frame- chapter Trading for COTS Components to Fulfil Ar-
work we have used the Eclipse platform and JADE chitectural Requirements. M.E. Sharpe, Inc.
(http://jade.tilab.com/) for agents.
ISO/IEC (1997). Information technology - open distributed
For space reasons, we have included here a part processing - trading function: Specification. Technical
of the Soleres-HCI framework. A more complete report, ISO/IEC 13235-1, ITU-T X.950.
version of the SOLERES project and details about Jouault, F. and Kurtev, I. (2006). On the architectural align-
the model transformations and implementations of ment of ATL and QVT. In SAC ’06: Proc. of the 2006
the trading agent (p.e., Java code) are available at ACM Symp. on Applied Computing, pages 1188–1195,
http://www.ual.es/acg/soleres. New York, NY, USA. ACM.
Lozano, M., Ramos, I., and González, P. (2000). User in-
terface specification and modeling in a object oriented
environment for automatic software development. In
4 ONGOING RESEARCH IEEE 34th Int. Conf. on Tech. of OO Lang. and Sys-
tems, pages 373–381. IEEE Comp. SP.
The following are the specific objectives of our ongo- Meyers, B. and Oberndorf, P. (2001). Managing Soft-
ing line of research. ware Acquisition. Open Systems and COTS Products.
Firstly, we plan to study the design of intelligent Addison-Wesley.
user interfaces. We will look into how to adapt the UI Nunes, N. (1998). Object modeling for user-centered de-
to different user’s profiles. velopment and user interface design: The wisdom ap-
Our scientific objective (complement to previous proach. PhD thesis, Univ. de Madeira.
one) is to study algorithms of dynamic-services com- Paterno, F. (1999). Model-Based Design and Evaluation of
position in UI-COTS components architectures. We Interactive Applications. Springer.
try to analyze this kind of component and its mar- Pinheiro, P. (2002). Object Modelling of Interactive Sys-
ketplace (i.e., how many exist, which types, how to tems: The UMLi Approach. PhD thesis, University of
define them, what kind of existing repositories, etc.). Manchester.
Furthermore, starting from previous works (Irib- Roberts, D., Berry, D., Isensee, S., and Mullaly, J. (1998).
arne et al., 2004), we plan to implement trading ser- Designing for the User with OVID: Bridging User In-
vices of UI-COTS components and their applications terface Design and Software Engineering. New Riders
in Cooperative Systems (specifically in environmental Publishing.
management systems.

ACKNOWLEDGEMENTS
This work has been partially supported by the EU
(FEDER) and the Spanish MEC under grant of the

262
 BUSINESS AND
SOCIAL APPLICATIONS
FULL PAPERS
 DEALING WITH BUSINESS PROCESS EVOLUTION
USING VERSIONS

Mohamed Amine Chaâbane, Eric Andonoff

Laboratoire IRIT, Université Toulouse 1, 2 rue du Doyen Gabriel Marty, 31042 Toulouse Cédex, France
[email protected], [email protected]

Lotfi Bouzgenda, Rafik Bouaziz

Laboratoire MIRACL, ISIMS, Route de Tunis, km 10, BP 242, 3021 Sakeit Ezzit, Sfax, Tunisia
[email protected], [email protected]

Keywords: Version, Business Process Evolution, Meta-model, Petri net with Objects.

Abstract: Competition in which enterprises and organizations are involved nowadays imposes them to often make
evolve their business processes in order to meet, as quickly as possible, new business or production
requirements. This paper proposes to adopt a version-based approach to support these dynamic changes of
business processes. This approach permits to keep chronological business process changes: it is then
possible to allow several instances of a same business process to own different schemas, each one
representing a possible schema for the considered business process. Consequently, this approach is very
suitable to deal with long-term business process evolution: it does not necessarily impose the adaptation and
migration of running instances of business processes to a new business process schema. The paper
contribution is threefold. First, it defines a meta-model for designing versions of business processes
considering the three dimensions of business processes: the informational, organizational and process
dimensions. Then, it introduces a taxonomy of operations for business process version management. Finally,
it proposes to formalize and visualize modeled versions of business processes using a Petri net-based
formalism, namely Petri net with Objects.

1 INTRODUCTION has mainly been addressed in the Workflow context

using two main approaches: an adaptive-based
Nowadays, the importance of business processes in approach and a version-based approach. The
enterprises’ and organizations’ Information Systems adaptive-based approach consists in defining a set of
(IS) is widely recognized. As a consequence, these operations supporting both workflow process
last few years, there has been a shift from data- schema changes, and adaptation and migration of
aware IS to process-aware IS (Aalst and al, 2007). their corresponding instances (Casatia and al, 1996),
However, even if important advances have been (Reichert and Dadam, 1998), (Kammer and al,
done in business process management, several 2000), (Rinderle and al, 2004). In this approach,
problems are still to be dealt. Among them, the only one schema is kept for all modelled workflow
business process evolution problem that can be processes. This approach has been investigated
posed as follows: how to support dynamic change of intensively and its implementations, ADEPT
business processes (Smith and Fingar, 2003), (Aalst (Reichert and Dadam, 1998) and JOpera (Heinis and
and al, 2003-a)? al, 2005), are probably the most successful
The competitive and dynamic worldwide Workflow Management Systems (WfMS) regarding
economic context, in which enterprises and workflow process’ schema evolution.
organizations are involved, lead them to often In the version-based approach, different
change and adapt their business processes in order to instances of a same workflow process can have
meet new business or production requirements. different schemas. Thus, it is possible to distinguish
Consequently, the business process evolution between temporary and permanent updates for
problem is really a relevant problem. This problem workflow processes since it is possible to keep track
of chronological workflow process changes, each

267
ICE-B 2008 - International Conference on e-Business

one representing a possible schema for the of business process versioning only considering,
considered workflow process. what is called in the workflow literature, “process
In the workflow context, where long-term model”. Such a model describes tasks involved in
processes are involved, adaptation and migration of the process and their coordination. But, using only
workflow process instances according to a new this model is not enough to have a comprehensive
schema are not always easy and are sometimes description of business processes (Aalst, 1999). Two
impossible (Casati and al, 1996). So, it is important others models have to be considered: the
to be able to manage different schemas for a organizational and the informational models. The
workflow process in order to allow several instances organizational model structures the business process
of this workflow process to own different schemas actors and authorizes them, through the notion of
(Kradolfer and Geppert, 1999). Thus, the version- role, to perform tasks making up the process. The
based approach is a promising solution to deal with informational model defines the structure of the
business process evolution. documents and data required and produced by the
Versions are used in several fields of computer process. These two models are glued together with
science in which was highlighted the need to the process model since, in addition to the tasks and
describe evolution of real world entities over time. their coordination, the process model also defines
Thus, versions are used in the database field mainly the required resources (information, actors) to
in object-oriented databases (Cellary and Jomier, perform the tasks.
1990), (Sciore, 1994), or scientific databases (Chen Consequently, this paper proposes to revisit the
and al, 1996) but also for specific database business process evolution problem using a version-
application fields such as computer aided design or based approach and considering both organizational,
computer aided manufacturing (Chou and Kim, informational and process models of business
1986), (Katz, 1990). Versions are also used in processes. More precisely, this paper introduces:
software engineering to handle software
ƒ A meta-model for designing versions of
configurations (Kimball and Larson, 1991). Versions
business processes;
are also considered in conceptual models such as the
ƒ A taxonomy of operations for business process
Entity Relationship model (Roddick and al, 1993) or
version management;
the OMT model (Andonoff and al, 1996).
ƒ A formalization and a visualization of versions
Although versions are used in several areas of
of business processes designed with the
computer science, to the best of our knowledge, only
previous meta-model.
few efforts have been put on version management in
the business process (workflow) context (in the The remainder of this paper is organized as
remainder of the paper, the terms workflow and follows. Section 2 introduces the Business Process
business process will be used equally). (BP) meta-model we use for designing business
We distinguish two main contributions about process, while section 3 introduces the Versioned
versions of business processes in literature. Business Process (VBP) meta-model we propose for
(Kradolfer and Geppert, 1999) have proposed to deal business process versioning. More precisely, section
with dynamic workflow evolution, i.e. modification 3 first recalls the notion of version, then presents the
of workflow process schemas in the presence of versioning kit we propose for handling versions of
active workflow process instances, introducing business processes, and finally explains how the kit
versions of workflow process schemas. This work is merged with the BP meta-model to define the
has defined a set of operations for workflow process VBP meta-model. This section also gives an
schema modification and, if possible, a strategy for example of business process versioning. Section 4 is
migration of workflow process instances. Recently, dedicated to the dynamic aspects of the meta-model:
(Zhao and Liu, 2007) have also defended the it presents a taxonomy of operations for business
advantages of a version-based approach to face process version management. Section 5 presents our
business process evolution. More precisely, this proposition for both formalization and visualization
work proposes to model versions of workflow of workflow process versions using a formal model,
process schemas using graphs. It also presents a set namely Petri Net with Objects (Sibertin, 1985).
of operations enabling updates of graphs and defines Finally, section 6 stands our contribution according
two strategies to extract versions of workflow to related works and then concludes the paper.
process schemas from these graphs.
We believe that these two propositions need to
be revisited. Indeed, both (Kradolfer and Geppert,
1999) and (Zhao and Liu, 2007) addressed the issue

268
 DEALING WITH BUSINESS
D PROCESS EVOLUTION USING VERSIONS

2 MODELING BUSINESS FlowMark meta-model proposes a very detailed

PROCESSES description of workflow processes along with
involved data flows and actors. However, these
As mentioned before, a business process meta-model meta-models are very complex, specially with
must allow the expression of three complementary respect to the organizational dimension.
aspects, usually described through three different Consequently, we have defined our own meta-
interacting models: the organizational, informational model which fulfils the previous requirements: (i) a
and process models. The main important model is comprehensive meta-model considering three
the process model which defines component tasks complementary aspects of business processes
and their coordination, but this model also refers to (organizational, informational and process models),
the organizational model and to the informational and (ii) a business process virtual meta-model as it
model defining required and produced resources defines the core (basic) concepts of the three com-
before and after tasks execution (Aalst, 1999). plementary aspects of business processes. This meta-
Another important requirement for such a meta- model is shown in the UML diagram of figure 1.
model is its simplicity and efficiency: it must be In this UML meta-model, a business process is
comprehensive and must define the core (basic) either a composite process or an atomic process. A
concepts of the three complementary aspects of composite process is itself recursively composed of
business processes: it must play the role of a atomic or composite processes. It also uses a control
Business Process Virtual Meta-model, i.e. a minimal pattern, which participates to the definition of
meta-model for the design of business processes. business process coordination. In our meta-model,
This idea of Business Process Virtual Meta-model is and as in, for instance (Manolescu, 2001), the main
the same as the one of Workflow Virtual Machine control patterns described in the literature are
introduced in (Fernandes and al, 2004) to deal with provided. Some of them are conditional (e.g. if,
the development of a Workflow Management while…), while others are not (e.g. sequence,
System (WfMS) that supports changes in its fork…). Their semantics is the following:
workflow definition language(s). ƒ Sequence pattern: it permits the execution of
But does such a meta-model for business process processes in a sequential order;
modelling (i.e. meeting the previous requirements) ƒ If pattern: it allows processes execution
already exist, or do we have to define a new one by according to a condition;
ourselves? ƒ Fork pattern: it spawns the parallel execution of
Despite the standization efforts of the Workflow processes and waits for the first to finish;
Management Coalition (WfMC), different workflow ƒ Join pattern: it spawns the parallel execution of
or business meta-models exist in literature. The used processes but waits for all of them before
vocabulary differs from one model to another, and completing;
yet, so far, the workflow and business process ƒ While and Repeat patterns: they cyclically
community seem to not have reached an agreement execute a process while or until a condition is
on which model to adopt, even if XPDL, BPMN and achieved.
BPEL are standards recommended by the WfMC. is_composed_of 1..* Non Conditional Control
Some business process and workflow meta- 2..*
Composite Process
0..* uses 1
Control Pattern
Pattern

models proposed in literature mainly focus on the Business Process

0..1
has_pre-conditions
0..*
Conditional Control Pattern

0..1
process model -i.e. tasks description and their
0..*
Atomic Process 0..1 has_post-conditions has
1
1..* Condition

coordination- (e.g. (Zhao and Liu, 2007), BPEL, 1..* 1..* 1..*
contains 1..* Action

XPDL). Others also consider the informational produces

consumes
references

Process Model

model in addition to the process model (e.g. (Casati Informational Model

0..* 0..*
1 Organizational Model
Role 1..* belongs_to 1..* Organizational Unit
and al, 1995), (Kradolfler and Geppert, 1999), Informational resource 0..* 0..*
1..*
is_member_of

(Vossen and Weske, 1999), (Aalst and al, 2004)).

1..*
Actor
played_by
Application Data 1..* uses
Finally, some meta-models have a comprehensive Process Data System Data
Not Human Human Internal External
1..*

approach for business process modeling considering Data Document Form Data Repository Database Software Machine

the three complementary aspects (e.g. FlowMark Figure 1: The Business Process meta-model.
and its successors MQSeries Workflow and
WebSphere MQ Workflow (Leymann and Roller, An atomic process corresponds to a task to
1999), Exotica (Mohan and al, 1995), OpenFlow perform. It can have pre-conditions and post-
(Halliday and al, 2001)). For instance, as illustrated conditions, and executes one or several actions. An
in (Rosemann and zur Muehlen, 1998), the atomic process is performed by a role (belonging to

269
ICE-B 2008 - International Conference on e-Business

the organizational model) and consumes and/or described by only one version. The definition of
produces informational resources (belonging to the every new entity version is done by derivation from
informational model). Informational resources a previous one. Such versions are called derived
correspond to system data, process data (i.e. data, versions (e.g. E1.v1 is a derived version from
document or form), and application data (i.e. E1.v0). Several versions may be derived from the
database and data repository). A role is played by an same previous one. They are called alternatives (e.g.
actor belonging to some organizational units. An E1.v2 and E1.v3 are alternatives derived from
actor is a human resource or not (machine or E1.v1).
software). Finally, an actor may be internal or A version is either frozen or working. A frozen
external. version describes a significant and final state of an
Going back to control patterns, our meta-model entity. A frozen version may be deleted but not
only includes low level (basic) control patterns; all updated. To describe a new state of this entity, we
the high level workflow patterns of (Aalst and al, have to derive a new version (from the frozen one).
2003-b) are not considerer here (they are much more A working version is a version that temporarily
complex than what we need). In this way, the meta- describes one of the entity states. It may be deleted
model we propose could be seen as a Business and updated to describe a next entity state. The
Process Virtual Meta-model gathering the core previous state is lost to the benefit of the next one.
(basic) concepts of business process models.
3.2 The Versioned Business Process
Meta-model
3 MODELING VERSIONS OF
This meta-model consists of a versioning kit to
BUSINESS PROCESSES handle versions of business processes, which is
merged to the BP meta-model previously introduced.
First, this section briefly recalls the version notion as
it is introduced in object-oriented databases and 3.2.1 Versioning Kit
software engineering. Then, this section presents the
Versioned Business Process (VBP) meta-model: it This kit is very simple: it is composed of a class and
consists of a versioning kit to handle versions of a set of properties and relationships that make
business processes which is merged with the BP classes of the previous meta-model “versionable”. A
meta-model introduced before. Finally, this section “versionable” class is a class whose instances are
illustrates the VBP meta-model instantiation to versions (Katz, 1990).
design versions of business processes. Thus, for each of these “versionable” classes, we
define a new class which contains versions, called
3.1 Concept of Version “Version of…”. We also specify two new
relationships: (i) the is_version_of relationship
A real world entity has characteristics that may which links a class to its corresponding “Version
evolve during its life cycle: it has different of…” class, and (ii) the derived_from relationship
successive states. In object-oriented database which describes version derivation hierarchies. This
systems that provide version management, this entity latter relationship is reflexive. The underlying idea
is described by a set of objects called versions. A of our proposition is to describe both entities and
version corresponds to one of the significant entity their corresponding versions as indicated in figure 2.
states. Then, it is possible to manage several entity Consequently, (i) versions are therefore involved in
states (neither only the last one as in classical the process definition, and (ii) a couple (version,
databases nor all the states as in temporal databases). entity) is obviously created when the first version of
E n titie s E 1 .v 0 E 1 .v 1 E 1 .v 2
an entity is created. Regarding properties of these
“Version of…” classes, we introduce the classical
V e r s io n s
E1
E n .v 1 E 1 .v 3

En E n .v 2
version number, creator name, creation date and
E n .v 0 E n .v 3
status properties (Sciore, 1994).
Figure 2: Representing entities with versions. 3.2.2 Merging the Versioning Kit with the
Business Process Meta-model
As illustrated in figure 2, the entity versions are
linked by a derivation link; they form a version Regarding the process model, we propose to keep
derivation hierarchy. When created, an entity is versions for only two classes: the Atomic Process

270
 DEALING WITH BUSINESS
D PROCESS EVOLUTION USING VERSIONS

and the Business Process classes. It is indeed work centres, for instance work centre#1 (Pc#1), has
interesting to keep changes history for both atomic a technical problem and consequently is removed
processes (i.e. tasks) and workflow processes since from the process, two solutions are proposed to
these changes correspond to changes in the way that attempt keeping the production output: fixing
business is carried out. At the atomic (task) level, unqualified products or using employees for manual
versions describe evolutions in activity realization production. The business process is then updated as
while at the business process level, versions describe shown in figure 4(c) and 4(d).
evolutions in work organization (i.e. coordination of
activities). We defend the idea that atomic process 4(a)
Start
4(b)
Start
4(c)
Start
4(d)
Start

(task) and business processes versioning is enough Cof Em

Schedule Production
Cof, E-Co Ss
Schedule Production
Cof, E-Co
Schedule Production
Ss Cof, E-Co
Schedule Production
Ss

to help organizations to face the fast changing Pc#1, Pc#1, Pc#2, Pc#1, Pc#2, Pc#2,
Po Ma Po Ma Po Ma Ma, Ms Po Ma Po Pc#1 Po Ma
environment in which they are involved nowadays. Produce Produce Produce Fix Unqualified Produce Produce Produce
Products (Manual)
Regarding the other models, it is necessary to Em Em Em Qs
handle versions for the Informational resource class Quality Checking Quality Checking Quality Checking Quality Checking

from the informational model, and versions for the Pac

Packaging
Pac
Packaging
Pac
Packaging
Pac
Packaging
Role class from the organizational model. Regarding
this latter model, it is also possible to handle End End End End

versions for the Actor and Organizational Unit Roles

Em: Enterprise manager Ss: Scheduling service
Informational Resource
Cof: Customer order form
classes. However, keeping changes history for these Pc: Production work centre
Pac: Packaging work centre
Ms: Maintenance service
Qs: Quality service
Po: Production order form
E-Co: Electronic customer order form
Ma: Machine
two classes is, in our opinion, quite useless to handle
versions of business processes. Figure 4: Change in the Production BP.
Figure 3 below presents the new obtained meta-
model in terms of classes and relationships. The solution we provide to model theses
is_composed_of 1..* Non Conditional Control
derivation hierarchies consists in instantiating the
derived_from Composite Process 0..*
1 Control
Pattern
Pattern
VBP meta-model. The Business Process, Atomic
2..* uses
Version of
Business Process
derived_from is_version_of Atomic Conditional Control Pattern Process, Role and Informational resource
1..*
is_version_of
1..*
Version of Atomic
1 Process 0..1
“versionable” classes and their “Version of…”
Process 1 has_pre-conditions 0..*
1 1..*
1..* 1..*
1..* 1
has_post-conditions
0..* Condition 1 has corresponding classes are involved in this
Business
Process contains 1..*
Action
instantiation, along with the Composite Process and
produces consumes references
Process Model
Control Pattern non “versionable” classes. This
Informational Model
0..* 0..*
is_version_of
Role Organizational Model instantiation is visualized in figure 5.
1 1..* 1
is_version_of derived 1..* 1..* Organizational Unit
1 Version of Version of belongs_to
_from 0..*
1..* Informational resource Role Version of
Informational Business Process
resource 0..* 1..* 1..* derived_from relationship
Business Process is_version_of
inverse relationship
derived_from Actor [vbp1, v1, bp1, nil, cp1, ...]
System Application 1..* played_by [vbp2, v2, bp1, vbp1, cp2, ...] [bp1, Production, {vbp1}, ...]
Process [vbp3, v3, bp1, vbp2, cp3, …]
Data Data Data uses [vbp4, v4, bp1, vbp2, cp4, …] structure of the
business process Version of Atomic Process
Not Human Human Internal External version
is_version_of relationship [vap1, v1, ap1, nil, {vr1}, {vir1}, …] is_version_of
1..* [vap2, v2, ap1, vap1, {vr8}, {vir1,vir2}, …] relationship
Data Document Form Database Data [vap3, v1, ap2, nil, {vr2,vr4}, {vir3}, …]
Repository Software Machine
Composite Process [vap4, v2, ap2, nil, {vr3,vr4}, {vir3}, …]
[vap5, v3, ap2, vap3, {vr2}, {vir3}, …]
[cp1, cpa1, {vap1, vap3, vap6, vap8}, ...] [vap6, v1, ap3, nil, {vr1},{}, …]
[cp2, cpa1, {vap2, cp5, vap6, vap8}, ...]
Figure 3: The Versioned Business Process meta-model. [vap7, v2, ap3, vap6, {vr5},{}, …] derived_from
[cp3, cpa1, {vap2, cp6, vap6, vap8}, ...] [vap8, v1, ap4, nil, {vr6},{}, …]
relationship
[cp4, cpa1, {vap2, cp7, vap7, vap8}, ...] [vap9, v1, ap5, nil, {vr2,vr4,vr7},{}, …]
[cp5, cpa2, {vap3, vap4}, ...]
[cp6, cpa2, {vap9, vap4}, ...]
[cp7, cpa2, {vap5, vap4}, ...] [ap1, Schedule Production, {vap1, vap2}, ...]

3.3 Example control pattern which Control Pattern

Version
[ap2, Produce, {vap3, vap4, vap5}, ...]
Atomic [ap3, Quality Checking, {vap6, vap7}, ...]
Process [ap4, Packaging, {vap8}, ...]
structures the [ap5, Fix Unqualified Products, {vap9}, ...]
composite process [cpa1, Sequence] of Role
[cpa2, Join]
[vr1, v1, r1, nil, ...]
In order to illustrate the VBP meta-model [vir1, v1, ir1, nil, ...]
[vr2, v1, r2, nil, ...]
[vr3, v2, r2, vr1, ...]
derived_from
relationship

[vr4, v1, r3, nil, ...]

instantiation, we propose to use the example Version of
Informational resource
[vir2, v1, ir2, nil, ...]
[vir3, v1, ir3, nil, ...]
[vr5, v1, r4, nil, ...]
[vr6, v1, r5, nil, ...]
is_version_of
relationship
is_version_of
[vr7, v1, r6, nil, ...]
introduced by (Zhao and Liu, 2007). Because of Informational resource
[vr8, v1, r7, nil, ...]
derived_from
inverse relationship

Role
space limitation, we only focus on the instantiation [ir1, Customer order form,{vir1}]
[ir2, Electronic customer order,{vir2}]
[ir3, Production order, {vir3}]
relationship
is_version_of
relationship
[r1, Enterprise Manager, {vr1}]
[r2, Production work center, {vr2,vr3}]
of the process model of this example. [r3, Machine, {vr4}]
[r4, Quality service, {vr5}]
[r5, Packaging work center, {vr6}]
This example describes a production business
is_version_of
inverse relationship [r6, Maintenance service, {vr7}]
[r7, Scheduling service, {vr8}]

process and involves a factory, which owns one

production pipeline following the business process Figure 5: Instantiation of the VBP Meta-Model.
shown in figure 4(a). It includes several activities:
production scheduling, production using a work
centre, quality checking and packaging. In order to
increase its productivity, the factory decides to add a
new work centre. The business process is then
updated as shown in figure 4(b). If one of the two

271
ICE-B 2008 - International Conference on e-Business

4 OPERATIONS FOR BUSINESS version is definitive and does not need additional
PROCESS VERSIONING updates. A frozen version (i.e. a version in a Frozen
state) can be deleted or can serve as a basis for the
creation of a new version using the derive event and
In this section, we introduce a taxonomy of
to-derive operation. This new created version has the
operations for business process versioning. These
same value as the version from which it is derived
operations are defined as methods in the “Version of
from: its state is Working.
…” classes (“versionable” classes). They correspond
In addition to the previous state chart, these
to classical operations for versions (Katz, 1990):
operations require further details. For instance, the
create, derive, delete, update and froze, but this
Create and Update operations permit to add and
taxonomy also includes operations for version
delete references to the components of versions.
selection. Of course, create, delete and update are
These components change according to the
also available for the other classes of the meta-
considered type of versions: versions of business
model (non “versionable” classes), but their
processes, versions of atomic processes (tasks),
presentation is out of the scope of the paper.
versions of informational resources or versions of
This section introduces the create, derive, delete,
roles. Regarding the Derive operation, it can trigger
update and froze operations first giving a state chart
the derivation of versions of its components. The
which indicates when these operations are available,
sections below gives additional details for these
and second detailing the actions they perform
operations.
according to the classes in which they are defined.
Moreover, this section also discusses about version
selection, more precisely business process version 4.2 Creating and Updating Versions
selection.
Table 1 and 2 below give the semantics of these two
operations (Create and Update) according to the
4.1 State Chart for Versions
classes in which they are defined. The four “Version
of…” classes are considered.
The UML state chart of figure 6 indicates when
these operations are available. Some of them are
Table 1: Creating and Updating Versions of Business
available whatever the state of versions on which Processes and Atomic Processes (task).
they are applied, while others are only available in
some cases. In this state chart, each operation is Business Process Atomic Process
described using the notation Operation:Event/Action 1. Change structure 1. Change conditions
whose meaning is “for Operation when Event is 1.1. add/delete 1.1. add/delete pre-
triggered then Action is performed”. composite process in conditions (has-pre-
the workflow process conditions relationship)
Delete:
delete/to_delete
structure
Update:
update/to_update 1.2. add/delete atomic 1.2. add/delete post-
Working Freeze:
freeze/to_freeze Delete: process in the work- conditions (has-post-
Create: delete/to_delete
create/to_create flow process structure conditions relationship)
Derive:
Frozen 2. Change pattern 2. Change action
derive/to_derive 2.1. choose a pattern 2.1. add/delete actions
for a composite process (contains relationship)
Figure 6: State Chart for Versions of Business Processes. (use relationship)
3. Change information
When the create event is triggered by the a 3.1. add/delete input
version designer, the to_create action is performed information (consumes
to both create the entity and its corresponding first relationship)
version. The state of the created version is Working. 3.2. add/delete output
In this state, the version can be updated (update information (produces
event and to-update action). relationship)
It also can be deleted (delete event and to-delete 4. Change role
action): its state is then the final state of the chart. It 4.1. add/delete roles
also can be frozen (freeze event and to-freeze (references relationship)
action): its state is then Frozen. Triggering the freeze These two tables indicate that Create and Update
event, the designer means that the considered operations change according to the classes in which

272
 DEALING WITH BUSINESS
D PROCESS EVOLUTION USING VERSIONS

they are defined. However, they share the same and version hierarchy selection: Select, Slice,
general idea that is to give values to properties and Display, among others... Because of space
relationships of the considered classes. Moreover, limitation, the paper only details the version
relationships referencing versions may only selection operation and illustrates its use for business
reference frozen versions (i.e. versions in the Frozen process version selection.
state).
4.4.1 Select Operation
Table 2: Creating and Updating Versions of Informational
Resource and Role. This operation allows the selection of versions. Its
syntax is: Select(Class,Predicate) where Class is a
Informational Role
name of a VBP class containing versions (i.e. a
Resource “versionable” class) and Predicate a condition
1. Change software 1. Change actors permitting the filtering of versions.
1.1. add/delete 1.1. add/delete actors The result of this operation is a set of versions
software (uses (played_by relationship) verifying the predicate along with versions and/or
relationship)
objects that are (directly or not) linked to it by a
2. Change the structure 2. Change organization
relationship. In other words, the result of the Select
of information 2.1. add/delete
operation is a set of instances of the VBP meta-
resource organizational units
model linked (directly or not) to a version belonging
(belongs_to
relationship) to the “versionable” class on which the Select
operation is performed. We call such a group of
instances VBP-instances. This notion of VBP-
4.3 Derivation of Versions instances corresponds to the notion of Configuration
introduced for handling versions in Software
The Derive operation allows the creation of a new
Engineering (Kimball and Larson, 1991). It is also
version from an existing frozen one. The new
close to the notion of Database Version introduced
created version is a working version (its state is
in (Cellary and Jomier, 1990) in order to reduce the
working). Before being updated, the value of this
complexity of version management in object-
new created version is the same than the derived
oriented databases.
one. Moreover, derivation of a version may trigger
the derivation of other versions, which are linked to 4.4.2 VBP-Trees for representing
the derived one. Figure 7 below illustrates this
VBP- Instances
derivation propagation.
Regarding business process version selection, the
Derive Derive
Role Derive result of a Select operation performed to the Version
Atomic Process
of Business Process class is a set of business process
Derive
Informational Resource Derive Derive versions verifying the predicate along with instances
Derive Business Process (versions and/or objects) of the Composite Process,
Control Pattern, Version of Atomic Process, Version
Figure 7: Derivation Propagation.
of Informational Resource and Version of Role
classes which are (directly or not) linked to them.
In this case, a VBP-instance corresponds to a
This propagation is due to the composition
business process version along with versions and/or
relationships existing between Business Process,
objects linked to it. It can be represented as what we
Atomic Process, Informational Resources and Role
call an VBP-Tree from which we distinguish two
classes. Thus, derivation of an Informational
kinds of nodes: terminal nodes (leaves) and non
Resource version or a Role version triggers the
terminal nodes. Terminal nodes correspond to VBP
derivation of its corresponding Atomic Process
atomic processes while non terminal nodes
version. In the same way, derivation of an Atomic
correspond to VBP composite processes. A non
Process version triggers the derivation of its
terminal node is described by the following data
corresponding Business Process version.
structure:
4.4 Selection of Versions ƒ NodeName: name of the node (corresponds to
the name of the corresponding composite
In addition to the previous presented operations, we process);
also propose specific operations for version selection

273
ICE-B 2008 - International Conference on e-Business

ƒ CPName: name of the control pattern used for Table 3: Mapping Rules from VBP-instance to VBP-Tree.
the composite process; VBP-Tree concepts
VBP meta-model
ƒ Condition: optional property associated to
concepts
conditional control patterns;
Instance of Version of VBP-Tree
ƒ SetOfNodes: set of nodes (terminal or non Business Process class
terminal) composing it. Instance of Composite Non Terminal node
A terminal node is described by the following Process class
data structure: Instance of Version of Terminal node
Atomic Process class
ƒ NodeName: name of the node (corresponds to
the name of the corresponding atomic Moreover, this function uses a set of functions
process); permitting the handling of processes and nodes:
ƒ SetOfActions: set of actions to perform; ƒ IsAtomicProcess(i) indicates if i is an instance
ƒ PreCondition: condition associated to the of the Version of Atomic Process class;
execution of the atomic process; it must be ƒ BuildTerminalNode(i) returns the
evaluated to true to perform actions of the corresponding terminal node of an atomic
atomic process that the node represents; process i taking into account the relationships
ƒ PostCondition: condition associated to the flowing from i (has_pre-conditions, …);
atomic process after execution of actions of ƒ BuildNonTerminalNode(i) returns the
the node; corresponding non terminal node of the
ƒ ConsumesInformation: set of informational composite process i using the relationship
resources required to perform actions of the flowing from i (uses);
node; ƒ AddNode(n, tr) adds the node n to a VBP-Tree
ƒ ProduceInformation: set of informational tr.
resources produced by the performing the
actions of the node; This function is the following.
ƒ PlayedBy: role defining a set of actors able to Function BuildVBP-Tree (i:VBP-
perform the actions of the node. Instance):VBP-Tree
Local n:Node
For instance, the VBP-Tree corresponding to the Begin
third version of the Production business process If IsAtomicProcess(i) Then
(vbp3 i.e. Production.v3) introduced in section 3.3 is n = BuildTerminalNode(i)
visualized in figure 8. BuildVBP-Tree = AddNode(n, tr)
This VBP-Tree illustrates the structure of the Else
considered business process distinguishing terminal –- i is a composite process
nodes (visualized as ellipses) from non terminal n = BuildNonTerminalNode(i)
nodes (visaulized as rectangles). BuildVBP-Tree = AddNode(n, tr)
In fact, figure 8 only gives a simplified view of For Each c ∈ IsComposedOf(i)
BuildVBP-Tree =
the VBP-Tree since nodes are not described in BuildVBP-Tree(c)
details (according their corresponding structures Next c
defined before). End If
End
Sequence

Schedule Production Packaging

Join
5 FORMALIZING BUSINESS
Quality Checking
PROCESS VERSIONS: FROM
Fix Unqualified Products Produce
VBP-TREE TO PNO
Figure 8: VBP-Tree for Production v3. Representing versions of business processes as
VBP-Tree is not sufficient to visualize and formalize
The function implementing the mapping from a the semantics of the modeled versions of business
VBP-instance to a VBP-Tree uses the mapping rules processes. To compensate this drawback, we
given in Table 3 below. propose to use a Petri net-based formalism, namely
Petri Nets with Objects (PNO) (Sibertin, 1985) for

274
 DEALING WITH BUSINESS
D PROCESS EVOLUTION USING VERSIONS

workflow process version visualization and piece of code in which transitions’ variables
formalization. may appear and object methods may be
This section first presents the PNO formalism invoked. These actions are executed at each
and gives the reasons of the choice of this language occurrence of the transition and they process
for workflow process versions. Then, this section the values of tokens. Finally, a transition may
explains the mapping from a VBP-Tree onto a PNO. include a set of emission rules i.e. side-effect
free Boolean expressions that determine the
5.1 Petri-net with Objects output arcs that are actually activated after the
execution of the action.
5.1.1 What are PNO?
5.1.2 Motivations for using PNO
Petri Nets with Objects (PNOs) (Sibertin, 1985) are
a formalism combining coherently Petri nets (PN) Petri nets are widely used for workflow specification
technology and the Object-Oriented (OO) approach. (Aalst, 1998). Several good reasons justify their use:
While PN are very suitable to express the dynamic ƒ An appropriate expressive power that permits
behavior of a system, the OO approach permits the the description of the different tasks involved
modeling and the structuring of its active (actor) and in a workflow process and their coordination;
passive (information) entities. In a conventional PN, ƒ A graphical representation that eases the
tokens are atomic, whereas they are objects in a workflow process specification;
PNO. As any PN, a PNO is made up of places, arcs ƒ An operational semantics making an easy
and transitions, but in a PNO, they are labeled with mapping from specification to implementation
inscriptions referring to the handled objects. More possible;
precisely, a PNO features the following additional ƒ Theoretical foundations enabling analysis and
characteristics: validation of behavioral properties and
ƒ Places are typed. The type of a place is a (list simulation facilities.
of) type of an (list of) object(s). A token is a Unfortunately, conventional Petri nets focus on
value matching the type of a place such as a the process definition and do not perfectly capture
(list of) constant (e.g. 2 or ‘hello’), an in- the organizational and the informational dimensions
stance of an object class, or a reference of business processes. As mentioned before, PNO
towards such an instance. The value of a place extend Petri nets by integrating high-level data
is a set of tokens it contains; structures represented as objects, and, therefore
ƒ Arcs are labeled with parameters. Each arc is provide the possibility to integrate in a coherent way
labeled with a (list of) variable(s) of the same the two missing dimensions. Thus, using PNO,
type, as the place the arc is connected to. The actors/roles of the organizational model are directly
variables on the arcs surrounding a transition represented as objects and they may be invoked
serve as formal parameters of that transition through methods in the action part of a transition. In
and define the flow of tokens from input to the same way, data and documents (from the
output places. Arcs from places to a transition informational model) are also represented by objects
determine the possible condition of the flowing in the PNOs and transformed by transitions.
transition: a transition may occur (or is Consequently, we use PNO as a graphical tool to
possible) if there exists a binding of its input visualize versions of business processes, and as a
variables with tokens lying in its input places; formal tool to define executable specifications in
ƒ Each transition is a complex structure made up order to analyze, simulate, check and validate
of three components: a precondition, one (or workflow process versions.
several) action(s) and emission rules. A
transition may be guarded by a precondition, 5.2 From VBP-Trees to PNOs
i.e. a side-effect free Boolean expression
involving input variables. In this case, the Table 4 and figure 9 give the mapping rules in order
transition is only permitted by a binding if this to obtain, from a VBP-Tree, i.e. a VBP-instance, the
binding evaluates the precondition to be true. corresponding Petri net with objects. We distinguish
Passing through a transition depends on the mapping rules for concepts from mapping rules for
precondition, on the location of tokens and control patterns. Table 4 introduces mapping rules
also on their value. A transition also includes for concepts while figure 9 presents mapping rules
one or several actions, which consists of a for control patterns.

275
ICE-B 2008 - International Conference on e-Business

Table 4: Mapping Rules for Concepts. Moreover, this function uses a set of functions
VBP-Tree concepts PNO concepts permitting the handling of a tree:
Name of a Terminal Name of a transition ƒ ListOfChildren(n) returns the children of a node
node n (non-terminal or terminal nodes);
SetofActions of a Actions of a transition ƒ ListOfLeaves(t) returns the terminal nodes
Terminal node (leaves) of a tree t;
PreCondition of a Pre-condition of a
Terminal node transition and also a set of functions for building transitions
PostCondition of a Emission rule of a and their coordination:
Terminal node transition ƒ BuildTransition returns the corresponding
ConsumesInformation Begin place of a transition to a node using mapping rules
of a Terminal node transition defined in table 4;
ProduceInformation of End place of a transition ƒ BuildPattern uses mapping rules defined in
a Terminal node
figure 9 to return the corresponding PNO
PlayedBy of a Begin place of a
according to the control pattern specified in a
Terminal node transition representing a
node;
role
ƒ AddTransition (tr,PNO): add a transition tr to a
Figure 9 below details how modeled control PNO;
patterns are represented using PNOs. In this figure, ƒ AddPattern (pa,PNO): adds a pattern pa to a
P, P1 and P2 correspond to business processes while PNO.
condition corresponds to a condition used in
This function is the following.
conditional control patterns. Finally, Empty is a
transition for which no actions are executed. Function BuildPNO (n:Node):PNO
Local c:Node; tr:Transition
Global t:VBP-Tree
Fork(P1,P2) Sequence(P1,P2)
Begin
If n ∈ ListOfLeaves(t) Then
P1
tr = BuildTransition(n)
P1 P2
BuildPNO =
AddTransition(tr,BuildPNO)
Else
P2 –- n is a non terminal node
Join(P1,P2)
pa = BuildPattern(n)
BuildPNO =
AddPattern(pa, BuildPNO)
P1 P2 If(Condition,P) For Each c ∈ ListOfChildren(n)
BuildPNO = BuildPNO(c)
Next c
condition not condition
AP Empty End If
Empty
End

While(Condition,P) Repeat(Condition,P) 6 CONCLUSIONS

condition not condition
AP As mentioned in the introduction, the problem stated
AP Empty
condition not condition
as “how to support dynamic change of business
process” has already been addressed in the workflow
context. We distinguish two main approaches to deal
Figure 9: Mapping Rules for Control Patterns. with this problem.
Concerning the adaptive-based approach,
We also provide a function for building a PNO relevant works in this area propose solutions to deal
from a VBP-Tree. This function uses mapping rules with workflow schemas changes, adaptation and
presented in table 4 and figure 9 for defining migration of their corresponding instances. (Casatia
transitions of the PNO and their coordination. and al, 1996) presents a workflow modification
language that supports updates of workflow

276
 DEALING WITH BUSINESS
D PROCESS EVOLUTION USING VERSIONS

schemas. It also defines a set of evolution policies the two other dimensions of workflow, that are the
that a workflow administrator can adopt to manage informational and the organizational dimensions.
instances of updated workflow schemas in order to Consequently, this paper revisits the dynamic
migrate (or not) them as instances of the new change of business process issue following a
schema. Three mains policies are defined: abort, version-based approach and considering the
flush and progressive. (Kammer and al, 2000) organizational, informational and process models of
investigates exception handling as a way to support business processes. More precisely, it introduces:
dynamic change to workflow process schemas.
ƒ A meta-model for designing versions of
Consequently, it introduces a taxonomy for
business processes;
exceptions and defines functionalities that Workflow
ƒ A taxonomy of operations for business process
Management Systems must have in order to be able
version management;
to deal with these exceptions. The ADEPTflex
ƒ A formalization and a visualization, using Petri
project (Reichert and Dadam, 1998), (Rinderle and
net with Objects, of versions of business
al, 2004) extensively studies process schema
processes, designed with the previous meta-
evolution. This work formally defines change
model.
operations for both process schemas and workflow
Our solution has the following advantages:
instances as well as related migration policies in
ƒ It permits a comprehensive modeling of
handling potential conflicts. We can also mention
business processes considering the three
van der Aalst’s work to address dynamic change of
dimensions of business processes;
workflow (Aalst, 2001). This work uses a generic
ƒ The VBP meta-model is simple: it only
process model to describe a family of variants of a
integrates core concepts for both business
same workflow process and the notion of inheritance
process modeling and business process
is used to link these different variants. In the same
versioning (our versioning kit is very simple),
vein, (Adams and al, 2006) proposes, for dealing
ƒ Dynamics aspects of business process version
with dynamic evolution in workflows, to use
management are investigated in depth
accepted ideas of how people actually work to
according to the state of the art for versions in
define sets of worklets (i.e. processes) and a stategy
databases;
for runtime selection of a specific worklet.
ƒ It provides rules and algorithms to derive
However, none of these works mention the
modeled versions of business processes onto
notion of workflow versions. Consequently, none of
Petri net with objects specifications.
them enables several different schemas of a same
workflow process to conjointly exist. As future work, we have planed to implement the
Relevant works from the version-based approach VBP meta-model in order to model version of
allow to different instances of a same workflow business processes and to derive versions of business
process to own different schemas. Two main processes specified using BPEL.
contributions are relevant from this approach. First,
(Kradolfer and Geppert, 1999) have proposed to deal
with dynamic workflow evolution, i.e. modification REFERENCES
of workflow process schemas in the presence of
active workflow process instances, introducing Adams, M., ter Hofstede, A., Edmond, D., van der Aalst,
versions of workflow process schemas. This work W., 2006. Worklets: A Service-Oriented
has defined a set of operations for workflow process Implementation of Dynamic Flexibility in Workflows,
schema modification and a strategy for migration of Int. Conference on Cooperative Information Systems,
workflow process instances. Second, and more Montpellier, France, pp. 291–308.
recently, (Zhao and Liu, 2007) have also defended Andonoff, E., Hubert, G., Le Parc, A., Zurfluh, G., 1996.
the advantages of a version-based approach to face Integrating Versions in the OMT Models. Int.
business process evolution. More precisely, this Conference on the Entity Relationship Approach,
Cottbus, Germany, pp. 472–487.
work proposes to model versions of workflow
Casati, F., Ceri, S., Pernici, B., Pozzi, G., 1995.
process schemas using a graph. It also presents a set Conceptual Modelling of Workflows. Int. Conference
of operations enabling to update this graph and on the Entity Relationship Approach, Goald Cost,
defines two strategies to extract versions of Autralia, pp. 341–354.
workflow process schemas from this graph. Casati, F., Ceri, S., Pernici, B., Pozzi, G., 1996. Workflow
However, these two works only consider the Evolution. Int. Conference on the Entity Relationship
process model of workflow. They do not integrate Approach, Cottbus, Germany, pp. 438–455.

277
ICE-B 2008 - International Conference on e-Business

Cellary, W., Jomier, G., 1990. Consistency of Versions in Roddick, J., Craske, N., Richards, T., 1993. A Taxonomy
Object-Oriented Databases. Int. Conference on Very for Schema Versioning based on the Relational and
Large DataBases, Brisbane, Australia, pp. 432–441. Entity Relationship Models. Int. Conf. on the Entity
Chen, I., Markowitz, V., Letovsky, , Li, P., Fasman, K., Relationship Approach, Arlington, Texas, USA, pp.
1996. Version Management for Scientific Databases. 137–148.
Int. Conference On Extended Database Technology, Rosemann, M., zur Muehlen, M., 1998. Evaluation of
Avignon, France, pp. 289–303. Workflow Management Systems: a Meta-model
Chou, H.T., Kim, W., 1986. A Unifying Framework for Approach. Australian Journal of Information Systems,
Version Control in a CAD Environment. Int. 6(1), pp. 103–116.
Conference on Very Large DataBases, Kyoto, Japan, Sciore, E., 1994. Versioning and Configuration
pp. 336–344. Management in Object-Oriented Databases. Int.
Fernandes, S., Cachopo, J., Silva, R., 2004. Version Journal on Very Large Databases, 3(1), pp. 77–106.
Supporting Evolution in Workflow Definition Sibertin-Blanc, C., 1985. High Level Petri Nets with Data
Language. Int. Conference on Current Trends in Structure. Int. Workshop on Petri Nets and
Theory and Practice of Computer Science, Merin, Applications, Espoo, Finland.
Czech Republic, pp. 208–217. Smith, H., Fingar, P., 2003. Business Process Management
Halliday, J., Shrivastava, SK., Wheater, SM., 2001. - The Third Wave. Megan-Kiffer Press.
Flexible Workflow Management in the OPEN-flow van der Aalst, W., 1998. The application of Petri Nets to
System. Int. Conference on Enterprise Distributed Workflow Management. Int. Journal on Circuits,
Object Computing, Seattle, Washington, USA, pp. 82– Systems and Computers, 8(1), pp. 21–66.
92. van der Aalst, W., 1999. Inter-Organizational Workflows:
Heinis. T., Pautasso. C., Alonso G., 2005. The JOpera An Approach Based on Message Sequence Charts and
Autonomic Workflow Engine, the 2nd International Petri Nets. Int. Journal on Systems Analysis, Modeling
Conference on Autonomic Computing (ICAC-05), and Simulation, 34(3), pp. 335–367.
Seattle, Washington, June 2005. van der Aalst, W., 2001. How to Handle Dynamic Change
Kammer, P., Bolcer, G., Taylor, R., Bergman, M., 2000. and Capture Management Information: an Approach
Techniques for supporting Dynamic and Adaptive based on Generic Workflow Models. Int. Journal of
Workflow. Int. Journal on Computer Supported Computer Systems, Science, and Engineering, 16(5),
Cooperative Work, 9(3/4), pp. 269–292. pp. 295–318.
Katz. R., 1990. Towards a Unified Framework for Version van der Aalst, W., Aldred, L., Dumas, M., ter Hofstede,
Modelling in Engineering Databases. Int. Journal on A., 2004. Design and Implementation of the YAWL
Computing Surveys, 22(4), pp. 375–408. System. Int. Conference on Advanced Information
Kimball, J., Larson, A., 1991. Epochs: Configuration Systems Engineering, Riga, Latvia, pp. 142–159.
Schema, and Version Cursors in the KBSA van der Aalst, W., Benatallah, B., Casati, F., Curbera, F.,
Framework CCM Model. Int. Workshop on Software Verberk, E., 2007. Business Process Management:
Configuration Management, Trondheim, Norway, pp. Where Business Processes and Web Services Meet.
33–42. Int. Journal on Data and Knowledge Engineering,
Kradofler, M., Geppert, A., 1999. Dynamic Workflow 61(1), pp. 1–5.
Schema Evolution based on Workflow Type van der Aalst, W., ter Hofstede, A., Kiepuszewski, B.,
Versioning and Workflow Migration, Int. Conference Barros, A., 2003-b. Workflow Patterns. Int Journal on
on Cooperative Information Systems, Edinburgh, Distributed and Parallel Databases, 14(1), pp. 5–51.
Scotland, pp. 104–114. van der Aalst, W., Ter Hofstede, A., Weske, M., 2003-a.
Leymann, F., Roller, D., 1999. Production Workflow: Business Process Management: A Survey. Int.
Concepts and Techniques. Prentice-Hall Press. Conference on Business Process Management,
Manolescu, DA., 2001. Micro-Workflow: A Workflow Eindhoven, The Nederlands, pp. 1–12.
Architecture Supporting Compositional Object- Vossen, G., Weske, M., 1999. The WASA2 Object-
Oriented Development. PhD Thesis, University of Oriented Workflow Management System. Int.
Illinois. Conference on Management of Data, Philadelphia,
Mohan, C., Alonso, G., Gunthor R., Kamath, M., 1995. Pennsylvania, USA, pp. 587–589.
Exotica: A Research Perspective on Workflow Zhao, X., Liu, C., 2007. Version Management in the
Management Systems. IEEE Data Engineering Business Change Context. Int. Conf. Business Process
Bulletin, 18(1), pp. 19–26. Management, Brisbane, Australia, pp. 198–213.
Reichert, M., Dadam, P., 1998. ADEPTflex: Supporting
Dynamic Changes of Workflow without Loosing
Control. Int. Journal on Intelligent Information
Systems, 10(2), pp. 93–129.
Rinderle, S., Reichert, M., Dadam, P., 2004. Disjoint and
Overlapping Process Changes: Challenges, Solutions
and Applications. Int. Conference on Cooperative
Information Systems, Agia Napa, Cyprus, pp.101–120.

278
 A THEORY-DRIVEN FRAMEWORK FOR CONSUMERS TO
ADOPT M-COMMERCE DEVICES

Vincent Cho, Humphry Hung and Y. H. Wong

Department of Management and Marketing, The Hong Kong Polytechnic University, Hung Hom, Hong Kong
[email protected], [email protected], [email protected]

Keywords: Mobile commerce.

Abstract: M-commerce (mobile commerce) is the buying and selling of goods and services through wireless handheld
electronic devices such as portable personal computers, mobile phones and personal digital assistants. This
paper proposes a theory-driven examination of the adoption of M-commerce devices (MCD) by consumers
in their on-line purchase processes. By integrating the concepts of the options model with the major ideas of
the technology acceptance model (TAM), we identify four M’s (merits, maturity, maneuverability and
mentality) as the influencing factors of the adoption of MCD. Based on the generic attributes of m-
commerce, we further identify two M’s, matching and mobility, as the antecedents of these influencing
factors. We then propose a conceptual model of the adoption of MCD by consumers. Because of the
ubiquitous nature of m-commerce, the proposed framework would have universal implications and would
make significant contributions to a more in-depth understanding of the spread and acceptability of m-
commerce.

1 INTRODUCTION wide speculation that m-commerce will surpass PC-

based e-commerce as the preferred method of choice
It is estimated that by 2007, the total number of the for digital commerce transactions. The industries
Internet users in the world will be over 1.4 billion and services affected by m-commerce will include:
and the percentage of wireless users is projected to 1) information services, which include the delivery
take up about 57% of the vast number (Magura, of financial news, sports, horse racing and traffic
2003). Most people anticipate that the next- updates to users; 2) financial services, which
generation commerce will emerge from traditional includes e-banking (when customers use their
commerce, to PC-based e-commerce, and eventually handheld devices to access their accounts and pay
to mobile commerce (Chircu & Kauffman, 2000; their bills) as well as brokerage and investment
Ellis-Chadwick et al., 2000, Miller, 2002). M- services, in which stock quotes can be displayed and
commerce (mobile commerce) is the buying and trading conducted from mobile devices; 3)
selling of goods and services through wireless telecommunications, in which service charges, bill
handheld devices such as mobile phone and personal payment and account reviews can all be conducted
digital assistants. It is an extension, rather than a from mobile devices; and 4) retail consumers are
complete replacement, of PC-based e-commerce given the ability and opportunity to place orders and
(electronic commerce) and allows users to interact pay through mobile devices.
with other users or businesses in a wireless mode, The emerging technologies behind m-commerce
anytime and anywhere (Balasubramanian et al, is based on the Wireless Application Protocol
2002; Samuelsson & Dholakia, 2003). It is very (WAP) and high speed wireless network such as 3G,
likely that PC-based e-commerce will still prevail 3.5G and 4G (Cowles, Kiecker, & Little, 2002;
for a relatively long period of time in spite of the Watson et al., 2002). These technologies enable
trend that more and more people will choose to users to download video/audio information
adopt m-commerce for their purchases (Miller, seamlessly. The focus of this research framework is
2002). on the consumers’ adoption of m-commerce devices
As content delivery over mobile devices (MCD), which are equipment and technologies that
becomes faster, more secure, and efficient, there is facilitate users to make use of m-commerce. MCD

279
ICE-B 2008 - International Conference on e-Business

include mobile phones, Personal Digital Assistants 2 M-COMMERCE DEVICE

(PDA), portable computer notebooks, Bluetooth, ACCEPTANCE
WAP and other facilities that can have access to the
wireless networks. Because of the need of the
The emergence of m-commerce requires relevant
standardization of the application, interface and
new technologies and attracts some current studies
inter-connectivity of all hardware and software, it is
on its adoption (Xu and Gutierrez, 2006; Dholakia
relevant to the adoption and usage of MCD
and Dholakia 2004; Bruner and Kumar, 2005;
(Dholakia and Rash 2004; Buellingen and Woerter
Okazaki, 2005; Harris et al., 2005). Empirical
2004). We expect that the heading towards a world
observation suggests that there is typically a
of mobile networks and wireless devices, which will
substantial lag between the discovery of a new
present a new perspective of time and space, is
technology and its adoption (Doraszelski, 2004).
definitely on its way.
The theories on the diffusion of innovation can be
traced back to Everett Rogers (1962). Since then,
1.1 Objectives many authors have worked further on this theory,
but the core remained the same: when a new
Several basic questions about MCD will be product/new technology is introduced, the target
addressed in this study. First, why should consumers market can be divided into five segments along an
adopt MCD? What are the influencing factors for axis of risk aversion: in the beginning there are the
consideration? Second, how do the MCD compare innovators, followed by the early adopters, the early
with the devices for other types of commerce such as majority, the late majority and the risk-allergic
e-commerce? Consumers will only adopt MCD laggards. This proposed framework is not on the
when there are some potential significant advantages diffusion process but primarily on the intention to
when comparing to old devices for other types of adopt new m-commerce technologies by early
commerce. There is still a lack of comprehensive adopters.
frameworks within which the adoption of MCD can Literature on the delayed acceptance of
be evaluated. Traditional viewpoints regarding this technology has stressed primarily on the benefits and
issue, especially those that are based on technology use of new technologies (Davis et al., 1989),
acceptance models, will need to be revisited and comparison between old and new technologies
revised when consumers are considering such an (Sheasley, 2000), role of sunk costs in existing
adoption. technology (Salter, 1966) or in complementary
In this proposed research, we intend to integrate technologies (Frankel, 1955). There are some
the major ideas of the technology acceptance models models that associate diffusion lags of new
(both TAM and TAM2) and the options model as technologies with the reduction of complementary
our basic framework for studying m-commerce. costs such as specific human capital (Chari and
Very little research has ever been done along this Hopenhayn, 1990), learning-by-doing (Parente,
direction. We can contribute to the literature by 1994; Jovanovic & Lach, 1989), and search costs
exploring and identifying the various options, or (Jovanovic & MacDonald, 1994). Other behavior-
independent variables, that will affect the decision of based models, such as the information cascades
buyers to adopt new technologies related to m- theory on the acceptance of new technology,
commerce. The research framework will be of suggests that an individual who adopts new
interest to marketers in m-commerce and also to technologies may do so based on the actions of
academics in the fields of marketing and IT. Both others and contrary to his or her private preferences
are keen to determine how they can perform further (Bikhchandi et al. 1992).
relevant research and position themselves well in the
next generation of m-commerce. Our proposed
2.1 Why TAM and Options Model are
framework will have both theoretical and practical
implications through knowing why and how relevant used Together
MCD are adopted for m-commerce.
In line with mainstream literature, we acknowledge
that TAM will be our primary research framework.
The technology acceptance model (TAM) is an
information systems theory that models how users
come to accept and use a new technology, with
reference to two major considerations, perceived
usefulness and perceived ease of use (Bagozzi et al.,

280
 A THEORY-DRIVEN FRAMEWORK FOR CONSUMERS TO ADOPT M-COMMERCE DEVICES

Features of M-commerce Devices

Standardization and common interface Free to move

Antecedents Matching Mobility

Options
Model
(Focusing on
comparisons
Technology
Acceptance among old and
Model new
Recognition by Easy to Use Significant Maturity of technologies
(Focusing on the
advantages of Peer groups improvement Technology and potential
new technologies) development)

Independent Mentality Maneuverability Merits Maturity

variables

Dependent
Acceptance of m-commerce devices
Variable
for m-commerce activities

Figure 1: A Conceptual Model of the Acceptance of M-commerce Devices (MCD).

1992; Davis et al., 1989). The former is about the possible upgrades and changes in the future of the
degree to which a person believes that using a new technology (Sheasley, 2000).
particular system will make his or her life easier, e.g. While using technology acceptance models
by enhancing his or her job performance or reducing (TAM) as our primary reference, we also
the workload, while the latter is the degree to which incorporate the important implications of the options
a person believes that it is not difficult to actually model into our basic framework for analyzing
use a particular system (Davis et al., 1989). An consumers’ adoption of MCD for m-commerce. We
extended version of the TAM model, referred to as observe that TAM is primarily about to what extent
TAM2, was later developed to explain perceived people will adopt new technologies with reference to
usefulness and usage intentions in terms of social the advantage and benefits (perceived usefulness and
influence and cognitive instrumental processes perceived ease of use). However, we observe that,
(Venkatesh and Davis, 2000). although it is very likely that new technologies will
The options model demonstrates that a new eventually replace old ones, the devices of old and
technology with a moderate expected improvement new technologies are very often being used at the
in performance can experience substantial delays in same time, based on people’s assessment of the
acceptance and price dropping in a competitive comparative merits of the two generations of
market (Bessen, 1999; Sheasley, 2000). Rather than technologies. We, therefore, consider that the
adopting a new technology that demonstrates only options model (which focuses on the comparison
marginal improvement, consumers have the option between old and new technologies) is a useful tool in
of not adopting until the new technology, in terms of our analysis of the acceptance of MCD for m-
performance and price, is substantially better than commerce.
the old technology. Consumers contemplating the
adoption of a new technology are, of course, aware
of the possibility of sequential improvement. They
consider not only the current technical level of the
new technology, but also their expectations of

281
ICE-B 2008 - International Conference on e-Business

3 THE INTEGRATED m-commerce, we also need to think about some

FRAMEWORK other aspects. With regards to the options model
(Bessen, 1999; Sheasley, 2000), we consider the
comparison between MCD and devices for other
In essence, options model focuses on the comparison
types of commerce, and in particular, the
between existing and old MCD while TAM places
comparative advantages of MCD to consumers.
emphasis on the generic attributes and utility of
Based on the options model, we identify two M’s,
MCD. We integrate the major ideas of these two
merits and maturity, in relation to the comparison.
models into our new proposed model as shown in
We identify the third influencing factor, merits,
Figure 1. Based on our theoretical framework, we
which is about the degree to which a buyer believes
identify four influencing factors: merits, maturity,
that the MCD can provide significant improvement
maneuverability, and mentality, which we consider
in the purchase process. Handheld mobile devices,
to be relevant to the decision of consumers in
such as PDA and other enhanced alphanumeric
adopting MCD. We also identify two generic
communicators, have supplemented mobile
antecedents of these influencing factors, mobility
telephones, thus expanding the range of MCD
and matching. This suggests an extent of influence
available for m-commerce transactions. With the
of these influencing factors and their antecedents to
abilities to be connected to digital communication
affect consumers’ adoption decision of MCD.
networks, MCD are considered to be in possession
of important comparative advantage of mobility.
3.1 Influencing Factors based on Merits can be measured by the comparative
Technology Acceptance Model advantages of the MCD in relation to the old devices
for other types of commerce.
With reference to technology acceptance model The fourth influencing factor, maturity, is the
(TAM, TAM2 and UTAUT) (Venkatesh & Davis, possibility that the technology of the MCD is mature
2000; Venkatesh et al. 2003), we consider whether enough so that there will not be any possible
the adoption of MCD will bring advantages to significant improvements at a later stage. While
consumers. We identify two M’s, maneuverability academic researchers and business practitioners
and mentality, for relating the acceptability of MCD recognize that the electronic market will penetrate
to users. and replace traditional type of commerce, there are
The first influencing factor, maneuverability, is still some reservations that will likely cause the early
related to the perceived usefulness in the adoption of adopters of new technologies some problems in
MCD and the degree to which a person can make the terms of the obsolescence of devices (Samuelsson &
best use of such MCD. Consumers will tend to adopt Dholakia, 2003). Most consumers will prefer
devices that are user-friendly and do not require adopting MCD with more mature technologies so
some intensive training of adoption (Prasanna et al, that there is no need for a high level of subsequent
1994). It would be measured by the usability of the upgrading of devices. Maturity can be assessed by
MCD. the perception that the relevant MCD can or cannot
The second influencing factor, mentality, is be upgraded.
concerned with the match between the new
technology and consumers’ own mindset, as well as 3.3 Generic Attributes of MCD
the appropriate recognition of their peer groups
(Bessen, 1999; Venkatesh & Davis, 2000). General In addition to the identification of the influencing
acceptance by the consumers, especially by their factors of the adoption of MCD, we also consider
peer groups, will be very important to consumers their antecedents, which are related to the very basic
when they consider using MCD for matching the and essential characteristics of MCD. We start our
devices of other people. Mentality can be evaluated analysis by considering two generic attributes of
by the perceived peer groups’ acceptance of MCD. MCD, mobility and matching.
Mobility is the most fundamental aspect of m-
3.2 Influencing Factors based on commerce because the name m-commerce arises
Options Model from the mobile nature of the wireless environment
that supports mobile electronic transactions
While mainstream literature on the adoption of new (Coursaris et al, 2003). Mobile wireless devices,
technologies is primarily based on the technology such as mobile phones, PDA, and portable computer
acceptance model, we consider that, in the context of notebooks, can have the ability to help users gain

282
 A THEORY-DRIVEN FRAMEWORK FOR CONSUMERS TO ADOPT M-COMMERCE DEVICES

access to the Internet. Based on these wireless by consumers in their purchase processes and will be
devices, m-commerce is a natural extension of e- of interest to the MCD market.
commerce but can provide some additional
advantages of mobility for consumers. Mobility is a
major prerequisite for the adoption of MCD. It is an ACKNOWLEDGEMENTS
antecedent of the influencing factors of the adoption
of MCD because people will consider adopting
This research was supported in part by The Hong
wireless connection because it can allow significant
Kong Polytechnic University under grant number A-
improvement (i.e. merits), easy to use (i.e.
PA6E.
maneuverability), and can be accepted by peer
groups (i.e. mentality). It can be measured by the
extent of access to wireless networks.
Matching describes the need for the standardized REFERENCES
and common interface of MCD (Coursaris et al,
2003). The unique characteristic of m-commerce Balasubramanian, S., Peterson, R.A, Jarvenpaa, S.L.,
very often requires both ends of this new type of 2002. Exploring the implications of M-commerce for
markets and marketing. Academy of Marketing
commerce to have a common interface. M-
Science Journal, 30, 4, 348-361.
commerce applications have the challenging task of Barnes, S. J., Scornavacca, E., 2004. Mobile Marketing:
discovering services in a dynamically changing The Role of Permission and Acceptance. International
environment. Effective mechanisms need to be in Journal of Mobile Communication. 2, 2, 128-139.
place for the interface between various types of Bessen, J., 1999. Real Options and the Adoption of New
MCD. Matching is an important antecedent of the Technologies, On-line working paper at
influencing factors of consumers’ adoption of MCD http://www.researchoninnovation.org/online.htm#real
because the need for standardization (i.e. matching) opt
is important for m-commerce technology which Bigne E., Ruiz C., Sanz S., 2005. The impact of Internet
user shopping patterns and demographics on consumer
allows for the interface of MCD with the wireless
mobile buying behaviour. Journal of Electronic
networks when the technology and interface is Commerce Research, 6, 3, 193 – 210.
mature (i.e. maturity). It also provides utility for Bisbal, J., Lawless, D., Wu, B., Grimson, J., 1999. Legacy
consumers for interacting with other devices (i.e. Information System Migration: A Brief Review of
merits). Matching can be measured by the degree Problems, Solutions and Research Issues. IEEE
that MCD can be compatible with each other. Software, 16, 103-111
Based on our conceptual framework, we identify Bruner, G.C., Kumar, A., 2005. Explaining consumer
the various influencing factors (i.e. 4 M’s) which acceptance of handheld Internet devices. Journal of
can affect consumers’ decision of the adoption of Business Research. 58, 5, 553 - 558
Buellingen, F., Woerter, M., 2004. Development
MCD in their purchases. It is possible to collect data
perspectives, firm strategies and applications in mobile
on whether consumers will consider the adoption of commerce. Journal of Business Research. 57, 12,
MCD, and at the same time, researchers can also 1402-1408.
investigate the reasons why they adopt or do not Chircu, A., Kauffman R., 2000. Reintermediation
adopt MCD, in terms of timing, opportunities, Strategies in Business-to-Business Electronic
changing trends and applications. Commerce, International Journal of Electronic
Commerce. 4, 4, 7-42.
Coursaris, C., Hassanein, K., Head, M., 2003. M-
commerce in Canada: An interaction framework for
4 CONCLUSIONS wireless privacy. Canadian Journal of Administrative
Sciences. 20, 1, 54-73.
We are proposing new insights and new adoption Cowles, D.L., Kiecker, P., Little, M.W., 2002, Using key
behavior in the ubiquitous world of m-commerce, informant insights as a foundation for e-retailing
which we believe, are still not yet fully understood theory development. Journal of Business Research.
by most marketers and scholars (Stevens & 55, 629-636.
Dholakia, R.R., Dholakia, N., 2004. Mobility and markets:
McElhill, 2000; Struss et al., 2003). Our conceptual
emerging outlines of m-commerce. Journal of
framework contributes to literature by suggesting the Business Research. 57, 12, 1391 – 1396.
new constructs: merits, maturity, maneuverability, Dholakia, N., Rash, M., 2004. Configuration m-commerce
and mentality, which we consider to be relevant to portals for business success. In N. Shi (Ed.), Mobile
the decision of consumers in adopting MCD. It also commerce applications (76-93). Hershey, PA: Idea
represents an examination of the adoption of MCD Group Publishing.

283
ICE-B 2008 - International Conference on e-Business

Ellis-Chadwick, F., McHardy, P., Wiesnhofer, H., 2000. Venkatesh, V., Davis, F.D., 2000. A theoretical extension
Online Customer Relationships in the European of the technology acceptance model: Four longitudinal
Financial Services Sector: A Cross – Country field studies. Management Science. 46, 2, 186-204.
Investigation. Journal of Financial Services Watson, R.T., Pitt, L.F., Berthon, P., Zinkhan, G.M. 2002.
Marketing. 6, 4, 333-345. M-commerce: Extending the Universe of Marketing.
Frolick, M.N., Chen, L., 2004. Assessing m-commerce Journal of the Academy of Marketing Science. 30, 4,
opportunities, Information Systems Management, 21, 329-343.
2, 53 – 62. Wen, H.J., Mahatanankoon, P., 2004. M-commerce
Harris, P., Rettie, R., Kwan, C.C., 2005. Adoption and operation modes and applications. International
usage of m-commerce: a cross-cultural comparison of Journal of Electronic Business. 2, 3, 301 – 315.
Hong Kong and the United Kingdom. Journal of Xu, G., Jairo, A., Gutiérrez 2006. An Exploratory Study of
Electronic Commerce Research. 6, 3, 210 – 225. Killer Applications and Critical Success Factors in M-
Hung, H., Cho, V., 2006. A Conceptual model of the Commerce. Journal of Electronic Commerce in
influencing factors for consideration when consumers Organizations. 4, 3, 63-79.
adopt M-commerce devices. Encyclopedia of Mobile
Computing & Commerce. working paper accepted for
publication.
Kim, J., Lee, I., Lee, Y., Choi, B., 2004. Exploring E-
business Implications of the Mobile Internet: A Cross-
national Survey of Hong Kong, Japan and Korea.
International Journal of Mobile Communication. 2, 1,
1-21
Liao, S.S., He, J.W., Tang, T.H., 2004. A framework for
context information management, Journal of
Information Science, 30, 6, 528 – 541.
Lin, H.H., Wang, Y.S., 2006. An examination of the
determinants of customer in mobile commerce
contexts, Information and Management, 43, 3, 271-
282.
Magura, B., 2003. What Hooks M-commerce Customers?
MIT Sloan Management Review, 44, 3, 9-10.
Miller, A.I., 2002. Einstein, Picasso: Space, Time, and the
Beauty That Causes Havoc. New York, Basic Books.
Mylonakis, J., 2004. Can Mobile Services Facilitate
Commerce? Findings from the Greek
Telecommunications Market, International Journal of
Mobile Communication. 2, 2, 188-198.
Okazaki, S., 2005. New Perspectives on m-commerce
research. Journal of Electronic Commerce Research.
6, 3, 160 – 165.
Peffers, K., Tuunanen, T., 2005. Planning for IS
applications: A practical, information theoretical
method and case study in mobile financial services.
Information & Management, 42, 3, 483-501.
Samuelsson, M., Dholakia, N., 2003. Assessing the market
potential of network-enabled 3G M-Business Services.
In: Nansi, S., (Ed), Wireless communications and
mobile commerce. Ideal Group Publishing, Singapore
& Hershey, PA.
Sheasley, W.D., 2000. Taking an Options Approach to
New Technology Development. Research Technology
Management. 43, 6, 37-43.
Shugan, S.M., 2004. The impact of advancing technology
on marketing and academic research. Marketing
Science. 23, 4, 469 – 476.
Stevens, G.R., McElhill, F., 2000. A qualitative study and
model of the use of e-mail in organizations. Electronic
Networking Applications and Policy. 10, 4, 271-283.
Struss, J., El-Ansary, A., Frost, R., 2003. E-Marketing. 3rd
edition, Prentice Hall, New Jersey.

284
 IT APPLICATIONS IN PRODUCTION PLANNING AND
CONTROL
A Survey of Medium Sized Business in German-speaking Europe

Jakob Lewandowski, Matthias Buhl and Burkhard Kittl

Institute for Production Engineering, Vienna University of Technology, Karlsplatz 13, 1040 Vienna, Austria
[email protected], [email protected], [email protected]

Keywords: Enterprise Resource Planning (ERP), IT survey, Manufacturing Execution System (MES), Medium-sized
business, Production Planning and Control.

Abstract: This paper addresses the use of IT-applications for production planning and control of medium-sized
businesses in German-speaking Europe. The article summarises challenges today and expectations for the
future. In particular, the kind of IT systems in use, experiences and satisfaction with IT support were key
questions of the study. Basis of the research was a telephone survey with 150 IT managers of representative
medium-sized enterprises in production, trade and service sectors.

1 INTRODUCTION business-software companies in the last years have

only brought minor improvement. Getting a better
Production and therewith combined planning and insight into the use of and the opinions about IT
controlling of resources are the core of every solutions in the industry was the impulse to start a
company in the manufacturing industry. Although survey to collect useful information for further
most of the factories enforce development and research projects.
adaptation of the technological processes, the
capabilities of IT applications for organizing and
handling the production process are still 2 APPROACH
underestimated. Many troubles are arising out of
linguistic barriers between suppliers/consultants and The study consists of face-to-face and telephone
users. There is no obligatory nomenclature and the interviews.
latest endeavours for standardisation are rarely used The subject of production planning and control is
or even known by many users. This fact leads to quite complex, so intensive preparation for the face-
misconceived coherences and in the worst case to to-face interviews was indispensable. Almost half of
prejudices regarding the qualities of IT systems. the project time was spend on literature studies as
Moreover, medium-sized enterprises don’t have well as developing and evaluating the questionnaire.
sufficient IT resources compared to bigger In the qualitative face-to-face interviews with
companies. For this segment the cost-benefit ratio of twenty-two representative enterprises, the authors
automating workflows in the production planning wanted to explore how comfortable the interview
and control level often is not in evidence. partners feel concerning terms and definitions in this
Furthermore the mid-sized business is confronted topic. In order to get a proper data base additional
with an unclear diversity of software applications for telephone interviews with 150 IT managers were
the production area. carried out. The point of interest was defined as
The topic of IT applications in production systems in “Business Planning and Logistics” (layer
planning and control has kept many developers and 4) and the “Manufacturing Operations and Control”
scientists busy for decades but it is more than ever (layer 3) according to the ANSI/ISA-95 standard
relevant, because there is still no satisfying solution (Instrument Society of America, 2000).
for bridging the divide between the shop floor and
management. Even the great efforts of leading

285
ICE-B 2008 - International Conference on e-Business

2.1 Objectives moment? Furthermore how satisfied are the

companies with their applications? Positive and
The basic objectives of the survey carried out in negative experiences and expectations essentially
German-speaking regions were: depend on the idea the enterprises have of concepts
like enterprise resource planning or manufacturing
ƒ Finding out the problems enterprises have in
execution systems. For this reason it was important
the area of production planning and control.
to find out how conclusive these concepts are for
ƒ Identifying and locating the use of IT
companies. Another critical issue was to ascertain
solutions.
the challenges entrepreneurs are facing in the
ƒ Pin-pointing the prejudices and expectations
production area. Above all, how far should processes
concerning IT systems
be automated respectively is it really desirable to
ƒ Assessing the experiences and best practices
have every process supported by an IT system in
with enterprise resource planning (ERP) and
order to stay on a competitive basis?
manufacturing execution systems (MES).

2.2 Target Group

While the European German-speaking market is
dominated by small to medium sized enterprises, the
businesses size chosen for the survey ranged from
50 to 500 employees. The majority of the companies
belonged to the continuous or discrete producing
industry, but the trade and service sector was
allowed also.

2.3 Survey Model

As a basis for the survey a theoretical framework
was used. It combined the tasks and functions of the
“Aachener model for production and control”
(Schuh, 2006) with the structure of the ANSI-ISA95
standard (Instrument Society of America, 2000). In
addition, a cross-check with further American and Figure 1: Functional hierarchy (Instrument Society of
European literature like MESA (MESA, 1997) or the America, 2000).
VDI-report (VDI, 2006) combined the particular
advantages of each model. The functional hierarchy 3.1 Business Planning and Logistics
of the ANSI/ISA-95 standard is exposed in Figure 1.
The survey model consisted of three main parts. The first part of the results is primarily addressing
The first one gave a hierarchical structured overview Layer 4 of the ANSI/ISA-95 standard (“Business
of the functions in the production and control of the Planning and Logistics”). This is usually the domain
manufacturing industry. In the second part the of enterprise resource planning systems. To get a
functions of order processing were drawn in a task general idea of what kind of applications are in use,
related flow-chart. The third part is a checklist Table 1 reveals the size of enterprises as well as the
referencing functions and processes for the face-to- type of ERP-system. The categories are standard
face interviews in order to allow free conversation software offered by global vendors, as well as
and obtain structured and comparable information at standard software sold by regional vendors and
the same time. individual software. The most remarkable point is
that more than 50% of companies still have
individual ERP-software in use, either as a stand-
3 RESULTS alone system or as an additional solution.
Concerning the age of the IT systems an
important finding was that more than 20% of the
The analysis of the collected data occurred on the
analysed ERP-systems were older than 10 years.
basis of several key questions. First of all, what kind
Due to this fact and the fact that the average age of
IT systems do medium-sized enterprises use at the
the ERP-systems in use is approximately eight years

286
 IT APPLICATIONS IN PRODUCTION PLANNING AND CONTROL - A Survey of Medium Sized Business in
German-speaking Europe

Figure 2: Companies where business processes run parallel to ERP-System (n=150, telephonic interviews).

the issue of individual software becomes more companies. Rare domains were production planning
coherent. and quality management. The reason for this is the
diversity of the manufacturing processes, even
Table 1: ERP-systems in use (n=150, telephone within an industry sector.
interviews). A conspicuous connection was detected between
Employees the type of ERP-system and the management of
<100 101-250 251-500 Total business processes. There were considerably less IT
Standard software global vendors supported business processes running parallel to the
In use 60,5% 61,8% 87,5% 68,0% ERP-system in companies using software by global
Rollout 2,6% 2,9% 0,0% 2,0% vendors. It seems obvious that ERP-systems of
Planned 1,4% 0,0% 0,0% 0,7% leading vendors are employed more efficiently.
Nonexistent 35,5% 35,3% 12,5% 29,3% Table 2 contains a summary of the criterions
Standard software regional vendors leading to the decision for a certain ERP-system.
In use 30,3% 35,3% 20,0% 28,6% The gist is that companies belonging to a concern or
Nonexistent 69,7% 64,7% 80,0% 71,3% group usually have no bearing on decisions because
Individual software most of these are predetermined. The second and
In use 63,2% 52,9% 42,5% 55,3% third most relevant motives were flexibility and
Rollout 5,3 5,9% 0,0% 4,0% sectoral advantages. The fact that both consultancy
Nonexistent 31,5 41,2% 57,5% 40,7% and license fees are rated below three percent
indicates that these costs are ranked as a non-
A moderate enhancement was determined relating to determining factor regarding the total cost of
the rollout time of ERP-systems. Compared with ownership.
studies in the last decade, the time period decreased. The information relating to how comfortable the
Concerning the fields of application like material medium-sized enterprises are with their IT systems
management or production planning, there was no can be regarded as inconspicuous, but even so
significant difference between standard and significant. The data content of Table 3 gives an
individual software. Material management, overview of the three groups of ERP-systems.
procurement, sales as well as controlling and There was no difference in the degree of
accounting were implemented by most of the functional coverage as well as satisfaction between

287
ICE-B 2008 - International Conference on e-Business

companies with standard or individual software in awareness of corporate responsibility for a

use. The user driven margin that individual systems successful IT implementation is obvious, the
naturally should have is not recognizable. capability and benefits of ERP-systems are
established. Basically there was a high degree of
Table 2: Criterions for selecting an ERP-System (n=150, trust in ERP-systems.
telephone interviews).
Table 3: Functional coverage of and satisfaction with
Employees
ERP-systems (n=150, telephone interviews).
<100 101- 251- Total
250 500 Standard Standard Individual
Concern- 23,7% 32,4% 37,5% 29,3% software / software / software
directive global sectoral
No 6,6% 0,0% 7,5% 5,3% vendors vendors
alternatives Degree of functional coverage
to specific Procurement 91,3% 90,9% 91,0%
ERP / Material
Unknown 4,0% 5,98% 2,5% 4,0% Management
reasons Sales/ 91,3% 90,4% 88,5%
Consultancy 2,6% 0,0% 2,5% 2,0% Distribution/
fees Service
Size of 0,00% 5,9% 12,5% 4,7% Production-/ 90,3% 92,0% 92,0%
vendor Project
Global 10,5% 5,9% 7,5% 8,7% planning
application Controlling 93,5% 90,9% 91,8%
of the Quality 91,6% 91,6% 87,0%
specific ERP Management
System 5,3% 5,9% 17,5% 8,7% Human 92,7% 85,1% 90,4%
security Resources
Vendor 1,3% 2,9% 0,00% 1,3% Satisfaction (1=very satisfying, 5=unsatisfied)
support Procurement 1,8 1,5 1,8
License fees 2,6% 0,0% 2,5% 2,0% / Material
Sectoral 19,7% 32,3% 12,5% 20,7% Management
advantage Sales/ 1,8 1,6 1,9
Flexibility 13,1% 8,8% 20,0% 14,0% Distribution/
N/a 18,4% 11,8% 0,0% 12,0% Service
Production-/ 1,7 1,8 1,9
Changing attitudes to IT support and the Project
adaptation of business processes were detected. In planning
the medium-sized businesses more than 27% of the Controlling 17 1,7 1,7
companies using software by global vendors are Quality 1,8 1,7 1,9
outsourcing their IT support. Worthy of mention is Management
that the smaller the amount of employees of a Human 1,8 1,7 1,8
company is the more likely IT support is outsourced. Resources
Asked for satisfaction with external IT support in
This confidence was reflected in the fact that
general the average evaluation number was 1,82
calculating the return on investment very often is no
(1=excellent, 5=insufficient).
issue to companies. They attach great importance to
The way medium-sized enterprises saw
the factors, which are impossible to measure (e.g.
expectations and requirements in the near future
transparency, communication, etc).
often depended on the management. In general,
The expectations in the ERP-system itself rise
companies agreed to the statement: “The more
after an adaptation phase, so long-term support is
business’ processes are automated the bigger the
very important for the companies as well as the
benefit for the company”. Furthermore, companies
vendors. In most cases the area of material
are poised to align business processes to save costs
management indicates improvement very quickly.
for customizing the IT system. These two findings
Users are noticing the limits of the ERP-systems in
were emanating from the telephone interviews as
the area of manufacturing. Reported reasons are the
well as from the face-to-face-interviews. The

288
 IT APPLICATIONS IN PRODUCTION PLANNING AND CONTROL - A Survey of Medium Sized Business in
German-speaking Europe

Figure 3: Requirements for IT software in the production area (n=22, face-to-face interviews).

inflexible structure of the ERP-systems, the 3.2 Manufacturing Operations and

unsatisfying reporting capabilities and unsolved Control
problems with machine and control interfaces. In the
face-to-face interviews it was found out that IT Compared to the layer of business planning and
trainings often are not mandatory in order to reduce logistics, the IT systems used for the shop floor
costs during the implementing phase. At least some show a lower level of integration respectively are
of the companies came to the conclusion that these not existent. Whereas almost each business (97%)
retrenchments have negative effects on the system uses ERP-software and the majority even works with
operation. an integrated standard software tool (more than
Finally the most important, because most often 80%), the situation of MES-software presents itself
mentioned demands on IT software from the survey absolutely different.
are summarised in Figure . The first four positions in The number of enterprises, which have a totally
the bar chart are demands for one integrated system, integrated IT System is limited to approximately
simple usability, central data management and 10%. Most of the questioned companies apply IT
flexible interfaces. Additional needs were assured systems just for certain activities in their shop floor.
vendor support, as well as monitoring and displaying They use a combination of standard software tools
the production process. like quality- or scheduling-software (approx. 40%),
In many companies management demands more individually developed software (30%) and MS
key performance indicators, this naturally militates Office products (30%). About 20% of the enterprises
in favour of integrated IT system. As a closing run the production area even without any IT system.
insight, companies know that there is still potential Remarkable is the fact that IT systems in use are
within the ERP-sector, thus it is not remarkable that quite old. More than half of the companies quote to
the IT budgets stay equal or will increase within the work with software which is older than ten years.
next 5 years. Moreover there has not been any rollout of an IT
system at the questioned companies in layer 3 since
2005, and the satisfaction with the existing IT
applications is quite positive (with an average of 1,4
using 1 as very satisfying and 5 as unsatisfying). The
companies polled in the face-to-face interviews

289
ICE-B 2008 - International Conference on e-Business

didn’t think it’s necessary to exchange their current Although most of the enterprises don’t have an
IT systems in the shop floor. Some of them had IT system for detailed scheduling, it is seen as a
already started IT projects in the shop floor, which critical business process for the ability to respond to
were cancelled later due to the confusing situation of the market.
available IT solutions and vendors on the market or Another topic in the shop floor was traceability.
not evident cost-benefit ratio. Many companies are confronted with this subject
A few enterprises try to fill the gap between the due to legal obligations or product claims. In many
layer of business planning and logistics and the cases the functionalities of ERP-solutions in this
manufacturing operations and control by extending regard are not satisfying and systems without IT
their existing ERP-system. But the majority thinks support are in use.
that the ERP-system is not flexible enough to cope A lot of enterprises use specialised IT systems
with their requirements in the shop floor, which are: for particular functionalities in their production. The
interfaces between these systems are a main, but
ƒ Clustering and analysing existent data.
unsolved problem. Attempts to solve this problem
ƒ Decentralising know-how
with Manufacturing Execution Systems were
ƒ The need for traceability.
stopped, because the offered systems were not able
ƒ The reduction of partial solutions and
to integrate the existing IT applications or
interfaces.
programmable controllers satisfyingly.
ƒ Increased transparency.
Many enterprises were aware of the fact that the
The general requirements for IT software in the implementation of new systems has to go together
production area are shown in Figure . The main with reengineering the processes.
points deal with system integration and data Further chances of development were assigned to
management. maintenance. This MES function turned out to be an
A lot of enterprises were not sure about the cost- exception, especially in the face-to-face interviews.
benefit of an integrated MES solution, because there More than 50% of end-users claimed that
are only a few running reference systems. This could maintenance is a very important element but only a
be the reason, why the integration of standard MES percentage under 10% declared to use a standard-
software is a low priority subject in many software for planning and controlling the
companies. maintenance processes at all.
The possibility of expanding the ERP-software Compared to a study of Trovarit (Trovarit, 2004)
was more often considered by companies which are the awareness level of MES has not changed
not so familiar with MES yet. Furthermore the significantly. More than half of the companies have
number of employees is an indicator whether an not heard about MES yet. In particular enterprises
enterprise is thinking about this alternative. The with less than 350 employees usually don’t know
smaller a business is, the more it tries to avoid a this term. Furthermore there is a relation between the
second integrated system next to ERP. Usually the date of rollout of the ERP system and the know-how
manpower of mid-sized businesses for of MES. Only companies, who had applied their
implementation and service of a second IT system is ERP system more than 10 years ago, have dealt
insufficient. On the other hand there were a intensely with the topic, or even used an MES
significant number of companies, who didn’t believe software.
that an extension of an integrated ERP system into
the shop floor is possible. They pointed out that an
ERP system is not able to cope with the 4 CONCLUSIONS
requirements of the shop floor. In particular
flexibility and planning intervals were mentioned. In general the potential for further development in
Gathering and clustering data from the shop floor the field of production planning and -control is
is a main topic in almost every enterprise. Most of estimated very high. Almost each of the questioned
the data is transferred in the direction from layer companies quoted to struggle with one or more
three to four. The opposite direction regarding order functions of this area. Up to the rough planning the
data is often realized by manual instructions or enterprise resource planning systems got well
documents. Only few companies, who have already established within the last decade. It emerged that
dealt intensely with the subject of MES, demand a when moving from planning to the control of the
bidirectional information flow. production process that the bigger part of the
medium-sized enterprisers places emphasis on

290
 IT APPLICATIONS IN PRODUCTION PLANNING AND CONTROL - A Survey of Medium Sized Business in
German-speaking Europe

individual software or “no software” solutions. concerning interfaces. These systems have to
Although there are many old systems in use the provide more and better connectivity in order to
companies are highly satisfied and prefer solutions allow easy integration of satellite systems. All in all,
decoupled from the ERP-system. In particular the companies are quite happy with their ERP-systems.
segments quality assurance and staff work time In the sector of MES the request for one fully
logging are concerned. integrated IT system is more likely to meet the idea
Based on the findings of this study, one may of an integration framework. Moreover enterprises
generally conclude that for the majority of the are reluctant to substitute successful partial software
medium-sized companies in the production industry solutions by a totally integrated MES. Together with
the term MES is largely unknown. Even if some the problem of low IT resources the step of
enterprises are familiar with the definition of MES exchanging the whole software landscape in the
they are not able to distinguish the functional shop floor seems to be too risky for medium sized
environments of a manufacturing execution system. companies. But the integration of capsuled MES
The most criticised items headed by this group are functions into a framework with the opportunity of
the complexity of the MES-market, as well as the step by step extension is a worthwhile alternative.
unfeasible commitments from software vendors. Another benefit of an integration framework is that
Also remarkable for the authors was that no MES functionalities like maintenance can easily use
enterprise polled in the face-to-face interviews had existing functionality of the ERP system.
implemented a new software-system within the last Information about the production process or
three years. The statement of uncertainty relating to machine parameters are usually collected by several
new software was considerable. Companies doubt systems in the shop floor and the ERP. The job of
that implementation of new MES will bring more the integrating framework is to allocate and analyse
benefits. The case that positive references are the information in order to provide it to a planning
missing has to be mentioned additionally at this tool.
point.
Production enterprises, having long lasting
experience with ERP-system are more willing to 5 FUTURE RESEARCH
evaluate functions in the area of MES. Their know-
how in implementing and maintaining this standard- In a test facility different MES scenarios will be
software system is instrumental in realising new IT simulated. In one scenario it is planned to cover the
systems. functionalities of an MES by using an off-the-shelf
In general there is an unexplainable gap between MES product. In another scenario we are going to
the demands and the willingness to invest in the use an IT framework such as xMII for the
shop floor. In other words enterprises agreed to the integration of dedicated software tools like
advantages of an automated workflow but almost scheduler, shop floor data acquisition or tool
none of the polled companies has an automated management.
workflow established or is willing to provide the
required resources.
Users that exhaust the functional limits of ERP-
system in the production area are increasingly REFERENCES
claiming for more flexibility and more transparency,
attributes very often expected from MES software. Instrument Society of America, 2000. ANSI/ISA–
In the first instance the fixed framework and the not 95.00.01–2000, Enterprise-Control System
Integration, Part 1: Models and Terminology, North
available real-time data are criticised, a standardised
Carolina.
bi-directional interface between ERP and control MESA International, 1997. MESA White Paper #06: MES
layer does not exist. A high potential is attributed to Explained: A High Level Vision, Pittsburgh.
this unsolved section. Schuh, G., 2006. Produktionsplanung und –steuerung,
Looking into the future there is an irrepealable Grundlagen, Gestaltung und Konzepte, Springer,
tendency to integrated systems. The majority of the Berlin Heidelberg, 3rd Edition.
medium-sized businesses prefer one fully integrated Trovarit AG, 2004. MES-Echtzeit in der Fertigung?!,
IT system as personal resources are regularly Aachen.
undersized. This fact offers a chance for ERP- VDI Berichte 5600, 2006. Manufacturing Execution
Systems - Fertigungsmanagementsysteme, Blatt 1,
systems. There is still an enormous potential for the
Beuth Verlag, Berlin, First draft.
development of ERP-systems in particular

291
 CONDITIONS FOR TECHNOLOGY ACCEPTANCE
Broadening the Scope of Determinants of Ict Appropriation

Pieter Verdegem and Lieven De Marez

Research Group for Media & ICT (MICT), Ghent University (UGent)
Interdisciplinary Institute for Broadband Technology (IBBT), Korte Meer 7, 9000 Ghent, Belgium
[email protected], [email protected]

Keywords: User research, technology acceptance, Ict appropriation, adoption determinants, usage determinants.

Abstract: Regarding the ICT industry, we have a fast evolving sector being under pressure due to a growing number
of failing innovations. Companies are forced to be the first on the market and for that reason thorough
insights in user preferences are indispensable. New technological innovations often fail because too much
attention is given to (technical) product-related features without taking into account the most important
parameters of user acceptance. In this paper we highlight some theoretical considerations on this matter.
First of all, we propose an approach in which more traditional and often scattered vision(s) on adoption
determinants are broadened into an integrated framework. The approach should provide a stronger base for
better targeting of (new) users of technologies. Second, we elaborate on this by rethinking these
determinants with regard to the later adopters. Later adopters (or even non-users) are often ignored in
technology acceptance research. However, especially for policy purposes, the understanding of why people
do not adopt or do not use ICT is strongly relevant in the light of the development of the information
society. Both approaches – focusing towards early as well as late adopters – are illustrated by case studies
starting from a common framework.

“Consumer research has shown that knowledge about (and as a consequence more failing technologies), all
the user tends to be limited … It is quite self-evident stakeholders involved are desperately seeking for
that both social and commercial policies will benefit accurate insights into adoption determinants as a
from accurate insight into the different parameters basis for more effective introduction and targeting
determining the success (acceptance and use) of these
technologies according to a user’s point of view.”
strategies (Lin, 1998: 95; Talukdar et al, 2002: 97;
(Burgelman, 2000: 236) Ziamou, 2002: 366; Chen et al, 2002: 706;
Venkatesh et al, 2003: 426). From a policy point-of-
view such insights into drivers and barriers for
adoption and usage of ICT are necessary in order to
1 INTRODUCTION set up adequate e-inclusion measures (Chaudhuri et
al, 2005: 737-739; Milner, 2006: 177; Trkman et al,
Conditions for ‘technology acceptance’ have always 2008: 102).
been a central pillar in all kinds of approaches of In this paper we introduce a framework that
studying the acceptance and appropriation of new could help to refine our thinking on this. First, we
innovations: ranging from the diffusion theory-based broaden the scope on adoption determinants by
approaches focusing on perceived technology integrating the existing but fragmented approaches
characteristics since the early 60’s, over more usage- into a more comprehensive one. This becomes more
oriented theoretical approaches since the 80’s to important for industrial and marketing purposes, as a
more industry-oriented studies/approaches focusing thorough understanding of the user – the customer –
on image- and network-related determinants. is necessary for acceptance. Second, we elaborate on
However, in today’s ICT-environment a broader this by paying attention to approaches that go
and more comprehensive framework for under- beyond adoption diffusion. More specifically, policy
standing determinants or conditions for technology makers are seeking to understand parameters that
acceptance is more than ever needed, in order to have an influence on the impact of ICT adoption and
obtain the necessary insights to face the challenges use, in order to formulate effective measures in the
of both ICT managers and policy makers. Due to the light of overcoming digital inequalities.
exponentially increased offer of ICT-innovations

292
 CONDITIONS FOR TECHNOLOGY ACCEPTANCE - Broadening the Scope of Determinants of Ict Appropriation

2 DETERMINANTS FOR ICT Downside of this increased attention is that

ACCEPTANCE researchers nowadays are confronted with a ‘lack of
overview’, since the increased multidisciplinary
interest entails a cluttered and inconveniently
2.1 Broadening the Scope on Adoption arranged entirety of determinants. Evidently, more
Determinants accurate insight into adoption determinants requires
an insight in more than the five determinants of
With ‘adoption determinants’ we refer to parameters Rogers’ diffusion theory, but it remains unclear how
that influence technology acceptance in terms of the many and which determinants should be taken into
actual adoption decision (De Marez, 2006: 189-192). account. Since a convenient overview of
For a long time and to a large extent influenced by (potentially) relevant adoption determinants for ICT
the dominant technological deterministic paradigm, innovations is still lacking to date (Busselle et al,
demographic variables were supposed to have an 1999; Randolph, 1999; Hadjimanolis, 2003 – an
important influence on that adoption decision (see exception is the work of Venkatesh et al, 2003) we
Rogers, 1983; Rogers, 2003). However, many conducted a meta-analysis on determinants for ICT
scholars have stated that – in addition to the more adoption (De Marez, 2006). Comparable to the
traditional parameters – this view should be development of UTAUT (Unified Theory of
extended to an approach based on ‘attitudinal’ Acceptance and Use of Technology, Venkatehs et al,
adoption determinants (Bergman et al, 1995; Plouffe 2003: 446-465), we started from different studies
et al, 2001; Atkin et al, 2003; Leung, 1998).. and existing theoretical models (in the field of
Attitudinal determinants are related with more communication, marketing as well as social
subjective perceptions of innovation characteristics psychology) whose central building block was
and personality traits. mainly diffusion theory’s set of five determinants.
The approach of this attitudinal adoption This resulted in an extension to 19 determinants, in
determinants was mainly inspired by the diffusion which we distinguish ten innovation-related
theory, in which innovations were supposed to have characteristics (perceptions), eight adopter-related
a set of five characteristics (relative advantage, characteristics, and the impact of the marketing
complexity, compatibility, trialability and strategy (see table 1 below).
observability) of which the subjective perception
determines one’s attitude towards the technology, Table 1: Extension of adoption determinants (De Marez et
and one’s innovativeness or timing of adoption al, 2007: 82).
decision (Rogers, 1983; Rogers, 2003). The
perception of each of these characteristics is ASSUMED
ADOPTION
assumed to have a strong relationship with the RELATION WITH
DETERMINANT
innovativeness of an individual. Innovators and early INNOVATIVENESS
adopters, for example, are assumed to have a higher INNOVATION RELATED CHARACTERISTICS
Compatibility +
perception of relative advantage than the (later) Complexity -
majority segments, together with a lower perception Cost -
of complexity of the innovation (contrary to the later Enjoyment +
adopters). Observability +
Over the years, the increasing attention paid to Relative advantage +
Reliability +
these ‘attitudinal’ adoption determinants resulted in Tangibles +
a considerable yet cluttered extension of the original Trialability +
set of five adoption determinants. The convergence Visibility +
with social psychology theories such as the Theory ADOPTER RELATED CHARACTERISTICS
of Reasoned Action (TRA) (Fishbein, 1967; Control/Voluntariness +
Image/Prestige +
Fishbein & Ajzen, 1975), (Decomposed) Theory of Innovativeness +
Planned Behaviour ((D)TPB) (Ajzen, 1991; Taylor (product) Knowledge +
& Todd, 1995) and Technology Adoption Model Opinion leadership +
(TAM) (Davis, 1986; Davis, 1989) in particular led Optimism +
to an extremely valuable - yet fragmented - increase Social influence +
Willingness to pay +
in (research on) adoption and determinant models.
IMPACT OF MARKETING STRATEGY
Some scholars consider one or two extra Marketing (impact) +
determinants (Holak & Lehmann, 1990), while
others considered eight (Plouffe et al.; 2001), ten Clearly, innovativeness and adoption decisions
(Choi et al., 2003) or more determinants. seem to be determined by more characteristics than

293
ICE-B 2008 - International Conference on e-Business

the original five initiated by Rogers’ diffusion why they are not willing to adopt. This could
theory. The perception of ‘relative advantage’ for provide insights in how to adjust the innovation (in
example, can express itself in several dimensions. all its dimensions: product, distribution,
The ‘perceived cost’ and ‘tangibles/aesthetics’ are communication) in order to stimulate appropriation
the most important of them. Most scholars relegate by the overall population. On the other hand, in view
to Rogers’ work in his conceptualization of of the pervasiveness of ICT in society and the
‘observability’ in terms of the perceived result increasing dependence on ICT in everyday life,
demonstrability, while some others distinguish the policy makers are obliged to think about policies
latter from ‘visibility’ as the degree to which the that prevent exclusion of groups of citizens in the
innovation is visible to others in its own right. It is development of the information society. Insights in
also important to account for the ‘perceived the parameters of adoption by later adopters is
enjoyment’ of using the innovation (the so-called therefore of crucial importance.
likeability), and ‘reliability’ as a dimension of The adoption of a certain technology (as for
perceived risk that is not covered by other which the determinants are discussed in 2.1.)
determinants (‘reliability’ in this context refers to however, cannot be the sole focus when studying the
‘performance risk’). ‘Innovativeness’, on the other factors that influence technology acceptance. This
hand, is the most important personality would be too much a technology deterministic and
characteristic. It covers a multitude of sub diffusion-based approach, mainly serving ‘industry
dimensions such as ‘venturesomeness’, ‘novelty purposes’ (how to approach the most interesting
seeking’, ‘cosmopolitanism’, ‘variety seeking’, segments of innovators, early adopters, early
‘information seeking’, etc. ‘Opinion-leadership’ majority as good and as soon as possible?). A more
needs to be considered as a separate dimension, just elaborated focus on technology acceptance not only
as a person’s ‘optimism’ towards technology, requires a focus on adoption, but also on usage
‘product knowledge’, ‘willingness (and ability) to determinants. In addition, a thorough understanding
pay’, the ‘perceived impact on one’s personal of technology acceptance not only asks for a focus
image’, the ‘perceived control’, ‘impact of social on the first segments in the diffusion curve, but also
influences’ and the ‘impact of marketing, advertising on the later segments in that curve (late majority and
and promotional strategies’. laggards).
If industry strategies nowadays require more Attention for digital inequalities is, both in
profound insight in more than the traditional five scholarly publications as well as in political studies
determinants, it will largely boil down to an insight and in the popular press and media is, an obvious
in these 19 determinants. It will probably never be result of the euphoric ‘cyberbole’ that characterized
the case that all these determinants are relevant, but much of the rhetoric of new technologies since the
if prior-to-launch research could reveal which mid-1980’s (Gunkel, 2003: 500). Hence, profound
determinants are the most important drivers and insights in why people lag behind in the adoption
barriers for which segments, this would allow to and use of new technologies, are important in view
adjust the approach of different segments. Question of the development of the information society for all.
remains, however, how to acquire such prior-to- More insights are necessary, especially when we can
launch insight? conclude that business strategies and policies that
were successful in, for instance, increasing internet
2.2 Elaboration of Determinants with penetration in the early days, may no longer be
Regard to ‘Later Adopters’ appropriate to reach the rest of the society. And this
is most probably so in societies where a majority of
Another challenge of research concerning the people are already connected to the internet. Thus,
acceptance of new technologies – especially for policies also need insights in the most important
policy strategies – is how to gain insight in the drivers and barriers that have an impact on the
profiles of later adopters. That are individuals to individual’s decision to appropriate an ICT product.
whom traditionally less attention is given in
innovation studies (Selwyn, 2003: 100-101; Roe &
Broos: 91). People who step later into the innovation
circle or who even resist to do this, are often left
aside. However, research of non or later adoption
could offer fundamentally added value. First,
industry or managers could learn substantially not
only of why people adopt a new technology but also

294
 CONDITIONS FOR TECHNOLOGY ACCEPTANCE - Broadening the Scope of Determinants of Ict Appropriation

3 CASE STUDIES Table 2: Operationalisation of determinants in 47 Likert

statements (applied to the mobile news/TV cases) (cont.).
3.1 Broadening for Industry & COST (RELATIVE ADVANTAGE)
Marketing Purposes 1. Subscription on mobile news/TV services seems expensive
to me.
5. Mobile news/TV services will probably cost too much for
The above-mentioned question was also the central many people.
question in two recent case studies conducted by EFFECTIVENESS (RELATIVE ADVANTAGE)
Research Group MICT-IBBT. Both studies were set 36. Mobile news/TV services will certainly make some things
easier for me.
up to acquire the necessary insights in attitudinal
ENJOYMENT
adoption determinants for two ‘mobile innovations’, 4. Mobile news/TV service seems very user friendly to me.
in preparation for their commercial launch in IMAGE PRESTIGE
Belgium. In the first case-study (2006) a sample of 33. Subscribing to mobile news/TV services applications
269 respondents was questioned about their attitudes would have a positive impact on my image and social status.
38. Subscribing to mobile news/TV services beams out a
towards a new ‘mobile news’-application in the certain standing.
context of the IBBT-project ROMAS. In the second INNOVATIVENESS
case study (2007) a representative Flemish sample of 6. I think to be among the first to subscribe to such mobile
405 respondents was questioned about their attitudes news/TV services.
towards mobile television services. In the first study, 34. Based on what I already knew about the application and
what I have learned today, I will certainly search for more
data were collected by means of an online survey information about subscribing to these services.
(after a two months period in which the respondents MARKETING STRATEGY
could test the mobile news application). In the 26. If I would subscribe to a mobile news/TV application, it
second study, data were collected by means of 40 would be important to me that it is provided by a well-known
'brand'.
minutes during CAPI-interviews (in which
27. If I would consider mobile news/TV adoption, I would
respondents were shown short movies on DVD in first check the ads, brochures and promotions.
order to familiarize them with mobile tv applications OBSERVABILITY — RESULT DEMONSTRABILITY /
and usage moments). In both studies potential COMMUNICABILITY
adopter segments (innovators up to laggards) for the 24. I am perfectly able to explain the strengths and the
weakness of mobile news/TV services to others
innovations were forecasted by means of the Product OBSERVABILITY — VISIBILITY
Specific Adoption Potential scale (De Marez, 12. One of the nice things of a mobile news/TV application is
Verleye, 2004a,b), and the 19 determinants were that it is something to show off with among friends.
transformed into a battery of 47 Likert statements 17. I see many people in my environment who use mobile
(cf. table 2), to be answered on 5-point agreement news/TV services.
OPINION LEADERSHIP
scales (varying from 1: ‘I do not agree at all’ to 5: ‘I 15. If mobile news/TV would be introduced on the market,
fully agree’). people in my environment will certainly come to me for
advice.
Table 2: Operationalisation of determinants in 47 Likert OPTIMISM
statements (applied to the mobile news/TV cases). 44. The fast technological developments are a good thing.
45. If you don't want to run behind, adoption of new
COMPATIBILITY — LIFESTYLE AND technologies is necessary.
PERSONALITY PERCEIVED RISK (FINANCIAL)
7. Consultation of mobile news/TV services fits my lifestyle 18. I fear that subscribing to a mobile news/TV application
39. If I buy a new mobile, it has to be a model that fits my would be way above my budget.
personality PERCEIVED RISK (IMPLEMENTATION)
COMPATIBILITY — (TECHNOLOGICAL) 23. If I would have to use such mobile news/TV applications
30. I am interested in subscribing to mobile news/TV services? on my own, I don't think I would manage.
but I would mind if that would imply an investment in a new PERCEIVED RISK (SOCIAL)
device. 21. If I would use mobile news/TV services, people in my
13. Mobile news/TV services are only interesting to me as a environment would look odd at me.
part of the subscription on other mobile services. PRODUCT KNOWLEDGE
COMPLEXITY/COMFORT LEVEL 19. I recently send something about mobile news/TV services
8. I fear that mobile new/TV services application offers or recently talked to someone about it.
different possibilities, which makes It rather complicated. 35. I consider myself well-informed about the possibilities and
20. The mobile news services application seems very user- (dis)advantages of mobile news/TV services.
friendly to me. RELATIVE ADVANTAGE
29. The mobile news/TV services application offers different 11. The advantages of mobile news/TV services are clearer to
possibilities, which makes it rather complicated. me than the disadvantages.
CONTROL/SELF-EFFICACY 40. I don't see where or when to use mobile news/TV services.
46. I have no problem to sort out on myself how mobile
news/TV application work and must be installed.

295
ICE-B 2008 - International Conference on e-Business

Table 2: Operationalisation of determinants in 47 Likert in four focus group interviews with the goal to select
statements (applied to the mobile news/TV cases) (cont.). the best way to translate the item into a statement.
RELATIVE ADVANTAGE All 269 (mobile news study) and 405 (mobile
11. The advantages of mobile news/TV services are clearer to television study) respondents completed the entire
me than the disadvantages. questionnaire. The most important results show a
40. I don't see where or when to use mobile news/TV services. striking difference between the attitudes or
RELIABILITY
31. I doubt the reliability and proper functioning of the mobile
determinants for both innovations. In the average
news/TV services application. agreement scores, for example, it can be noticed that
SOCIAL INFLUENCE a determinant as ‘tangibles’ (14, 25) is more
9. Most people in my environment will certainly be enthusiast important for mobile television than for mobile
about the mobile news/TV application. news. Regarding ‘reliability’ (31) people seem to be
SOCIAL INFLUENCE — COMPLIANCE
2. If 'mobile news/TV usage' would be considered as 'trendy' more sceptical for mobile news, while the ‘perceived
in my environment, I would certainly consider subscribing to control’ (46) seems to result in a higher score for
it. this new mobile application. ‘Product knowledge’
10. My direct environment will probably expect me to be one (19, 35) on the other hand is lower for mobile
of the first to use mobile news/TV services.
32. Even if I am interested, I would not subscribe if my
television; etc … . With an R² ranging between .503
environment would be negative about mobile news/TV and .795 for the earlier adopters and early majority,
applications. these 47 ‘determinant operationalisations’ certainly
SOCIAL INFLUENCE — IDENTIFICATION seem to be a good set of variables to explain the
47. If I would use mobile news/TV services, it would certainly variance in the dependent variable ‘adoption
tell something about me and my personality.
SOCIAL INFLUENCE — INTERPERSONAL
intention’. Even for the later adopter segments this
COMMUNICATIONS R² still ranges between .34 and .42. Detailed
3. Before subscribing to a mobile news/TV application, I information about the psychometric reliability and
would like the advice of some people. validity can be found in De Marez et al (2007: 86-
16. Mobile news/TV services will certainly be a topic of
88).
discussion among my friends and family.
SOCIAL INFLUENCE — NETWORK Thus, for both technological innovations, this set
EXTERNALITIES of attitudinal determinant statements explains
37. I am interested in subscribing to mobile news/TV services, adoption intentions quite well, but there remain
but only if there are sufficient people in my direct environment many differences in the significant determinants for
doing so. Otherwise, the application wouldn't have much value
to me. the different innovations and adopter segments.
TANGIBLES (RELATIVE ADVANTAGE) ‘Lifestyle compatibility’ (39) for example is only
25. As the mobile news/TV services is presented and testable significant for the mobile television’s innovators, not
now it has an attractive design and style. for mobile news. Also the ‘cost perception’ (1) is
14. If I would consider buying a new mobile, design would be only significant in the mobile television case
a very important buying argument to me.
TRIALABILITY — PHYSICAL (laggards). ‘Trialability’ (41) then is significant in
41. I would like to try out mobile news/TV services before both cases, but not for the same segments. Other
subscribing to them. determinants such as the perceived impact of
TRIALABILITY — VICARIOUS adoption on one’s ‘image’ (33) was only significant
28. Before subscribing or adopting mobile news/TV services I
prefer to look around for a while and see how others are
for mobile news’ innovators and laggards. So, we
experiencing the application. can notice many differences in attitudes, as well
VOLUNTARINESS when compared over the two cases, as compared
42. If I would subscribe to mobile news/TV services, it would over the different adopter segments. This
completely be my own decision. No one would influence me emphasizes the need for a product- and segment
in making that decision.
WILLINGNESS-TO-PAY
specific approach when studying adoption
22. Even if it costs a bit more, mobile news/TV is something I determinants.
really want.

The transformation of determinants into a scale

3.2 Elaboration for Policy Purposes:
of 47 items is the combined result of desk research Analysis of Non-Adopters
and qualitative research by means of focus group
interviews. A first phase of desk research resulted in The need for more profound insights in why people
a long list of statements of 19 determinants used in do not use ICT innovations, for instance computer
other studies and models (both diffusion theory and internet, is an important question for policy
based models as well as social psychology based makers. For instance, as more people are connected
models). In addition to this, the long list was verified and taking full advantage of new possibilities that
are offered via internet, government cannot ignore

296
 CONDITIONS FOR TECHNOLOGY ACCEPTANCE - Broadening the Scope of Determinants of Ict Appropriation

those groups that are not yet connected. So, policy possible to determine a hypothetical ‘turning point’
makers should at least take the initiative to set up for ICT use, namely the point at which the benefits
measures that can help people – who risk to be will outweigh the cost of appropriating an ICT
excluded – to enhance their participation in the product for a certain category of users.
information society. On a practical level, in order to set up effective
Research Group MICT-IBBT was commissioned e-inclusion measures, the advantage of this method
with this research question by Fedict (The federal is that groups of individuals with relatively
public agency for information and communication homogeneous ASA-profiles, can easily be identified
technology) of the Belgian federal government. As and reached by policy makers. Very often they are
the responsible agency for stimulating ICT represented by professional or social organizations
acceptance and use in society, Fedict needed a that know how to reach them and are willing to
scientific supported base for setting up new cooperate with government. A specific offering can
initiatives. The research results presented below then be proposed to these groups, taking into
draw on the experience that the Belgian government account the specificities of their ASA-profile and
has acquired through the ‘Internet for all’ project in socio-economic background.
2006. The latter was set up in collaboration with The approach draws upon the assumption that
ISPs, PC manufactures and retailers, and entailed the members of socio-demographically and socio-
provision of an affordable package (€750 - €1000) economically homogeneous groups yield similar
deal to customers, consisting of a PC, an internet perception in terms of access, skills and attitudes
connection plus a training session. It was calculated towards ICT. This hypothesis was tested by means
that the project contributed to 16% of the increase of of a quantitative survey (personal interviews with
new internet connections over a period of one year 184 respondents). The research population was
(Verdegem & Verhoest, 2008: 38). A critical composed of a theoretical sampling, meaning that
evaluation of the ‘Internet for all’ project revealed we selected individuals based on a limited number
different elements, two of which inspired our of characteristics, i.e. variables of which previous
research. The first was merely the confirmation of research has shown that they are of major
what could be expected. Not all of the groups in importance for (non-)adoption of ICT. In the
society were equally well served by the campaign. research we recruited individuals (non-users) from
For instance, for some individuals the proposed ten groups, varying from single mothers with
offering was too expensive. The second source of children to physical therapists. This resulted in
inspiration was an incidental call of a representative certain prototypical profiles, exemplary for the
of a professional organization of physical therapists societal diversity without being representative for
that proposed to target the campaign also towards the overall population (for detailed information see
the members of his organization. These two Verdegem & Verhoest, 2008).
observations triggered a reflection that inspired the In order to map the respondents’ perceptions of
new policy approach and adjoining research. computer and internet use at home, we presented
The new approach is articulated around the them with a list of statements. The statements were
concept of ‘relative utility’, a sociological based on the same adoption determinants that are
reinterpretation of the economic concept of mentioned above (see table 2). A number of these
‘marginal utility’. Contrary to the other case studies statements aimed at obtaining information about the
illustrated in this paper, of which the goal was to respondent’s specific ASA-profile: 1) positive or
broaden the insights concerning adoption negative attitudes towards computer and internet at
determinants, this case is focused on the elaboration home; 2) the presence or lack of skills and
and interpretation of parameters of ICT competences towards using ICT and 3) the presence
appropriation. By paying attention to both the or absence or barriers to access ICT. Other
adoption as well as the usage decision we wanted to statements served as measurement scales to gain
provide input for measures that would help to insights in more generic factors such as, for
stimulate ICT adoption and use. Following the example, the influence of social networks or
relative utility approach, the assumption is that the marketing strategies of the ICT industry.
specific combination of conditions in terms of access Based on the answers of the respondents on the
to ICT, skills to master the devices and attitudes statements cluster analysis revealed five distinctive
towards the technology, has an impact on whether groups of domestic non-users of computer and
people will use ICT or not. More specifically, based internet:
on the combination of perceptions of people towards
access, skills and attitudes (ASA) it becomes • Incapable refusers;
• Self-conscious indifferents;

297
ICE-B 2008 - International Conference on e-Business

• The willing but incapable; these parameters, who have an impact on technology
• Skilled ICT-lovers with limited access; acceptance, should be reconsidered. The described
• Price sensitive pragmatists. elaboration contains both an exercise of broadening
and deepening.
The clusters demonstrate that non-adopters or non- First of all, it is important to examine which
users should not be seen as one generic group. Each determinants are of major relevance in order to
profile represents a different combination of the forecast how new innovations should be brought to
factors investigated, in which each factor carries a the market to persuade the potential (first) adopters,
different weight. Statistical testing was also or those interested in the product. Not only the
conclusive about the relationship between the group product development in terms but also the targeting
membership (from theoretical sampling) and the and marketing campaigns strongly ask for accurate
membership of the ASA-profiles (Pearson Chi- insights into user preferences. Particularly in the
Square p ≤ 0,01). As such, we found empirical (pre-)launch phase.
foundation for the assumption that homogeneous In addition, the framework of adoption
groups – in terms of socio-demographic and socio- determinants should also be re-evaluated with regard
economic characteristics – result in generic ASA- to later adopters. People who enter the adoption
profiles. process in a later stadium – or who even resist to
Following on this quantitative research adopt – may have clear reasons for that. However,
qualitative in-depth interviews and focus group deep understanding of who is making less (or even
interviews were organized to refine our thinking of no) use of information technologies remains weak.
why people do not use ICT and to examine which Nevertheless, this is of major importance for both
leverages could lift them over the turning point policy makers, as well as for ICT managers.
between non-usage and usage. The results of both So, in a nutshell, our approach contains both a
research stages show the advantage of the approach managerial as a policy relevance. Furthermore, we
proposed and offers the opportunity for policy also hope that this paper contributes to both
makers to set up measures to stimulate later adopters theoretical reconsiderations as well as the
to ICT appropriation. These measures could be more methodological foundation of technology acceptance
effective as they are based on strategies of research.
segmentation and differentiation, taking into account
the different profiles of these individuals. The
elaboration of adoption and usage determinants is
thus necessary to gain insight in a group of ACKNOWLEDGEMENTS
individuals that are often ignored in innovation
research. This work was supported by the IBBT projects
ROMAS (Research on Mobile Applications &
Services) and MADUF (Maximizing DVB Usage in
Flanders). Both projects are funded by the
4 CONCLUSIONS Interdisciplinary Institute for Broadband Technology
(IBBT) and a consortium of companies. The last
Our research results clearly show the need for a case study is funded by the Federal Public Agency
thorough understanding of user attitudes towards for Information and Communication Technology
ICT acceptance. As more technological innovations (Fedict – Belgian Government).
are introduced in rapid succession and an increased
number of those innovations is failing, accurate
insights in the determinants towards adoption and
use become increasingly important. We could state REFERENCES
that both our theoretical reconsiderations as well as
the empirical foundations of them, provide ICT Ajzen, I. (1991). The theory of planned behavior.
Organizational Behavior and Human Decision
managers as well as policy makers with useful input
Processes, 50(2): 179-211.
in support of their innovation strategies. As a matter Atkin, J.A., Neuendorf, K., Jeffers, L.W. & Skalski, P.
of fact, the development of an information society (2003). Predictors of audience interest in adopting
for all serves both economic as social purposes. digital television. The Journal of Media Economics,
The approach proposed started from the same 16(3) 159-173.
common framework, i.e. more traditional adoption Bergman, S., Frissen, V. & Slaa, P. (1995). Gebruik en
determinants who are founded by technological betekenis van de telefoon in het leven van alledag. In:
deterministic inspired paradigms. We illustrated that Rathenau Instituut, Toeval of noodzaak? Geschiedenis
van de overheidsbemoeienis met de informatie-

298
 CONDITIONS FOR TECHNOLOGY ACCEPTANCE - Broadening the Scope of Determinants of Ict Appropriation

voorziening (277-325). Fatima Reeks, Amsterdam: Holak, S.L. & Lehmann, D.R. (1990). Purchase intentions
Meboprint. and the dimensions of innovation: an exploratory
Burgelman, J.C. (2000). Innovation of communication model. Journal of Product Innovation Management,
technologies: some general lessons for the future from 7(1), 59-73.
the pas. In: Cammaerts, B. & Burgelman, J.C. (ed.) Leung, L. (1998). Lifestyles and the use of new media
Beyond competition: broadening the scope of technology in urban China. Telecommunications
telecommunications policy (229-238). Brussel: VUB Policy, 22(9) 781-790.
Press. Lin, C.A. (1998). Exploring personal computer adoption
Busselle, R., Reagan, J., Pinkleton, B. & Jackson, K. dynamics, Journal of Broadcasting & Electronic
(1999). Factors affecting internet use in a saturated- Media, 42(1), 95-112.
access population. Telematics & Informatics, 16(1-2), Milner, H. (2006). The digital divide: the role of political
45-48. institutions in technology diffusion. Comparative
Chaudhuri, A., Flamm, K.S. & Horrigan, J. (2005). An Political Studies, 39(2), 176-199.
analysis of the determinants of internet access. Plouffe, C.R., Vandenbosch, M. & Hulland, J. (2001).
Telecommunications Policy, 29(9-10), 731-755. Intermediating technologies and multi-group adoption:
Chen, L.D., Gillenson, M.L. & Sherrel, D.L. (2002). a comparison of consumer and merchant adoption
Enticing online consumers: an extended technology intentions towards a new electronic payment system.
acceptance perspective. Information & Management, Journal of Product Innovation Management, 18(2),
39(8), 705-719. 65-81.
Choi, H., Choi, M., Kim, J. & Yu, H. (2003). An empirical Randolph, C. (1999). Why do we hate Microsoft and still
study on the adoption of information appliances with a use word? Proceedings of the 10th Australasian
focus on interactive TV. Telematics and Informatics, Conference on Information Systems, 736-748.
20(2), 161-183. Roe, K. & Broos, A. (2005). Marginality in the
Davis, F.D. (1986). A technology acceptance model for information age: the socio-demographics of computers
empirically testing new end-user information systems: disquietude. A short research note. Communications:
theory and results. Unpublished PhD dissertation, the European Journal of Communication, 30(1), 91-
Cambridge: MIT Sloan Management School. 96.
Davis, F.D. (1989). Perceived usefulness, perceived ease Rogers, E.M. (1983). The diffusion of innovations (3rd ed.).
of use and user acceptance of information technology. New York: The Free Press.
MIS Quarterly, 13(3), 319-340. Rogers, E.M. (2003). Diffusion of innovations (5th ed.).
De Marez, L., Verleye, G. (2004a). ICT-innovations New York; The Free Press.
today: making traditional diffusion patterns obsolete, Selwyn, N. (2003). Apart from technology: understanding
and preliminary insight of increased importance. people’s non-use of information and communication
Telematics and Informatics, 21(3): 235-260. technologies in everyday life. Technology in Society,
De Marez, L., Verleye, G. (2004b). Innovation diffusion: 25, 99-116.
the need for more accurate consumer insight. Talukdar, D., Sudhir, L. & Ainslie, A. (2002).
Illustration of the PSAP-scale as a segmentation Investigating new product diffusion across products
instrument. The Journal of Targeting, Measurement and countries. Marketing Science, 21(1), 97-114.
and Analysis for Marketing, 13(1): 32-49. Taylor, S. & Todd, P. (1995). Decomposition and
De Marez, L. (2006). Diffusie van ICT-innovaties: crossover effects in the theory of planned behavior: a
accurater gebruikersinzicht voor betere introductie- study of consumer adoption intentions. International
strategieën. Unpublished PhD dissertation, Journal of Research in Marketing, 12(2), 137-155.
Department of Communication Studies, Ghent Trkman, P., Blazic, B.J. & Turk, T. (2008). Factors of
University. broadband development and the design of a strategic
De Marez, L., Vyncke, P., Berte, K., Schuurman, D. & De policy framework. Telecommunications Policy, 32(2)
Moor, K. (2007). Adopter segments, adoption 101-115.
determinants and mobile marketing. Journal of Verdegem, P. & Verhoest, P. (2008). The ‘relative utility’
Targeting, Measurement and Analysis for Marketing, approach for stimulating ICT acceptance: profiling the
16(1), 78-95. non-user. European Journal of ePractice, Nr. 3, May
Fishbein, M.A. (1967). Readings in attitude theory and 2008, http://www.epracticejournal.eu/document/4717.
measurement. New York: Wiley. Venkatesh, V., Morris, M.G., Davis, G.B. & Davis, F.D.
Fishbein, M.A. & Ajzen, I. (1975). Belief, attitude, (2003). User acceptance of information technology:
intention and behavior: an introduction to theory and toward a unified view. MIS Quarterly, 27(3), 425-478.
research. Reading MA: Addison-Wesley. Ziamou, P. (2002). Commercializing new technologies,
Gunkel, D.J. (2003). Second thoughts: toward a critique of consumers’ response to a new interface. Journal of
the digital divide. New Media & Society, 5(4), 499- Product Innovation Management. 19(5), 365-374.
522.
Hadjimanolis, A. (2003). The barriers approach to
innovation. In: Shavinina, L.V. (ed.) The handbook of
innovation (559-573). Oxford, UK: Pergamon,
Elsevier.

299
 BARRIERS TO MOBILE BANKING ADOPTION
A Cross-national Study

Tommi Laukkanen
Department of Business and Management, University of Kuopio, Kuopio, Finland
[email protected]

Pedro Cruz
Instituto Superior de Gestao, Lisbon, Portugal
[email protected]

Keywords: Innovation resistance, Adoption, Mobile phone, Banking.

Abstract: The objective of this study is to explore barriers to mobile banking adoption in two distinct European
countries namely Finland and Portugal. Even successful innovation may face various types of resistance that
may paralyse customers' desire to adopt or use the innovation. We investigated the country effect to five
adoption barriers namely usage, value, risk, tradition and image, derived from the earlier literature. An
Internet questionnaire was developed and 3.597 usable responses were collected. A confirmatory factor
analysis was implemented with SEM to build the constructs’ latent score levels. Using non-parametric
difference tests we concluded that the resistance is significantly lower among the Portuguese online bank
customers in terms of four out of the five barriers. The results can be used for a better understanding and
enhancement of adoption of this specific case of m-commerce.

1 INTRODUCTION mobile banking adoption in these two European

countries and how the countries differ in terms of
The wide penetration and rapid diffusion of mobile barriers to the service adoption.
phones has opened opportunities for new First we describe the Internet and mobile
innovations in the services sector. One such communications market both in Finland and
innovation is mobile banking representing one of the Portugal. Thereafter, we summarise the relevant
most promising, while still marginally adopted, m- literature on innovation resistance and banking
service. Previous studies have shown that mobile technologies. Finally, the findings are presented and
banking increases efficiency and convenience in bill concluding remarks drawn.
paying, for example, as the service can be used
wherever wanted enabling time savings and
immediate reactions to unexpected service need 2 INTERNET AND MOBILE
(Laukkanen and Lauronen, 2005; Laukkanen, COMMUNICATIONS MARKET:
2007a).
Finland has long been seen as the most FINLAND VS. PORTUGAL
successful European country in terms of the
adoption and use of mobile services (Bouwman et During the last decade the penetration of mobile
al., 2007). However, even though already around phones has been dramatic. In addition, the diffusion
two thirds of the Finns pay their bills over the of Internet-connected computers has been
Internet, mobile banking has not yet received the remarkable in the 21st century. These advances in
attention of the masses. In general, Finland is communication technologies have reshaped the
referred as one of the leading European countries in service development and revolutionised the service
terms of Internet banking adoption, while, for consumption. In Finland the amount of Internet-
example, Portugal is lacking far behind (Eurostat, connected computers per 100 persons have grown
2007). In this study we investigated what inhibits from roughly 20 in 2002 to nearly 50 in 2006

300
 BARRIERS TO MOBILE BANKING ADOPTION - A Cross-national Study

(Figure 1). Compared to Portugal, the relative paper we explore the adoption barriers to mobile
amount of these devices is over three times higher in banking in the light of consumer resistance to
Finland. These figures may partly explain the higher innovations.
Internet banking adoption rates in Finland, even
though the growth rate of these devices in Portugal
has been dramatic during the last years. 3 LITERATURE
50
40 Albeit the pro-innovation bias (Sheth, 1981; Ram,
1987; Rogers, 2003) that majority of the diffusion
30 literature has, there may be product and service
20 categories or market segments where innovation
10 resistance is predominant (Gatignon and Robertson,
1991). The literature on innovation resistance aims
0
2002 2003 2004 2005 2006 to explore the various reasons that inhibit innovation
Finland Portugal adoption. Sheth (1981) suggests that the two key
factors explaining the phenomenon are habit or
satisfaction with an existing behaviour and
Figure 1: Internet-connected computers per 100 persons perceived risks associated with innovation adoption.
(Statistics Finland, 2008).
He states that the inclination toward an existing
behaviour is related to the typical human tendency to
1,2 strive for consistency and status quo rather than to
1,1 continuously search for new behaviours.
Consequently Ellen et al. (1991), note that
1,0
satisfaction with current performance increases
0,9 resistance to alternatives and reduces the likelihood
0,8 of adoption. They further highlight the role of
perceived self-efficacy which means the perceived
0,7
2002 2003 2004 2005 2006
ability or skill to successfully perform a given task.
Finland Portugal These lines of thought assume that consumers base
their decisions on two aspects: perceived benefits
over existing methods and perceived risks associated
Figure 2: Mobile telephone subscription per capita
with innovation adoption.
(Statistics Finland, 2008).
Ram and Sheth (1989) suggest a more
comprehensive view on innovation resistance by
Although the distribution and penetration of
explaining the phenomenon with five adoption
mobile handsets in Finland is argued to be among
barriers namely usage, value, risk, tradition and
the highest in the world, making the country an
image.
interesting test-market for new mobile services
(Bouwman et al., 2007), the number of mobile
phone subscriptions is even higher in Portugal with
3.1 Usage Barrier
1,16 connections per capita compared to 1,08 in
Ram and Sheth (1989) suggest that the usage barrier
Finland. The relatively low number of computers
relates to the situation in which an innovation is not
connected to the Internet and a great number of
compatible with existing workflows, practices or
mobile phones make Portugal a highly potential
habits. In the context of technological innovations,
market for mobile services such as banking.
however, this construct parallels with complexity
The future of mobile communications relies
which, according to Rogers (2003), refers to the
heavily on services. However, the optimistic and
degree to which an individual considers an
experimental mood that we witnessed in the
innovation to be relatively difficult to understand
beginning of the century has been replaced by a
and use.
cautious atmosphere in which fewer risks are taken
The small size of mobile devices including small
in the development and marketing of new third
screens and tiny multifunction keypads may be
generation mobile services (Bouwman et al., 2007).
troublesome to use and hamper the usability of
Therefore, insight into the reasons why consumers
mobile services. Earlier studies on mobile banking
are not adopting mobile services is needed. In this
show that the smaller screens appear adequate in

301
ICE-B 2008 - International Conference on e-Business

information-based mobile services, such as request adverse the innovation is (Dunphy and Herbig,
for account balance service, but those banking 1995).
services that involve transactions require a bigger As with many other technological innovations,
screen size (Laukkanen, 2007b). For example, some there appear to be security and privacy concerns to
bank customers consider bill payment via mobile mobile banking among some bank customers (Luarn
handheld device to be difficult and time consuming and Lin, 2005). Safety measures of personal details
as the device enables only a limited amount of and financial information by the bank are one of the
information processing and for this reason, the critical factors for the commercial success of mobile
whole bill is not visible on the display inhibiting the banking (Brown et al., 2003). A portable list of PIN
progress in the service process (Laukkanen and codes may also pose security threats as it may be
Lauronen, 2005; Laukkanen, 2007a). Moreover, lost by a customer and found by an untrustworthy
some studies highlight the importance of simple party (Kuisma et al., 2007).
authorization mechanisms in mobile banking Moreover, the extent to which a person believes
(Laukkanen and Lauronen, 2005) while some report a new technology will perform a job consistently
inconvenience due to changing PIN codes among and accurately (i.e. reliability) is highly important
some bank customers as the codes need to be carried risk-related factor in technology-based financial
along (Kuisma et al., 2007). service innovations (Lee et al., 2003). Mobile
phones, for instance, may be limited in
3.2 Value Barrier computational power, memory capacity and battery
life, limiting the use of mobile services (Siau and
The degree to which an individual believes that an Shen, 2003).
innovation is better than the idea it supersedes
determines the individual's decision to use the 3.4 Tradition Barrier
innovation (Rogers, 2003). This is called relative
advantage which is a related concept with the value The tradition barrier is related to the change an
barrier referring to the performance and monetary innovation may cause in a consumer's daily routines.
value of an innovation in comparison to its Thus, if the consumer considers routines important
substitutes (Ram and Sheth, 1989). in his/her daily behaviour, the tradition barrier will
In similar vein, the greater the perceived most likely be high. Moreover, the tradition barrier
advantage that mobile banking offers over other may arise when an innovation is incompatible with
ways of banking, the more likely it is to be adopted the consumer's existing values, norms and past
(Brown et al., 2003). The earlier studies show that experience (Ram and Sheth, 1989). Thus, an
the option to check the movements or transactions of innovation needs to be well-suited with the existing
an account wherever wanted increases customers' values and norms in order an individual to adopt the
feeling of control over their financial affairs adding innovation (Rogers, 2003).
value to service consumption (Laukkanen and Kuisma et al. (2007) showed that some
Lauronen, 2005). consumers resist Internet banking due to their habit
However, if an innovation does not offer greater of paying bills via bill paying ATMs. Alternatively,
performance to existing alternatives, it is not a customer may need social interaction and enjoy
worthwhile for consumers to change their behaviour talking to bank personnel as a strong desire to deal
(Ram and Sheth, 1989). The extent to which an with human tellers is found to discourage consumer
individual believes that using mobile banking is from adopting self-service technologies in banking
uneconomical, for instance, has a negative effect on (Marr and Prendergast, 1993). Thus, it may be that
the intention to use mobile banking (Luarn and Lin, in mobile banking the tradition barrier arises if an
2005). individual simply prefers to deal directly with the
bank clerk instead of using new banking
3.3 Risk Barrier technologies.

The risk perceptions in technological innovations 3.5 Image Barrier

usually arise due to the uncertainty to the
technology’s capability to deliver its expected The image barrier arises from unfavourable
outcome (Im et al, 2008). Thus, the diffusion of associations to the identity of the innovation, such as
innovation is likely to take the longer the more risk the country of origin, brand or the product category
to which the innovation belongs (Ram and Sheth,

302
 BARRIERS TO MOBILE BANKING ADOPTION - A Cross-national Study

1989). In the case of technological innovations, for

instance, image barrier may derive from a negative Usage barrier
image of new technology in general and of a product
class in particular. H1
In the late 90’s Fain and Roberts (1997) argued Value barrier
that the image barrier in online banking derives from H
a negative hard-to-use image of computers and the Mobile
Risk barrier H3
Internet. We argue that this may well be the case in banking
mobile banking today as some consumers may H
perceive the mobile technology to be too difficult to
Tradition barrier
use and therefore instantly form a negative image of H5
the service related to the mobile technology.
Image barrier
3.6 Hypotheses Development
Following the earlier literature on innovation
Figure 3: Conceptual model.
resistance a research model was designed (Figure 3).
According to the Eurostat’s (2007) statistics, Finland Table1: Measure development of the barriers.
is among the leading European countries in terms of
individuals’ Internet banking adoption with 63 Statements measuring the barriers
percent adoption rate in 2006. Portugal, for example, Usage barrier
represents the opposite with only 10 percent B1. In my opinion, mobile banking services are easy to use *
adoption rate. Based on these facts we hypothesise B2. In my opinion, the use of mobile banking services is
that the resistance to electronic banking services, convenient *
B3. In my opinion, mobile banking services are fast to use *
including mobile banking, is significantly lower
B4. In my opinion, progress in mobile banking services is clear *
among the Finns compared to Portuguese bank
B5. The use of changing PIN codes in mobile banking services is
customers. This leads us to the following convenient *
hypotheses: Value barrier
H1: Usage barrier to mobile banking is significantly B6. The use of mobile banking services is economical *
lower among the Finns compared to Portuguese B7. In my opinion, mobile banking does not offer any advantage
compared to handling my financial matters in other ways
H2: Value barrier to mobile banking is significantly B8. In my opinion, the use of mobile banking services increases
lower among the Finns compared to Portuguese my ability to control my financial matters by myself *
H3: Risk barrier to mobile banking is significantly Risk barrier
B9. I fear that while I am paying a bill by mobile phone, I might
lower among the Finns compared to Portuguese make mistakes since the correctness of the inputted information
H4: Tradition barrier to mobile banking is is difficult to check from the screen
significantly lower among the Finns compared B10. I fear that while I am using mobile banking services, the
battery of the mobile phone will run out or the connection will
to Portuguese otherwise be lost
H5: Image barrier to mobile banking is significantly B11. I fear that while I am using a mobile banking service, I
lower among the Finns compared to Portuguese might tap out the information of the bill wrongly
B12. I fear that the list of PIN codes may be lost and end up in
H6: Overall resistance to mobile banking is the wrong hands
significantly lower among the Finns compared B13. I trust that while I am using mobile banking services, third
to Portuguese parties are not able to use my account or see my account
information *
Tradition barrier
B14. Patronizing in the banking office and chatting with the
4 DATA AND METHODS teller is a nice occasion on a weekday
B15. I find self-service alternatives more pleasant than personal
customer service *
Based on the theory of innovation resistance and the Image barrier
existing literature on banking technologies, B16. In my opinion, new technology is often too complicated to
especially on mobile banking, a survey questionnaire be useful
was designed. The five adoption barriers were B17. I have such an image that mobile banking services are
difficult to use
examined with 17 statements expressed in Table 1.
*Reversed scale
A seven-point Likert scale ranging from totally
disagree (1) to totally agree (7) was used.

303
ICE-B 2008 - International Conference on e-Business

The questionnaire was first designed in Finnish the effects being 0.51 and 0.31 respectively. On the
and thereafter translated to English. The English other hand, tradition had a negative influence to the
questionnaire was then translated to Portuguese. The overall resistance, with a standardized structural
questionnaires both in Finland and Portugal were coefficient of –0.15.
placed in a log-out page of large banks’ online
service. Due to a vast number of online banking Usage barrier
users in Finland the questionnaire was open much
longer in Portugal than in Finland. In Finland the 0.93
questionnaire was open for 72 hours between Value barrier
November 6th and 9th 2006, whereas in Portugal the
0.84
questionnaire was open for 2 weeks, between June
Mobile banking
28th and July 13th 2007. The surveys generated a Risk barrier 0.31
resistance
total random sample of 3597 usable responses
– 0.15
without missing values.
The Finnish sample is slightly male dominated Tradition barrier
(53%) and relatively young with 36.7 percent of the 0.51
respondents being less than 35 years old. Finnish
sample consisted of a total number of 1.494 valid Image barrier
responses of which 28 percent (419 cases)
represented mobile banking users. The Portuguese Figure 4: Standardized parameter estimates.
sample is largely male dominated (61%) and even
younger than the Finnish sample with 59.3 percent The latent scores were calculated using
of the respondents being less than 35 years old. A Kolmogorov-Smirnov Z and Mann-Whitney U tests.
total number of 2.103 valid responses were obtained Both tests showed statistically significant differences
from Portugal with 32.7 percent (688 valid cases) of to the five constructs and overall resistance level
mobile banking users. (Table 2). Apart from the tradition barrier, the
In the data analysis phase the scales of positively results indicated higher values for Finland in all the
formed statements were reversed so that the scales barriers explored.
of all statements were comparable. Thus, a higher
mean of a statement determines higher resistance of Table 2: Resistance levels across countries (latent scores).
the respondent. A Structural Equation Model (with
AMOS 7.0 software) was estimated and its fit and Finland Portugal
Sig.
n=1494 n=2103
constructs’ reliability was checked. Latent scores
Overall resistance 4,052 3,516 0,000
were also computed at the constructs’ level. Using
Value barrier 3,058 2,781 0,000
non-parametric tests (Kolmogorov-Smirnov Z and
Image barrier 2,345 2,159 0,000
Mann-Whitney U), the differences between the
Tradition barrier 3,555 4,047 0,000
countries were assessed (variables and latent scores Risk barrier 3,879 3,594 0,000
did not follow normal distribution). Usage barrier 4,241 3,588 0,000

Table 3: Resistance levels across countries among non-

5 RESULTS users (latent scores).
Finland Portugal
The constructs’ Cronbach’s Alphas indicated n=1075 n=1415
Sig.
satisfactory internal consistency reliability Overall resistance 4,516 3,907 0,000
(usage=0,92; value=0,60; risk=0,80; tradition=0,59; Value barrier 3,420 3,083 0,000
image=0,65) and the estimated structural model Image barrier 2,583 2,287 0,000
showed an acceptable fit (χ2=2896,10; d.f.=115; Tradition barrier 3,514 3,828 0,000
p=0,00; CFI=0,90; RMSEA=0,08; GFI=0,91). All Risk barrier 4,114 3,752 0,000
coefficients revealed to be significant. Usage barrier 4,727 4,011 0,000
The second order structural model showed that
the resistance to the adoption of mobile services In addition, differences between countries were
mainly derives from usage (standardized structural computed only for non-users of mobile banking
impact=0.93) and value (0.84) barriers. Image and (Table 3). Out of the total number of 3597 responses
risk barriers also influenced the overall resistance, 2490 respondents represented this group of

304
 BARRIERS TO MOBILE BANKING ADOPTION - A Cross-national Study

customers with 1075 and 1415 observations in research is needed related to the role of
Finland and Portugal respectively. The results of innovativeness and demographic variables in mobile
these responses follow the research results of the banking adoption. Moreover, Finland and Portugal
total sample. Therefore, the hypotheses H1, H2, H3, represent very divergent countries in terms of
H5 and H6 are rejected and only the hypothesis H4 cultural dimensions (e.g. Hofstede, 1980), hence
is supported by the data in terms of both the total providing good means to study the effect of culture
sample and the sample of non-users. (Kivijärvi et al., 2007). Future research could
investigate the role of culture in consumer resistance
to technological innovations.
6 CONCLUSIONS AND FUTURE
RESEARCH REFERENCES
The structural equation model showed that usage Bouwman, H., Carlsson, C., Molina-Castillo, F.J.,
and value barriers are the most intense determinants Walden, P., 2007. Barriers and drivers in the adoption
of overall resistance to mobile-banking, followed by of current and future mobile services in Finland.
image and risk barriers respectively. These results Telematics and Informatics, 24 (2), 145-160.
suggest that functional usability and relative Brown, I., Cajee, Z., Davies, D., Stroebel, S., 2003. Cell
advantage compared to other ways of banking are phone banking: predictors of adoption in South Africa
currently the most powerful inhibitors of mobile - an exploratory study. International Journal of
banking adoption. Interestingly, tradition appeared Information Management, 23 (5), 381–394.
to be a negative determinant of resistance. Dunphy, S., Herbig, P.A., 1995. Acceptance of
Furthermore, Portuguese online bank customers innovations: The customer is the key! Journal of High
showed less resistance in terms of usage, value, risk Technology Management Research, 6 (2), 193–209.
and image to adopting mobile banking services than Eurostat, 2007. European business: Facts and figures,
Eurostat, Luxembourg.
their Finnish counterparts. However, Portuguese
Ellen, P.S., Bearden, W.O., Sharma, S., 1991. Resistance
online bank customers showed greater preference for to technological innovations: an examination of the
personal service, indicating more traditional banking role of self-efficacy and performance satisfaction.
behaviour compared to Finns. This idiosyncrasy of Journal of the Academy of Marketing Science, 19 (4),
the Portuguese could mean a high pre-disposition to 297-307.
adopt new service channels alongside with more Fain, D., Roberts, M.L., 1997. Technology vs. consumer
traditional ones. behavior: the battle for the financial services customer.
Compared to Finland, the relatively low Journal of Direct Marketing, 11 (1), 44-54.
resistance scores to mobile banking among the Gatignon, H., Robertson, T.S., 1991. Innovative decision
Portuguese may reflect the fact that Portugal has processes. In Robertson, T.S., Kassarjian, H.H. (Eds),
simultaneously a low number of internet-connected Handbook of Consumer Behavior (pp. 316-48).
computers and a high mobile penetration, a situation Prentice-Hall. Englewood Cliffs, NJ.
very auspicious for mobile services (Narinder, Hofstede, G., 1980. Culture’s Consequences, Sage
2007). Another explanation for such surprising result Publications, Thousand Oaks, CA.
might be related to the sampling method as only Im, I., Kim, Y., Han, H.-J., 2008. The effects of perceived
online banking users participated in the study. risk and technology type on users’ acceptance of
technologies. Information & Management, 45, 1-9.
Rogers (2003) argues that adopter categorisation is
Kivijärvi, M., Laukkanen, T. and Cruz, P., 2007.
based on innovativeness, i.e. the degree to which an Consumer trust in electronic service consumption: a
individual is relatively earlier in adopting new ideas cross cultural comparison between Finland and
than other members of a social system. The fact that Portugal. Journal of Euromarketing, 16 (3), 51-65.
only the Innovators and Early Adopters of the total Kuisma, T., Laukkanen, T., Hiltunen, M., 2007. Mapping
population in Portugal have so far adopted Internet the reasons for resistance to Internet banking: a means-
banking, and that in Finland the diffusion of the end approach. International Journal of Information
innovation has already reached the Late Majority, Management, 27 (2), 75-85.
may have resulted that, in general, the Portuguese Laukkanen, T., 2007a. Internet vs. mobile banking:
sample consisted of more innovative individuals comparing customer value perceptions. Business
than the Finnish sample. Process Management Journal, 13 (6), 788-797.
In general, innovativeness is related to Laukkanen, T., 2007b. Measuring mobile banking
demographics such as age. In our study the customers’ channel attribute preferences in service
Portuguese sample consisted of much younger consumption. International Journal of Mobile
respondents compared to the Finnish sample. Future Communications, 5 (2), 123-138.

305
ICE-B 2008 - International Conference on e-Business

Laukkanen, T., Lauronen, J., 2005. Consumer value

creation in mobile banking services. International
Journal of Mobile Communications, 3 (4), 325-338.
Lee, E-J., Lee, J., Eastwood, D., 2003. A two-step
estimation of consumer adoption of technology-based
service innovations. Journal of Consumer Affairs, 37
(2), 256-282.
Luarn, P., Lin, H-H., 2005. Toward an understanding of
the behavioral intention to use mobile banking.
Computers in Human Behavior, 21 (6), 873-891.
Marr, NE., Prendergast, GP., 1993. Consumer adoption of
self-service technologies in retail banking: is expert
opinion supported by consumer research. International
Journal of Bank Marketing, 11 (1), 3-10.
Narinder, C, 2007. Fast growth of Mobile
Communications in India: Lessons for emerging
markets. 6th Annual Global Mobility Roundtable, Los
Angeles, USA.
Ram, S. 1987. A model of innovation resistance. Advances
in Consumer Research, 14 (1), 208-212.
Ram, S., Sheth, JN., 1989. Consumer resistance to
innovations: the marketing problem and its solutions.
Journal of Consumer Marketing, 6 (2), 5-14.
Rogers, E., 2003. Diffusion of Innovations, Free Press.
New York, 5th edition.
Sheth, J.N., 1981. Psychology of innovation resistance: the
less developed concept (LDC) in diffusion research.
Research in Marketing, 4 (3), 273-82.
Siau, K., Shen, Z., 2003. Mobile communications and
mobile services. International Journal of Mobile
Communications, 1 (1/2), 3-14.
Statistics Finland, 2008. Information Society, Retrieved
April 13, 2008, from
http://www.stat.fi/tup/maanum/hakemisto_en.html

306
 BUSINESS AND TECHNICAL WORKFLOWS FOR E-BUSINESS
IN A VIRTUAL CLUSTER OF ISPS

Jane Hall and Klaus-Peter Eckert

Fraunhofer FOKUS, Kaiserin-Augusta-Allee 31, 10589 Berlin, Germany
[email protected], [email protected]

Keywords: Workflow technologies, e-Business, virtual organization, ISPs.

Abstract: Although workflow technologies have existed for some time, their true potential is only now being unlocked
with the emergence of Web services technology and XML in conjunction with expanding Internet use.
There has been extensive support for Web services and the standards have matured sufficiently for
widespread adoption, replacing proprietary standards for interfaces and data that hindered the integration
and automation of business processes, especially between partners wanting to cooperate in a virtual
organization. This paper discusses the application of workflow technologies in supporting the e-Business of
small Internet Service Providers (ISPs) collaborating in a virtual cluster. Not only business processes but
also innovative technical processes are being executed as workflows in the marketing, deployment and
operation of tailored ISP services. A scenario depicts the actual processes used and illustrates how the
software being developed supports e-Business for ISPs collaborating in a virtual cluster.

1 INTRODUCTION for ISP collaboration and in particular highlights an

innovative part of the work by showing how
An ISP here is a small firm providing Internet and technical processes that have been carried out
communication services to business users, manually by ISPs can be automated using workflow
themselves mainly small and medium enterprises technologies. Only by using such automated
(SMEs). e-Business is of direct concern to these processes can small ISPs hope to remain competitive
ISPs for in the volatile and challenging marketplace and agile in the marketplace of today and the use of
of today it is imperative for them to improve their workflow technologies here can demonstrate their
efficiency, to reduce their costs and to automate the potential in such applications.
selling and provisioning of the services they offer if The structure of this paper is as follows. First
they are to survive. For e-Business to be successful, some context is given to the rationale behind a
full advantage must be taken of the technologies virtual cluster of small ISPs. Then the application of
allowing open interoperability and supporting cross- workflow technologies to support the operation of
organizational cooperation. the cluster and the collaboration between cluster
This paper discusses the use of workflow partners in marketing and deploying services are
technologies to support a virtual cluster of small discussed. The workflow techologies selected and
ISPs in marketing and delivering services to the platform for modeling, specifying and executing
customers. The research has involved an them are presented. The business and technical
investigation into how workflow technologies could workflows developed for running on the platform
be used in this context, whether this was a feasible are examined. An e-Business scenario is outlined to
objective, the challenges and issues involved, and show how the VISP software infrastructure is being
how a workflow platform could be designed and used for marketing and deploying composite ISP
implemented to support the objectives. Many of the services. The work is being carried out within the
issues are generic and applicable to many areas, they European IST project VISP (IST-FP6-027178) and
are however discussed in this paper within the conclusions are drawn on the experience gained in
framework of small ISPs collaborating in a virtual the project of utilizing workflow technologies for
cluster. This paper illustrates some of the main ideas ISP e-Business.
and issues in developing a workflow support system

307
ICE-B 2008 - International Conference on e-Business

2 RATIONALE FOR A VIRTUAL 3 WORKFLOW TECHNOLOGIES

CLUSTER OF ISPS
Workflow has been defined as “the automation of a
The ISP market is changing rapidly, is quite volatile business process” (Workflow Management
and very competitive. A small ISP lacks financial Coalition, 1999). Ever since the existence of widely
resources and has few personnel, which means it available computing power, there have been
cannot cover all areas but does have expertise in numerous developments aimed at automating and
specialized areas. It offers a limited range of services streamlining the activities in a business process in
and so cannot cross-sell additional services to order to improve the efficiency of the organization,
existing customers. No small ISP on its own can both internally and vis-à-vis other organizations (van
provide the geographic coverage and the wide range der Aalst et al., 2003).
of specialized services that customers are now A major objective of the VISP project is to
requesting. develop an innovative software platform enabling a
Collaboration enables individual business entities cluster of small ISPs to collaborate and operate as a
to pool resources and competencies to provide single business entity using workflow technologies
value-added products and services (Camarinha- to support the dynamic implementation and
Matos, Afsarmanesh, 2006). The benefits of provisioning of tailored services. The provisioning
collaboration in virtual organizations generally are and operation of ISP services by cooperating cluster
relevant also for small ISPs (Kürümlüoglu et al., partners requires the introduction not only of
2005). A solution for a small ISP is therefore to form business but also of technical workflows in the
a virtual cluster with other small ISPs for e-Business cluster. However, little work has been undertaken on
purposes. In this way, a greater variety of services automating the activities in a technical process and it
can be offered as an ISP can integrate its own was one of the research topics to undertake work
services with those of its partners in the cluster, here so that technical processes could also be
extend its geographic coverage and become more automated.
visible in the market. As customers are requiring Workflow technologies are being adopted to
increasingly specialized services, often for a market provide the software infrastructure required by the
of one, tailored services can be offered by the cluster cluster when undertaking e-Business transactions
to its customers by composing new services from the both with customers and between the partners
different services offered by various cluster partners. themselves. Partners in the cluster can use
The services offered by ISPs require a long-term collaboration-based process models to describe their
relationship with the customer as they are provided cooperation. A workflow will specify how the
over months or years. Unlike other forms of virtual individual roles participate in an end-to-end process.
organization (Camarinha et al., 2005), a VISP These processes are modeled as workflows using
cluster is therefore intended to be a long-living formal languages in order to be deployed and
entity, although partners can join and leave. Two executed on distributed workflow engines.
operational modes are envisaged for the cluster. The There has been extensive support for Web
Community mode is where each partner owns its services and the standards have matured sufficiently
own customers and the cluster is not visible for widespread adoption, especially between
externally. The partner serves its customers and the partners wanting to cooperate in a value network
cluster is used dynamically as a pool of services for (Keen and McDonald, 2000). Integration both within
subcontracting. In the Virtual Enterprise mode it is the enterprise as well as between enterprises is not
the cluster that is visible to the customer and the only easier but also cheaper. These developments
cluster owns the customer relationship, the customer clearly have an impact on the VISP idea of
data and the customer transaction. When a customer cooperating roles in an SME cluster, which requires
makes an enquiry about a service in this mode it is technologies that automate processes across
sent to one partner, which communicates with the organization boundaries. This availability is a
customer on behalf of the cluster. The cluster is a significant element in making the VISP vision a
federation of independent partners collaborating in a reality.
decentralized manner to conduct e-Business. Each Work that had been undertaken in the area of
partner is a separate entity that joins in the workflows, especially between organizations, could
collaboration to the extent that it wishes. be used as a basis on for the work in VISP for the
ISP service domain. This included projects such as
CrossWork, which was concerned with cross-

308
 BUSINESS AND TECHNICAL WORKFLOWS FOR E-BUSINESS IN A VIRTUAL CLUSTER OF ISPS

organizational processes in the automotive industry well as to the emerging standards for human
(http://www.crosswork.info); CrossFlow, which interaction, BPEL4People and WS-HumanTask.
investigated workflows in dynamically formed Additionally the submission introduces a
virtual organizations (http://www.crossflow.org); standardized XML-based exchange format for
and Astro, which has been developing tools to BPMN models, which was lacking in the previous
support distributed processes during their lifecycle version.
(http://astroproject.org). During the lifetime of the VISP project, most
A top-down approach is being adopted using vendors of BPEL development and execution tools
workflow technologies. The high-level process can have migrated from BPEL 1.1 to WS-BPEL 2.0.
be recursively decomposed into sub-processes until Because the expressiveness of WS-BPEL and the
the required level of detail is obtained for execution number of built-in functions increased, it became an
on a workflow engine. Various modeling languages option to use the new version of the language and
were available and an investigation was undertaken supporting tools. In this situation, the mapping chain
to select those most appropriate for the VISP work has to be extended from BPMN to BPEL 1.1 and
(Eckert et al., 2006). BPMN was adopted as a then further to WS-BPEL. Theoretically both
modeling language for the high-level business and mappings cannot be fully performed automatically.
technical process modeling (BPMN, 2006). It is a A mapping from BPMN to BPEL is only possible
standard containing a standardized mapping to for a subset of BPMN models, thus not every valid
BPEL4WS, or BPEL for short (Andrews et al., BPMN model is “BPEL-valid”. Additionally not
2003), and it was felt that its graphical notation is every BPEL process can be mapped to a WS-BPEL
more intuitive for non-IT specialists acquainted with process. Fortunately the selected tool supports a
the graphical notation of traditional business best-effort mapping that works for most
flowcharting notations. Despite the weaknesses and specifications and requires only minor manual
deficiencies of BPMN (Wohed et al., 2006), it was improvements.
available on the market in tools complying with the
standard at the time of language and tool selection 3.1 VISP Workflow Platform
for VISP. The ‘ideal’ solutions were not available
and a pragmatic decision was to select a technology As the VISP project is basing its software
that seemed to have potential and was suitable for infrastructure on workflow technologies, the
VISP’s purposes. software platform that it is developing will allow the
The BPMN flows are at a high level and may be cluster to specify, model, deploy and execute
decomposed to several levels of detail but they workflows that support the operation of the cluster
remain in BPMN. Mappings from BPMN to BPEL and the provision of tailored services to customers.
according to the standard are then undertaken and so This software platform consists of two major parts.
tools are required that not only support the The Workflow Modelling and Specification
specification work in a particular language but that Platform (WfMSP) is designed to support service
can map to another language and/or import and and workflow specification, modeling, choreography
export such languages. BPEL also has its strengths and orchestration. The Workflow Execution
and weaknesses (Wohed et al., 2003), but is in Platform (WfEP) executes and controls the
widespread use for implementing workflows. It was workflows (see Figure 1).
selected as it met the VISP requirements in the area Informal textual descriptions provided by
of workflow technologies and also because the only domain experts and following Cockburn’s style of
standardized mapping was from BPMN to BPEL. writing use cases (Cockburn, 2001) are the primary
The tools selected for these languages were also input to the WfMSP. They have to be formalized,
investigated and selected on the basis of their first as BPMN models, then as abstract BPEL
suitability for VISP aims in the first phase of the skeletons and emerge as an executable workflow
project as well as their adherence to the standard that can be deployed on the WfEP. The workflow
specifications of BPMN and BPEL. engines are the coordinating point of the WfEP that
In the last year both standards used in VISP have are responsible for executing and controlling the
been significantly improved. OASIS published workflows specified by the WfMSP. The WfEP
version 2 of WS-BPEL (BPEL, 2007) and several interfaces either directly or through mediation
devices with partners’ ERPs and with network and
big IT companies proposed an initial submission of
system components.
BPMN 2.0 (BPMN, 2008) to OMG. BPMN 2.0 will
comprise standardized mappings to WS-BPEL as

309
ICE-B 2008 - International Conference on e-Business

Figure 1: The VISP workflow platform.

3.2 VISP Workflows The technical processes are categorized in two

groups, the Administrative Technical Processes
Workflows are being developed in the VISP project (ATPs) and the Toolbox Technical Processes
to support the provisioning and operation of services (TTPs). The ATPs perform all operations required to
offered by the partners in a VISP cluster. Processes support a single ISP service instance lifecycle and to
are being specified, modelled, and executed as interface them to the corresponding business
workflows of activities in both the business and the processes. They administer an instance of an ISP
technical domains (see Figure 2). service and are invoked by the business processes.
A set of business workflows is being developed They can be defined and used without the VISP
within the project able to deal with business system and an adapter to the VISP system has been
processes in a dynamic cluster of partners and developed that maps and forwards all the necessary
therefore providing for multilateral relationships and information stored in the VISP repositories and
not only for binary relationships. These use standard workflow engines to the ATPs and TTPs.
processes where available, such as those from the The TTPs act directly on low-level Web services
OAGIS specifications together with their associated found in the mediation servers of the various
WSDLs (OAGIS, 2008) and are also based on the network elements. They usually contain just one
use cases produced as part of the requirements work atomic operation to be performed on one low-level
undertaken within the project. Web service. They are invoked by the ATPs and are
The technical processes in VISP are those that specific to a service instance type.
interact directly with network elements. Technical The relationship between the business processes,
processes cover all technical activities related to the ATPs and TTPs can be seen in the service instance
lifecycle of an ISP service in order to instantiate, lifecycle where business processes are responsible
commission, activate, deactivate and decommission for handling generic operations on composed
the service. Further administrative activities such as services, or service sets, for example, “instantiate a
testing, technical location transfer, suspend and service set”. Business processes first invoke the
resume are also being included. The project is trading mechanism and determine the cluster partner
providing formalized workflow specifications of that is best suited to provide the “Simple Call
technical processes that are currently manually Service” VoIP. Then they invoke ATPs to perform
executed in order to be able to process them operations on each of the service instances in the
automatically in a standardized way, something that service set, in this case “instantiate Simple Call
has been possible for business processes but not so Service”. Each ATP can call directly one or more
far for technical processes. TTPs to carry out atomic administrative operations
on service instances.

310
 BUSINESS AND TECHNICAL WORKFLOWS FOR E-BUSINESS IN A VIRTUAL CLUSTER OF ISPS

Figure 2: Business and technical processes.

Figure 3: Overview of the VISP components.

In this example, the ATP invokes the TTP that no longer have their scarce human resources tied up
initializes the new user account in the network in time-consuming tasks. The VISP project is
infrastructure. The TTPs help to isolate the upper therefore taking the technical knowledge of network
layers of processes from any changes to the WSDL engineers, preparing textual specifications of the
description as only the TTPs using the specified steps involved in their work, modeling these steps in
WSDL file need to be changed and not the ATPs BPMN, mapping this to abstract BPEL, refining to
themselves (see Figure 3). The parameters needed to executable BPEL and then running on the partners’
invoke the TTPs are either stored in the VISP workflow engines.
repositories or can be retrieved from the ISP’s Technical processes have not been standardized
employees, for example from network engineers, and so this work constitutes an innovative aspect of
utilizing the VISP internal worklist management the project. The intention is to automate where
system. possible not only the business but also the technical
The technical processes are intended to automate aspects of provisioning and delivering ISP services.
as much as possible of what network engineers have This can save scarce human expertise for more
been undertaking manually to date. The need to be complex tasks and thus enable ISPs to be more
efficient and competitive means that small ISPs can competitive and efficient in the marketplace.

311
ICE-B 2008 - International Conference on e-Business

4 REALISING E-BUSINESS WITH If a cluster partner is not able to provide a service of

WORKFLOW TECHNOLOGIES the service set, an appropriate cluster member is
selected by means of trading mechanisms performed
This section introduces the implementation of the according to a particular economic model. Trading is
VISP software infrastructure and the specific issues thus part of the service instantiation process and has
encountered. Implementation work is being carried a business (quote, contract) as well as a technical
out incrementally, with the first release concerned (instantiation) result. Details of the instantiated
with how a simple VoIP service can be sold to a service set are stored in the Service Instance Base,
customer and how a VoIP system already accessible which stores all trading and deployment information
through Web services can be controlled from the concerning the service instances comprising the
VISP software infrastructure. service set. The technical part of the offer can be
The e-Business scenario comprises the sale and transferred to an ERP for a full offer preparation, if
provisioning of a VoIP service for a customer and it applicable. This can include all contractual terms
demonstrates how the various parts of the software and conditions, billing and payment details and SLA
infrastructure are utilized to support this. A information.
The offer is then made to the customer and
prospective customer enquires about the ‘simple call
further negotiations may ensue. If the offer is
service’ VoIP via a cluster communication channel
accepted, the sales representative starts the
and a sales representative from one of the partners is commissioning of the service set. Commissioning of
selected to process the customer’s request on behalf the service set has to be performed according to the
of the cluster. The sales representative checks the requirements of the offer and in particular its timing.
customer in the cluster’s Consumer Directory and if The resources previously reserved are allocated to
not already in the repository, registers the customer the service set and once commissioning has been
there. This requires infrastructure support to search carried out, the service set can be activated on the
the Consumer Directory and, if necessary, to update date agreed with the customer. Commissioning and
it with the prospective customer’s details. Business activation of the service set are carried out
workflows have been implemented with the automatically with the technical workflows
appropriate GUIs to do this. developed in VISP.
In order to meet the customer request, the sales A request for service termination would typically
representative checks the VISP cluster Service imply three steps, deactivation, decommissioning
Knowledge Base (SKB) for possible solutions and and deinstantiation of the service instance so that its
combinations of services. The SKB contains status becomes ‘historic’. Again, the technical
technical information about the services in the workflows in the infrastructure carry this out.
cluster and workflows have been developed to add, An example of a centralized configuration of the
delete or modify a service as well as to browse the VISP infrastructure for a two partner virtual
SKB. The Market Directory can also be searched to enterprise mode is depicted in Figure 4. This
ascertain which partners are offering which services configuration shows two partners A and B with their
and the availability of these services. Once the sales representatives and administrators. All actors
component services required to support the VoIP are using partner-specific front ends that are
service for the customer have been selected, the connected to a common GUI server running the user
sales representative groups the services provided by interface to the VISP application that is written in
cluster partners into a candidate service set. A Orbeon, an open source forms solution using
service set is the result of the composition of the XForms and Ajax technology. The Orbeon server
services into one group and enables the selection and can access the VISP repository services running as
combination of services, characteristics and values Web services on the repository server. Depending on
to be validated. The sales representative stores the the mode of the VISP cluster, either partner-specific
definition of the service set in a personal Service Set repositories and/or global repositories can be
Catalogue and validates it. queried. Additionally the Orbeon server invokes the
The sales representative uses the VISP business workflows running on an activeBPEL
workflows to instantiate the service set. Instantiation server as Web services.
is a means of reserving resources so that if the These workflows invoke ATPs and TTPs
customer accepts the offer, the service can be running in the same engine via their WSDL
provided to the customer. Instantiation is therefore interfaces. The last component shown in Figure 4 is
based on the confirmation of resource reservations the mediation server (MS_1) to the VoIP technical
from partners contributing services to the service set.

312
 BUSINESS AND TECHNICAL WORKFLOWS FOR E-BUSINESS IN A VIRTUAL CLUSTER OF ISPS

Figure 4: The configuration of the VISP infrastructure.

infrastructure that can be accessed by VISP TTPs common conventions to ensure their interoperability
and by the ISP network operators via a proprietary and reusability. This comprises agreements on
administration interface (Jacinto). In operational namespaces, message definitions, and binding styles.
environments, the repository servers and workflow The WS-I Basic Profile was therefore adopted
engines will be replicated and distributed to the throughout the project (WSIO, 2004).
organizational domains of the cluster partners. A big challenge in VISP is to interface generic
business processes with dynamic technical
processes. During the design time of the VISP
application it is known that for each ISP service,
5 CONCLUSIONS corresponding ATPs with specific WSDLs have to
be invoked. Unfortunately these WSDLs are
The research work of the VISP project has unknown during design time. Thus a dynamic
investigated the feasibility of using workflow invocation mechanism has been developed that
technologies in an e-Business environment of ISPs. generates corresponding SOAP messages from
This paper has shown how workflows can support e- information stored in the VISP repositories using
Business in a virtual cluster of small ISPs XSLT transformations and retrieving endpoint
collaborating to offer and deliver services to their addresses from the VISP UDDI. Such a dynamic
customers. A platform is being developed within the invocation concept can be reused in every dynamic
VISP project to model, specify and execute the SOA environment.
business and technical workflows that have been The work in the VISP project is ongoing.
realized. In assessing the research work, the Additional ISP services are being made available for
following points can be made. offer and the trading models used are to be extended.
Experiences from the work in the project show More work is also required on ERP integration via
that it is a challenge to extract the knowledge of OAGIS 9.x WSDL interfaces, and the configuration
experts in a form that is amenable to automation. of the software needs to be based on a distributed
The modeling and specification of textual architecture to replace the centralized one currently
specifications based on the knowledge of network adopted.
engineers therefore has to undergo several Workflows are an interesting and innovative
reiterations. technology for supporting e-Business in the ISP
Also, although there is a standardized mapping domain and the marketing and provisioning of ISP
from BPMN to BPEL, the fact that BPMN is a services. Such technologies are being used
graph-oriented language and BPEL a block- increasingly for both intra- and inter-enterprise
structured language has meant that conventions had cooperation and ISPs, with their knowledge of
to be established that were based on experience Internet and Web technologies, are well-placed to
within the project as well as on work undertaken use these technologies themselves. The research and
elsewhere (Ouyang, 2008). When using Web service development work within the VISP project has
technology it was also found necessary to agree on shown has shown that an infrastructure based on

313
ICE-B 2008 - International Conference on e-Business

workflow technology is feasible and can provide an Applications, Case Studies, P. Cunningham and M.
effective automated environment allowing small Cunningham, ed. IOS Press, Amsterdam, pp.1631-
ISPs to cooperate in offering and delivering services 1638.
to customers. In addition, several technical solutions Keen, P., McDonald, M., 2000. The eProcess Edge:
developed in the project can be transferred to SOA Creating Customer Value and Business Wealth in the
implementations in different application domains. Internet Era, Osborne/McGraw-Hill, Berkeley, CA.
Kürümlüoglu, M., Nøstdal, R., Karvonen, I., 2005, “Base
Despite the challenges encountered in
Concepts”. In Virtual Organizations: Systems and
implementing the workflow technologies for VISP
Practices, ed. L.M. Camarinha-Matos, H.
objectives, an analysis of the project results to date Afsarmanesh andM. Ollus, Springer Science+Business
show that workflow technologies have potential and Media, New York, pp.11-28.
can be used to support a virtual cluster of small ISPs. OAGIS (Open Application Group Integration
The VISP application is therefore being prototyped Specification), 2008. Business Object Documents,
and will be further developed for use in a http://www.oagi.org/downloads/oagis/oagis92_registra
commercial environment. tion.htm.
Ouyang, C., Dumas, M., ter Hofstede, A.H.M., van der
Aalst, W.M.P., 2008. “Pattern-based Translation of
BPMN Process Models to BPEL Web Services”,
ACKNOWLEDGEMENTS International Journal of Web Services Research, 5(1),
pp.42-62.
This work was carried out with partial funding from van der Aalst, W.M.P., ter Hofstede, A.H.M., Weske, M.,
the EU through the IST project VISP (IST-FP6- 2003. “Business Process Management: A Survey”. In
027178). Further information on VISP can be Business Process Management International
obtained from http://www.visp-project.org. Conference, BPM 2003, Eindhoven, The Netherlands,
June 26-27, 2003, Proceedings, ed. W.M.P. van der
Aalst, A.H.M. ter Hofstede, and M. Weske, LNCS
2678, pp.1-12, Springer-Verlag, Berlin.
REFERENCES Wohed, P., van der Aalst, W.M.P., Dumas, M, ter
Hofstede, A.H.M., 2003, “Analysis of Web Services
Andrews, T., et al., 2003. Business Process Execution Composition Languages: The Case of BPEL4WS”. In
Language for Web Services, version 1.1, Conceptual Modeling – ER 2003. 22nd International
http://download.boulder.ibm.com/ibmdl/pub/software/ Conference on Conceptual Modleing, Chicage, IL,
dw/specs/ws-bpel/ws-bpel.pdf. USA, October 13-16, 2003, Proceedings, ed. I-Y
BPEL, 2007. Web Services Business Process Execution Song, S.W. Liddle, T.W. Ling and P Scheuermann,
Language, version 2.0, http://docs.oasis- LNCS 2813, Springer-Verlag, Berlin, pp.200-215.
open.org/wsbpel/2.0/OS/wsbpel-v2.0-OS.html. Wohed, P., van der Aalst, W.M.P., Dumas, M, ter
BPMN, 2006. Business Process Modeling Notation Hofstede, A.H.M., Russell, N., 2006, “On the
(BPMN) Specification, OMG Document dtc/06-02-01, Suitability of BPMN for Business Process Modelling”.
Final Adopted Specification. In 4th International Conference on Business Process
BPMN, 2008. Business Process Modeling Notation Management (BPM 2006),5-7 September 2006,
(BPMN) Specification 2.0, OMG Document bmi/2008- Vienna, Austria, ed. S. Dustdar, J.L. Faideiro and A.
02-06, Initial submission. Sheth, LNCS 4102, Springer-Verlag, Berlin, pp.161-
Camarinha, L.M., Silveri, I., Afsarmanesh, H., Oliveira, 176.
A.I., 2005. “Towards a Framework for Creation of Workflow Management Coalition, 1999. Terminology &
Dynamic Virtual Organizations”. In Collaborative Glossary, Document WFMC-TC-1011, Issue 3.0,
Networks and Their Breeding Environments (PRO- http://www.wfmc.org/standards/docs/TC-
VE’05), 26-28 September 2005, Valencia, Spain, ed. 1011_term_glossary_v3.pdf.
L.M. Camarinha-Matos, H. Afsarmanesh and A. Ortiz, WSIO, 2004. Web Services Interoperability Organization,
Springer Science+Business Media, New York, pp.69- Basic Profile Version 1.0, http://www.ws-
80. i.org/Profiles/BasicProfile-1.0-2004-04-
Camarinha-Matos, L.M., Afsarmanesh, H., 2006. 16.html#R2110.
“Collaborative Networks: Value creation in a
knowledge society”. In Proceedings of
PROLAMAT’06, Shanghai, China, 14-16 June 2006,
Springer, New York.
Cockburn, A., 2001. Writing Effective Use Cases,
Addison-Wesley, Boston, MA.
Eckert, K-P., Glickman, Y., Hall, J., et al., 2006.
“Workflow Technologies for a Virtual ISP”. In
Exploiting the Knowledge Economy: Issues,

314
 FACTORS AFFECTING THE USAGE OF T-GOVERNMENT
SERVICES
An Exploratory Study

Michele Cornacchia, Filomena Papa, Stefano Livi1, Bartolomeo Sapio

Enrico Nicolò and Gaetano Bruno
Fondazione Ugo Bordoni, Via Baldassarre Castiglione, 59 - 00142, Rome, Italy
[email protected], [email protected], [email protected], [email protected], [email protected]
1
University of Rome "La Sapienza", Faculty of Psychology 2, Via dei Marsi, 78 – 00185, Rome, Italy
[email protected]

Keywords: Adoption of technology, interactive public services, digital terrestrial television (DTT), end-users
evaluation, field study, home equipment, ICT influence on home life, predicting model, questionnaire,
research methodology, residential users, T-government, usability, usage, use behaviour.

Abstract: Developing humanly acceptable information and interactive systems is a complex process where testing
from the outset, the setting of operational criteria for efficiency, effectiveness and satisfaction, the
development of prototyping etc., has to be considered as the central activity of a field implementation.
Design and operative alternatives can be effectively selected and long-term targets can be seen to be
attainable. However, sometimes it happens, especially in the domain of information system implementation,
that obtained datasets can be effectively utilised to reach more tailored purposes. The case of interactive T-
government services at the Municipality of Parma is analysed aiming to redesign the analysis model about
the factors affecting the usage. To this respect one of the most recognized ICT predicting models in the
literature is utilized. Final results show the success of the methodology and encourage further developments
and applications in contiguous areas of ICT designed for residential population.

1 INTRODUCTION Hence there can be different orientations, even with

respect to a similar target of an investigation, which
The following study has been designed to be an can carry to a wide range of final resultant datasets,
exploratory study into aspects of T-government i.e. different for study typology, for structure, for
services usage among residential users. It aims to procedures to submit/collect data, for user profile,
generate new groups of variables from a pre-existing for task design, for boundary conditions, and so on.
dataset which can address the analysts towards T-government stands for a wide set of services
different and reasonably more tailored purposes. The addressed to citizens (e.g. about health, education,
method and the first results here presented are part tourism, bills payment), delivered by a Public
of a larger work still in progress, moreover having Administration and accessible by Digital Television.
the objective to investigate the usability and The piloting research program on DTT (Digital
economic factors affecting adoption and usage of T- Terrestrial Television) in Italy has been developed
government services. through a number of initiatives (Section 2.1), some
With reference to the emerging ICT usage, it of them issued to explore the potential impact on
may happen, especially in the domain of information population of the new digital services associated
system implementation, that achieved datasets, even with the broadcasted TV channels. In that field of
rich and widely aimed to support the design of investigation, T-government interactive services
technologies compatible with users’ abilities and were introduced to be early tested with a real sample
needs, cannot be applicable to more complex and of residential private adopters of the digital TV
different kinds of analyses. Reasons can have decoder. A large amount of data were processed to
different justifications, nevertheless it is usual that investigate how residential end-users reacted to the
different theoretical positions do not all share the usage of T-government services and they perceived
same views even in the same research domain. their added value. Specifically, the results pointed

315
ICE-B 2008 - International Conference on e-Business

out how, in some cases, the usage of the services government projects promoted by Centro Nazionale
was to some extent low assessed. A significant per l’Informatica nella Pubblica Amministrazione
difference in the usage was also found between (CNIPA) and Fondazione Ugo Bordoni (FUB).
informative and interactive services.
These findings have been the input for the 2.1 T-government Field Investigations
current investigation about the factors affecting the Developed in Italy
usage. The study was carried out by comparing the
UTAUT model (Unified Theory of Acceptance and 2.1.1 CNIPA Projects
Use of Technology, Venkatesh et al. 2003) with the
real data of the basic questionnaire, the one applied In Italy, in 2004 CNIPA allocated 7 million euro in
to the Italian T-government project named “Parma order to co-fund T-government projects based on
Municipality Services” (Section 2.2), in order to DTT and addressed to the general public. As in the
redesign the analysis model. case of the advanced applications co-funded by FUB
The UTAUT model (Section 3) is a proved, (3 million euro), such a competitive call came from
reliable representation from social psychology, a joint initiative of the Minister for Communications
concerned with the determinants of acceptance and and the Minister for Innovation and Technology.
usage behaviour of new adopters of emerging The experimentation carried out by CNIPA is
information technology. It was applied as reference mainly oriented to transfer on the DTT platform
to cluster and re-code the relevant variables from the existing e-government services. Twenty-nine
pre-existing dataset into new groups of indicators informative applications were selected and co-
(Section 3.1). The indicators thus generated, were funded, with a total amount of investments
those of T-government, concerning respectively exceeding 32.5 million euros. Correspondingly, 13
interactive services and informative contents. At last regions, 25 provinces, 164 municipalities and 15
they were tested both for significance and construct mountain communities along with 40 local and
power to check if they effectively meet the reference national broadcasting corporations were involved.
model basic requirements (Section 4), so verifying These funded projects aimed at the experimental
that structural differences may be sometimes implementation of several services for the citizens.
overcome to start more complex analyses. The services considered within projects can be
grouped into five classes:
ƒ social services: health care, social assistance and
2 T-GOVERNMENT SERVICES disability, third age, jobs and occupations;
ƒ environment and tourism: environment, tourism,
T-government services are herewith categorised as traffic congestion and cultural events;
“informative services” (e.g. provision of general ƒ education: nursery schools, kindergartens,
information) and “interactive services” (e.g. schools and universities;
searching for specific information, e-mail, chat, form ƒ relationships with Public Administration:
filling and form sending, payments). institutional activities;
Most of T-government services that have been so ƒ “Agorà”: citizens and elections.
far developed are informative services. Within several projects, interactive services
One of the opportunity given by T-government is exploiting back channel (reverse/return channel) and
to promote the use of ICT-based public services by smart cards have been experimented.
large groups of people (e.g. the elderly), who have As to the users involved, particular attention has
no Internet access or the required skills of using. For been devoted to people having limited opportunities
those people, the past experience with TV and of accessing the Internet, users with disabilities, old
remote control may be a key qualification to become people and people looking for their first job.
effective users of the above services.
On the other hand some studies (Damodaran,
2.1.2 FUB Projects
2002) make evident the gap of knowledge about the One of the objectives of the six T-government
human aspects of T-government services: usage and projects co-funded by Fondazione Ugo Bordoni
usability, user satisfaction, the capability of services (FUB) was to experiment high interactivity T-
to reach all citizens and the capability to include government services, realised by Digital Terrestrial
them in the benefits of information society. Television (DTT), involving real users.
On these topics some field investigations have The high interactivity T-government services
been developed in Italy in the framework of the T- provide the user with some of the following features:

316
 FACTORS AFFECTING THE USAGE OF T-GOVERNMENT SERVICES - An Exploratory Study

ƒ remote interactivity using the return channel; could be related also to the fact that some
ƒ high performance return channel (e.g. broadband services with “high interactivity” are mainly
or wireless); referring to periodical or occasional activities
ƒ user authentication using a smart card (e.g. (e.g. bill payment for phone etc, payment of
electronic identity card, services regional card or fines, change of family doctor) which cannot
other kind of smart cards); justify a continuous use in the short time period
ƒ on line payments. of the field experiments (two-three months on
The experimented services belong to different average). However, the reasons of low usage
application areas: demographics, utilities and fines, should be investigated in depth.
education, T-health, T-learning, employment, T- ƒ In general, technical problems were not
producing relevant perception of disturbances
commerce, T-banking.
and troubles and the users easily handled the
In order to investigate usage, usability and socio-
provided equipment. Overall the users were quite
economical aspects of T-government services, an
satisfied with T-government services in
explorative field investigation was developed in
comparison to different channels to perform the
each of the six projects.
same task as the Internet and the traditional
The field investigations were realised using a
office desk. However, in some projects, technical
common framework referring to human factors
problems produced relevant perception of
discipline (Papa & Spedaletti, 2001). The following
disturbances and troubles mainly due to the lack
main usability aspects related to interactive services
of the digital television signal and to the
were identified (Papa & Spedaletti, 2004): perceived
interruption of the connection with the return
usefulness, perceived ease of use and attractiveness,
channel. Other problems were connected with
training and user support (human support, user
current input devices of the digital television, for
manual, support provided by DTT, call center), user
instance user difficulties in the input of
perception of technical disturbances and troubles
alphanumeric data by the remote control or by
(due to television signal, set top box, return
the virtual keyboard.
channel), security and privacy perception
ƒ Services developed in three of the six projects
(confidentiality of personal data, security of
are still “on air” after the end of the project.
payments), impact of the equipment in the house,
These services have the common feature of
users’ satisfaction about the service including the
being provided in the context of a public
comparison of different channels to perform the
administration (municipality, regional
same task (e.g. DTT versus Internet, DTT versus
administration, public schools).
traditional office desk).
ƒ One of the initial objectives of the projects was
Service usage is related to whether a given
to promote the use of Public Administration
service is used or not and to the service utilisation
services by groups of people who have not the
level. Service usage is evaluated collecting
needed skills to use a computer and the Internet
subjective data provided by the user. The adopted
but are familiar with television and with the
indicators for service utilisation level are (Davis,
remote control. To this respect this objective was
1993): frequency of use, time duration of the
only partially reached since the investigation has
session, kind of use (shallow or intensive). Socio-
shown that the most intensive users of T-
economical aspects included: user profile (including
government services were people using a PC and
income and social network information), TV and
the Internet at home.
Internet usage (including other entertainment
technologies), and scenarios (including interest As mentioned before, the reasons of the low
levels, willingness to pay for equipment and utilisation of interactive services need to be
services, decision factors). investigated more in depth. In particular, the
identification of factors affecting usage of
The main results of the field investigations realised
interactive T-government services could provide
in the FUB projects can be summarised as follows:
ƒ The role of the information and communication useful elements to increase their usage in the future.
campaign is very important in ensuring a high
adhesion to the field experiment by the users of 2.1.3 The Field Investigation Developed in
the panels and, in some projects, in increasing the Municipality of Parma
the service usage.
ƒ In general, the usage of T-government services FUB provided the six projects with a common
has been quite low. A relevant difference in framework and some general tools:
usage was found between informative services ƒ guidelines to evaluate usability of T-government
and interactive services. A possible explanation services with high interactivity;

317
ICE-B 2008 - International Conference on e-Business

ƒ a first questionnaire for the users, about usability ƒ demographic characteristics to reflect the main
and usage aspects; features of citizens in the Municipality of Parma
ƒ a second questionnaire for the users, concerning (in terms of age, gender, and city district);
socio-economical aspects. ƒ potential interest toward the interactive services
Each question or group of questions had the aim under experimentation (holders of driving
to evaluate indicators related to usability aspects, licence and electronic identity card).
socio-economical aspects and usage. Within the It was decided to choose the sample age ranging
above common framework, each project utilised from 20 to 45 years, people in general skilled in the
different tools and techniques for data collection, use of new information technology. The 200
adapting the above general tools to the particular citizens were randomly selected in this age range.
context of the project. Each project was also Finally, 181 citizens (88 males and 93 females) took
responsible for data collection. In the following, the part at the experiment for the all planned period of
field experiment carried out in the Project “Parma two months and a half.
Municipality Services” is described.
Procedure. In the first phase, the T-government
T-government Services Implemented. The project applications were implemented and broadcasted in
has developed both informative and interactive the territory of Parma Municipality.
services. The former provide information about: In the second phase a randomly chosen users
Parma Municipality organisation; services offered to panel of 200 citizens was selected with the above
the citizens by the Municipality; cultural initiatives mentioned criteria. The recruitment officially started
in progress in the city. when the municipality sent a letter to the users’
The interactive services allow the user (after panel explaining objectives and modalities of the
authentication through Electronic Identity Card): trial. The users were then invited to contact the call
ƒ to pay fines using a credit card; center to choose between self-installation and
ƒ to visualise the state of a demographic dossier personnel assisted installation of the decoder.
(e.g. for changing place of abode); In case of self-installation, users were asked to
ƒ to visualise the state of a request for education pick up by themselves the Set-Top-Box (STB) at the
services provided by the municipality (e.g. Municipality offices. A user guide was also released
nursery school). to aid home installation and services use. In case of
A screen shot is presented in Figure 1. assisted installation, a technician brought the STB to
the people residence and there trained the users.
Tools and Techniques for Data Collection. Data
were collected mainly using a paper-based
questionnaire filled in by the users. The
questionnaire was both delivered and collected by
Municipality of Parma personnel.
Results. The first results of the field investigation
can be summarised in the following points:
ƒ The usage of interactive services has been quite
low. In particular only 20% of the users
answering to the questionnaire completely used
the interactive services. 35% used the interactive
services in a shallow way. A possible
Figure 1: Data input to select a demographic dossier. explanation could be related to the fact that the
services are referring to occasional activities
During the field experiment a call center was (e.g. payment of fines).
available to the users for any information and help in ƒ Technical problems were not producing relevant
the utilisation of the services. perception of disturbances and troubles and the
users easily handled complex equipment. Overall
The Users Panel. A sample of 200 people was the users were quite satisfied with T-government
selected from a group of 4.000 citizens of Parma services in comparison to different channels to
Municipality. The selection followed two main perform the same task, as the Internet and
criteria: traditional office desk. However, the sample
involved in the field study was composed by a

318
 FACTORS AFFECTING THE USAGE OF T-GOVERNMENT SERVICES - An Exploratory Study

large part of young people using ICTs. Of course presented with new technological devices, a number
it is not possible to generalise such indications to of factors influence their decision about how and
the whole Parma population and, even less, to when they will use them, notably two main key sets
the whole Italian population. of constructs: Perceived Usefulness and Perceived
Ease of Use (Davis, 1989).
Perceived Usefulness is defined as the user’s
3 THE FACTORS OF USAGE “subjective probability that using a specific
application system will increase his/her job
Since many years the study/assessment of usage of performance within an organisational context”.
emerging ICTs requires to be carried out through the Perceived Ease of Use refers to “the degree to which
analysis of more and more reliable models (Taylor & the users expects the target system to be free of
Todd, 1995). efforts”. Both complex constructs predict the attitude
toward using the system, in that order defined as
3.1 Theoretical Background “the users desirability of using the system”.
There can be also strong behavioural elements
In this paper we assume that the main forecasting with which the models assume that when someone is
feature of the most recognized models in the forming an intention to act, then he/she is free to act
literature, i.e. the “Technology Acceptance Models” without limitations. UTAUT extends TAM by
(TAM) (Malhotra & Galletta, 1999) and UTAUT introducing the terms of social influence and
(Venkatesh et al., 2003) in particular, could be cognitive instrumental processes and, above all, it
applied to analyse other experimental datasets, built ultimately unifies the main competing user
up by different criteria other than, of course, to be acceptance models, namely eight theoretical
suitable with ICT contexts of use. This hypothesis, approaches sharing the same basic concepts. As
whether verified, should expand the possibilities to shown in Figure2, the key dependent variable is the
analyse the usage constructs of the original sets of intention and/or the use of technology, thus the final
variables and as well as to come to more satisfying purpose of the model is that of understanding and
results. In fact, it is very difficult to attain data due providing explanation of the phenomenon “use of
to the high complexity and variability especially information technology”, taken as dependent
about socio-technical systems (Dillon, 2000) and/or variable. The role of intention, as an antecedent of
when technology is being oversold (Cornacchia, behaviour (i.e. of use), is the actual critical factor of
2003). The relevant literature on this topic describes the model and was analysed and proved in literature.
the development of several models of technology
acceptance by the users and many extensions to the
basic constructs (Malhotra & Galletta, 1999;
Venkatesh & Davis, 2000), mostly built with the
behavioural elements (Ajzen, 1996) of who is
forming an intention to act (Bandura, 1986) and the Figure 2: Basic concept underlying user acceptance
inclusion of some kinds of constraints (limited models.
ability, learning and usage (Bagozzi et al, 1992),
time, environmental, organisational, unconscious In the formulation of the unified model four
habits, and so on) which influence the individuals’ constructs were identified which may play a
actions (Compeau et al, 1999; Pierro et al, 2003). significant role as direct determinants of the user
This study thus aims to give a methodological acceptance and usage behaviour: Performance
answer in measuring the effects of re-coding the Expectancy, Effort Expectancy, Social Influence and
experimental T-government variables dataset, the Facilitating Conditions. In addition, Age, Gender,
one described in the previous paragraph. The Experience and Voluntariness of Use were the
procedure is carried out by matching the T- moderators of the model.
government simple sets of variables with/into new
ones of the usage and technology acceptance of 3.2 Recoding T-government Variables
TAM and UTAUT.
Actually the TAM and UTAUT both apply to the In the current exploratory study all direct
adoption and use processes of emerging ICT in order determinants of UTAUT constructs were considered
to understand the human choices and technology in the application with experimental T-government
acceptance. The models suggest that when users are dataset. To this extent, the questionnaire items were

319
ICE-B 2008 - International Conference on e-Business

recoded into aspects dealing with usage of “STB”, ƒ Manual for service use
“informative contents”, “interactive contents” and ƒ Information on the cost of connection
others. During the matching and re-coding process, ƒ Call center and contact center
the following four step scheme was applied: Moderators
1. set up the reference sets of complex constructs ƒ Gender
from UTAUT; ƒ Age
2. search for simple items in the questionnaire that ƒ Income and education
can replicate the causal structure of the unified ƒ Experience (skills in using TV, Internet and
model UTAUT, according to its requirements entertainment appliances)
and basic conditions; Arranged by the above recoded factors, since
3. re-code by grouping them into new sets; this was an exploratory study, dependent variables
4. test statistically both for significance and considered in testing the significance of the new
meaningfulness of the new constructs of causal constructs (step 4 of the procedure) were:
variables. ƒ cont_utente_servizi (STB Use)
Following steps 2 and 3 in the above mentioned ƒ cont_interattivi (Interactive Service Use)
scheme, factors of UTAUT model were replicated ƒ cont_informativi (Informative Service Use).
using the indicators adequately selected from the T- Besides, in carrying out assessment, following
government questionnaire. The result of such independent variables constructs were applied:
recoding process is herewith summarised. ƒ Performance expectancy
Use Behaviour ƒ Effort expectancy
ƒ Use of Parma municipality services ƒ Social influence
ƒ Use of DTV services ƒ Facilitating conditions
ƒ Frequency of use The other constructs left over, i.e. Behavioural
ƒ Time duration of each session Intention and Moderators, were not taken into
ƒ Type of use (intensive/shallow) consideration at this preliminary stage of the study.
ƒ Comparison with alternative channels Finally, ahead of computing each predictor of the
ƒ Weekly interactions novel UTAUT model, the three dependent variables
Behavioural intention were standardised in order to make them comparable
ƒ Willingness to pay for equipment
each other.
ƒ Willingness to pay for services
ƒ Interest levels toward DTV services
Performance expectancy
ƒ Perceived service usefulness 4 RESULTS
ƒ User satisfaction
ƒ Comparison with alternative channels In order to better understand the active relations
ƒ Factors influencing adoption among the new generated set of variables, either for
ƒ Privacy and security the general or for the information or for the
Effort expectancy interactive usage of T-government services, a
ƒ Perceived ease of use regression analysis was applied according with the
ƒ Attractiveness UTAUT reference model. This choice was suitable
ƒ Perception of technical problems given the low variability affecting the new
- Loss of television signal dependent variables. To this regard, it was pointed
- Too much time for application loading out how people didn’t use many of the several
- Loss of connection by the return channel available on-line facilities, but they typically used no
- Problems in establishing the connection more than one at time.
- Too much time to reach the objective
- Technical troubles in using equipment 4.1 Predictors Estimation
ƒ Interaction with equipment
ƒ Service functions Regression analysis allowed to estimate the different
ƒ Impact of equipment in the house weights of the UTAUT model factors applied to the
Social influence novel T-government usage structure, that is the
ƒ Social network
influence each of them may have in controlling the
Facilitating conditions
weight of a single predictor.
ƒ Decoder installation
ƒ User support provided by DTT Three models were thus arranged and tested,
ƒ User manual for STB install using as criteria the assessment respectively of STB

320
 FACTORS AFFECTING THE USAGE OF T-GOVERNMENT SERVICES - An Exploratory Study

Use, of Interactive Service Use, and of Informative UTAUT factors. Whether a predictor should be
Service Use. As shown in Table 1, set taken for a supplementary analysis, the best rank is
cont_utente_servizi (STB Use) as dependent Effort Expectancy.
variable, all coefficients with the exception of
Facilitating conditions were significant (overall 4.2 The Final Model
model: Adjusted R Square=.13; p<.000). Hence, it
can be said, the more subjects perceived high values By means of regression analysis it was demonstrated
of these three predictors in adopting the STB the if and how much the values taken by the dependent
more positive was the evaluation of this dependent variables, as well arranged from the T-government
variable (Beta=.279, .311; .236; overall p<.001). dataset, either depended on or were determined by
the correspondent values taken by the independent
Table 1: Dependent variable: general STB Use. variables, in our case the main determinants of
UTAUT. Figure 3 summarises the results of the
Predictors Beta t Sig. positive dependences found in the analysis.
Performance expectancy .279 3.406 .001
Effort expectancy .311 3.793 .000
Social influence .236 3.368 .001
Facilitating conditions .046 .648 .518

In the second model of Table 2, the evaluation of

Interactive Service Use, set cont_interattivi as
dependant variable, gave nearly the same results
than the previous one (overall model: Adjusted R Figure 3: T-government factors usage: final model.
Square=.14; p<.000). Therefore, still excluding
Facilitating conditions, the more subjects perceived The figure also shows that Facilitating Conditions
high values of the three predictors in using the hasn’t effective links to the hypothesised usage
Interactive Service the more positive was the constructs.
evaluation of this dependent variable (Beta=.315,
.285; .264; overall p<.001).

Table 2: Dependent variable: Interactive Service Use. 5 CONCLUSIONS

Predictors Beta t Sig. In this paper the main feature of a forecasting model
Performance expectancy .315 3.900 .000 has been applied in order to give a further
Effort expectancy .285 3.528 .001 representation of a field research dataset about ICT
Social influence .264 3.811 .000 information and interactive services proposed for
Facilitating conditions .032 .456 .649 residential users. Specifically, data coming from an
Italian T-government field experience at the
Finally, as shown in Table 3., the evaluation of Municipality of Parma were re-processed to bring to
Informative Service Use, set cont_informativi as more significant results towards new purposes, i.e.
dependent variable, found that no one predictor the identification of most relevant factors affecting
turned out to be effective (overall model: Adjusted R usage. In such a way, the low usage of interactive
Square=.003; p=n.s.); (Beta=.107, .153; .081; .024; services scored at Parma, and as well as of
overall p>.05). informative services and decoder adoption, was
explained in terms of “use of information
Table 3: Dependent variable: Informative Service Use. technology” key dependant variable, in UTAUT
mentioned as direct determinant of user acceptance
Predictors Beta t Sig. and usage behaviour. The determinants Performance
Performance expectancy .107 1.222 .223 Expectancy, Effort Expectancy, Social Influence and
Effort expectancy .153 1.749 .082 Facilitating Conditions were reformulated by means
Social influence .081 1.086 .279 of T-government data to assess if and how much the
Facilitating conditions .024 .315 .753 information system was humanly acceptable
developed. Results demonstrate that the procedural
According to this last model, Use of Informative scheme can be successfully applied to recode most
Services is not a variable depending on the given basic questionnaire items into new constructs

321
ICE-B 2008 - International Conference on e-Business

referable to UTAUT. The resultant model achieved Compeau, D. R., Higgins, C. A., and Huff, S., 1999.
is statistically significant for the first three constructs Social Cognitive Theory and Individual Reactions to
above mentioned, while it reveals to be partially Computing Technology: a Longitudinal Study, MIS
Quarterly (23:1), pp. 145-158.
adequate for the last construct. Results encourage
Cornacchia, M., 2003. Usability, a way to distinguish from
also to go ahead in further developments on some the good, the bad, and the irrelevant in the web, Cost
important issues. 269, Conference Proceedings, Media Centre Lume,
At this stage of the study about the usage of T- University of Art and Design, Helsinki (Finland),
government services, the UTAUT key moderators pp.159-163, September.
(i.e. age, gender, experience, voluntariness) were not Damodaran, L.(edited by), 2002. Analogue to digital
included in the process. However, a further analysis switch over: a scoping study for the digital television
has been planned to aim at finding whether it is project , March.
possible to incorporate in the model the variables Davis F., 1993. User acceptance of information
“income” and “education”, either as new moderators technology: system characteristics, user perceptions
and behavioral impacts, Int. J. Man-Machine Studies,
or as new determinants. AP, Vol. 38, pp.475-487.
This study is antecedent to a development also in Davis, F. D., 1989. Perceived usefulness, perceived ease
the area of forecasting the T-government services of use, and user acceptance of Information
diffusion processes for residential users. Actually, Technology, MIS Quarterly, 13(3), 319-340.
re-modulated datasets and UTAUT constructs can Dillon, A., 2000. Group dynamics meet cognition:
effectively feed a microsimulation model (Turk et al, combining socio-technical aspects and usability
2008), where different policy measures (such as engineering in the design of information systems, in E.
subsidisation or inclusion of different services) Coakes,. (et al.) Eds., The new Socio Tech: Graffiti on
the Long Wall, London: Springer-Verlag, pp 119-126.
could be tested from different aspects, for instance
Malhotra, Y., Galletta, D. F., 1999. Extending the
their cost, time and relevance. Technology Acceptance Model to Account for Social
Influence: Theoretical Bases and Empirical
Validation, Proceedings of 32th Hawaii International
ACKNOWLEDGEMENTS Conference on System Sciences (IEEE 99).
Papa, F., Spedaletti, S., 2001. Broadband cellular radio
telecommunication technologies in distance learning:
The T-government projects promoted by FUB and a human factors field study, Personal and Ubiquitous
CNIPA were co-funded via a competitive call Computing, 5, 231-242.
coming from a joint initiative of the Italian Ministry Papa, F., Spedaletti, S., 2004. Metodologia per la
of Communication and the Italian Ministry for valutazione dell’usabilità dei servizi interattivi offerti
Innovation and Technology. The field investigation dalla televisione digitale terrestre, FUB Internal
of the Project “Parma Municipality Services” was Report, December.
developed in the framework of the agreement Pierro, A., Mannetti, M., Livi, S., 2003. Self-Identity and
between Fondazione Ugo Bordoni and the project the Theory of Planned Behavior in the Prediction of
Health Behavior and Leisure Activity, Self and
partners, Municipality of Parma, Itcity and
Identity, 2: 47-60.
Enterprise Digital Architects. The authors would like Taylor S. and Todd, P. A., 1995. Assessing IT usage: The
to thank: Sebastiano Trigila (Chief of the FUB T- role of prior experience, MIS Quarterly (19:2), pp.
government program) for the supervision of the 561-570.
projects; Roberto Azzano (ANFoV) and Giulia Turk, T., Sapio, B., Cornacchia. M., Livi. S., Nicolò. E.,
Berni (ANFoV) for their contribution to the Papa. F., 2008. Microsimulating the Adoption of
supervision of user trials. Digital Television and T-Government Services,
Workshop “Digital Television Revisited: Linking
Users, Markets and Policies”, Budapest, May 15th.
Venkatesh, V., Davis, F. D., 2000. A theoretical extension
REFERENCES of the technology acceptance model: Four longitudinal
field studies. Management Science, (46:2), 186-204.
Ajzen, I., Fishbein, M., 1980. Understanding attitudes and Venkatesh, V., Morris, M. G., Davis, G. B., Davis, F. D.,
predicting social behaviour, Eaglewood Cliffs, NJ: 2003. User Acceptance of Information Technology:
Prentice-Hall. Toward a Unified View, MIS Quarterly vol. 27 No. 3,
Bagozzi, R. P., Davis, F. D., & Warshaw, P. R., 1992. pp. 425-478, September.
Development and test of a theory of technological
learning and usage, Human Relations, 45(7), 660-686.
Bandura, A., 1986. Social Foundation of Thought and
Action: A Social Cognitive Theory, Prentice Hall,
Englewood Cliffs, NJ.

322
 BUSINESS PROCESSES MANAGEMENT USING PROCESS
ALGEBRA AND RELATIONAL DATABASE MODEL

Kelly Rosa Braghetto, João Eduardo Ferreira

Department of Computer Science, University of São Paulo, Rua do Matão 1010, 05508-090 São Paulo, Brazil
[email protected], [email protected]

Calton Pu
College of Computing, Georgia Institute of Technology, Atlantic Drive 801, 30332-0280 Atlanta, U.S.A.
[email protected]

Keywords: Business Process Management, Workflows, Process Algebra, Relational Databases.

Abstract: Integrating information systems with tools that manage workflows and business processes is not always
a simple task. This difficulty becomes more accentuated when the execution control assumes countless
business processes. This work presents NavigationPlanTool (NPTool), a tool to control the execution of
business processes that can be easily integrated into the information systems. NPTool is supported by
Navigation Plan Definition Language (NPDL), a language for business processes specification that uses pro-
cess algebra as formal foundation. NPTool implements the NPDL language as a SQL extension and offers
two other important services: processes instantiation and process instances execution monitor. This paper
describes the NPTool showing how the process algebra features combined with a relational database model
can be used to provide a scalable and reliable control in the execution of business processes.

1 INTRODUCTION vide mechanisms that allow them to be easily used

within the information systems that manage business
Business Processes Management (BPM) involves processes.
methods, techniques and tools to support the project, This work presents the NavigationPlanTool
the execution, the management and the operational (NPTool), a tool that offers mechanisms for the repre-
analysis of business processes (Leymann et al., 2002). sentation and execution control of business processes
While the traditional definitions of workflow place supported by a process algebra formalism (Fokkink,
emphasis on the execution of operational processes, 2000). NPTool uses the Navigation Plan Defini-
BPM also gives support to the diagnosis phase, al- tion Language (NPDL) (Braghetto et al., 2007) and
lowing the processes to be analyzed in order to detect a relational database to specify business processes
flaws and possible improvements to the project. and to control their instantiations and executions.
In this context, associating formal frameworks NPTool implements NPDL as an extension of SQL;
with the process project phase is valuable since they this implementation allows an easy integration with
provide non-ambiguous models, improve the diagno- traditional information systems, which generally al-
sis capability and enable a reliable execution control ready have mechanisms that facilitate the access to
of the process. RDBMSs. The storage of the processes data in a rela-
Although there are a number of tools based on tional database adds scalability to the execution con-
formal frameworks directed to the management of trol provided by NPTool. Moreover, the processes
workflows and business processes, integrating these definitions can be reused in different applications. In
tools with other applications isn’t always easy. Some this context, the database maintained by NPTool can
of these tools are exclusively developed for mod- be viewed as a common repository of processes.
eling and simulation of processes, as occurs in Section 2 presents a summary of the related works
CPN/Tools (Beaudouin-lafon et al., 2000). In other and the reasoning behind the use of NPDL. Sec-
cases, the tools effectively carry out the execution tion 3 describes the services offered by the Naviga-
control of business processes, but they do not pro- tionPlanTool; in particular, Section 3.1 describes the

323
ICE-B 2008 - International Conference on e-Business

data structures in the relational database model to • Sequential Composition (“.”). the process term
represent the processes, and Section 3.2 explains the A.B means that the activity B will be enabled for ex-
mechanism for execution control. Using a real appli- ecution after the completion of the activity A. An ac-
cation as example, Section 4 illustrates the execution tivity can be an atomic action or a subprocess;
control applied to an instance of a process. Finally, • Alternative Composition (“+”). the process
Section 5 discusses the contributions of this work. term A + B means that initially both activities A and
B will be enabled for execution, but only one of them
can be executed;
2 RELATED WORK • Parallel Composition (“||”). the process term
A||B means that the activities A and B can be executed
Management technologies for business processes parallelly.
were developed to meet the following needs: (1) to These basic operators were directly extracted from
promote the separation between the specification of ACP (Algebra of Communicating Processes) and have
the process and its implementation; and (2) to take their semantics formally defined by a set of transition
away from the applications the responsibility of the rules that will be presented in the Section 3.2. In or-
execution control of business processes. Among the der to represent all control-flow patterns identified by
various existing languages to define business pro- Aalst et al in (Aalst et al., 2003), the NPDL was im-
cesses, it is not easy to see a formal pattern as a rep- proved with additional operators:
resentation basis, one that is capable of expressing in • Interleaved Parallel Composition (“|∗”). the
a non-ambiguous way the semantics associated to the process term A|∗ B means that the activities A and B
existing constructions in these languages. The for- can be executed in any order (e.g., A.B + B.A), but not
malisms regarded as natural candidates for this role in parallel;
are the Petri nets and the process algebras. Works • Multi Merge Composition (“&”). the process
such as (Aalst, 1998), (Puhlmann and Weske, 2005) term A&B means that the activity B will be enabled
describe the use of these formalisms in the specifica- for execution after the completion of each thread of
tion of business processes. control of activity A;
The Yet Another Workflow Language • Discriminator Composition (“∧”). the process
(YAWL) (Aalst and Hofstede, 2005) is the most term A ∧ B means that the activity B will be enabled
well succeeded approach for business process specifi- for execution after the completion of the first thread
cation based on Petri nets. It was developed intending of control of activity A;
to provide means for defining all the original control- • Unlimited Repetition (“?∗”). the process term
flow patterns described in (Aalst et al., 2003). Also A?∗ means that the activity A can be executed an un-
with the intention of representing business processes restricted number of times;
with a formal basis, Navigation Plan Definition • Number Limited Repetition (“?n”, where n
Language (NPDL) (Braghetto et al., 2007) was is a positive integer number). the process term A?5
created. NPDL is based on the concept of navigation means that the activity A must be executed five times;
plan of RiverFish architecture (Ferreira et al., 2005) • Function Limited Repetition (“? f ”, where f is
as well as in the operators of ACP (Fokkink, 2000), a a function that returns a positive integer number). the
member of the family of process algebras. The navi- process term A? f1 means that the activity A must be
gation plan was formalized in (Ferreira et al., 2006) executed the number of times calculated by function
as a set of all business processes demanded from f1 at execution time;
an application in order to achieve business goals. • Conditional Execution (“%r”, where r is a
Like in processes algebras, processes in NPDL are rule, e.g. a boolean function). the process term %r1 A
defined by algebraic expressions. The expression of means that the activity A will be enabled for execution
a process is built based on NPDL operators and steps if the return value of the rule r1 is true at execution
(atomic actions or processes); the operators indicate time;
the execution order of the steps. For completeness, • Negative Conditional Execution (“%!r”,
we will provide an overview of NPDL operators. where r is a rule, e.g. a boolean function). the process
term %!r1 A means that the activity A will be enabled
2.1 NPDL Operators for execution if the return value of the rule r1 is false
at execution time.
The basic control-flow patterns described in (Aalst Providing more details about NPDL is beyond the
et al., 2003) can be easily represented with three basic scope of this paper; the definition of the language
operators of NPDL: and the specification in NPDL of each one of the 20

324
 BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL DATABASE MODEL

control-flow patterns described in (Aalst et al., 2003) 3.1 The Relational Data Structures
can be seen in (Braghetto et al., 2007). Created by NPDL Interpreter
The relational data model does not have appropriate
3 NavigationPlanTool structures for representing processes and, therefore,
it requires an additional data structure to achieve this
NavigationPlanTool (NPTool) provides methods for task.
storing actions and processes in a relational database In (Braghetto et al., 2007), the NPDL syntax was
and for controlling the instantiation and execution of defined as an extension of the SQL syntax. The main
these processes. The tool offers operations like cre- goal of implementing NPDL as a SQL extension was
ation/removal of instances and services for monitor- to enable a RDBMS to create and handle business
ing the navigation plan execution. These services are processes, providing to information systems an easy
also responsible for storing logs of the execution of access to these features. The data structure kept stored
navigation plans in the database and for recovering in a database by NPTool is represented in the extended
executions that have been interrupted before comple- entity-relationship diagram shown in Figure 1. The
tion. diagram shows the data structures needed for repre-
The programming language used in the implemen- senting business processes and for controlling their
tation of NPTool was Java (Java 2 Platform Standard instantiations and executions.
Edition - J2SE 5.0). NPTool extends JDBC API - In NPDL, the navigation plan of a process
Java DataBase Conectivity Application Programming is defined by an algebraic expression formed by
Interface. JDBC enables Java programs to execute steps composed by operators that indicate the ex-
SQL commands and to interact with databases that are ecution order of these steps. The relationship-
compatible with SQL standard. The usage of JDBC set NAVIGATION PLAN REL between the entities-set
turns NPTool into a RDBMS independent tool. Since PROCESS and STEP represents this definition. The
it was developed as a library of functions, the NPTool specialization of STEP is total and disjoint, i.e., each
can be easily integrated into other Java applications. entity in STEP is a process, or an action, or an op-
NPTool is composed of three services: erator, or a rule, or a function or a number. As
• NPDL Interpreter. receives an input command described in Section 2.1, rules are always associ-
and makes the lexical, syntactic and semantical anal- ated with the operators “%” and “%!”, whereas func-
ysis. There are two possible situations: (1) the com- tions and numbers are associated with operator “?”.
mand is a NPDL valid command, and (2) the com- The entity-set PROCESS INSTANCE and the relation-
mand is not a NPDL valid command. In the for- ship INSTANCE LOG REL are associated with the in-
mer case, the interpreter will translate the command stantiation and control of process execution. An entity
to pure SQL commands before the submission to the of PROCESS INSTANCE is always associated with an
RDBMS. In the latter case, the command will be entity of PROCESS, as the relationship PROCESS INS-
directly passed to the RDBMS. The translated SQL TANCE REL shows; an instance represents a process
commands will be executed over a relational database request. The relationship INSTANCE LOG REL repre-
environment whose tables are created by the inter- sents the data related to execution status of the steps
preter to store processes, actions and instances data; that compose the navigation plan of a process in-
• Process Instantiation Service. provides func- stance. Each entity of INSTANCE LOG REL represents
tions for creating process instances. A process in- the execution of a step in a specific process instance.
stance represents a request to a specific process. All In Section 3.2, the implementation of the service
the instance data, as well as process definition data Process Instance Execution Monitor will be detailed
associated with the instance, are stored in a database; to show how process algebra properties and this re-
• Process Instance Execution Monitor. is re- lational database model were used for developing a
sponsible for linking a process instance to its execu- scalable and reliable engine that controls the business
tion data (navigation plan). This service contains the processes execution.
functions that control the execution of the navigation
plan of a process instance. 3.2 The Execution Control of Business
Processes in NPTool
When an execution of a process instance is started,
the navigation plan associated with the instance is re-
covered from database. According to the algebraic

325
ICE-B 2008 - International Conference on e-Business

Figure 1: Extended entity-relationship diagram of process data.

expression that represents the navigation plan, an ex- of the monitor service, there will be records related to
pression tree of the instance is built. Expression trees the instance in the log and they must be loaded to the
are used in the execution monitor service to determine navigation tree. This is made by updating the status
the execution order of the steps in the navigation plan. of the tree nodes associated with the steps indicated as
In this work, the expression tree of a process instance started, finished or canceled in the log. Each step per-
was called navigation tree. A navigation tree node can formed in the navigation plan of an instance results in
represent one of these three elements: a NPDL opera- the insertion or update of a record in the log.
tor, an action or a process. A navigation tree is a com-
plete binary tree; its internal nodes represent binary v √ v v √ v
x −→ x −→ x0 y −→ y −→ y0
operators of NPDL, whereas its leaf nodes represent v √ v v √ v (1)
x + y −→ x + y −→ x0 x + y −→ x + y −→ y0
actions or processes. As the navigation tree nodes rep- √
v v
resent the possible execution steps, an important at- x −→ x −→ x0
v v (2)
tribute of a node is its current status. A node, in a spe- x.y −→ y x.y −→ x0 .y
cific point of the instance execution, can be in one of
the following states: not started (N), started (S), fin- v √ v v √ v
x −→ x −→ x0 y −→ y −→ y0
ished (F) or canceled (C). The Figure 2 shows an hy- v v v v (3)
xky −→ y xky −→ x0 ky xky −→ x xky −→ xky0
pothetical example of one of the possible navigation
tree for an instance of process P = (a + b).c.(dke). f . Figure 3: Transition rules for basic operators of process al-
gebra. The variables x and y in the rules range over basic
process terms, while v ranges over the set of atomic actions.

The “navigation” through the tree of an instance

determines the execution order of the steps of this
instance. The algorithms of the monitor service of
NPTool use the semantics of NPDL operators and the
transition rules of process algebra to visit a naviga-
tion tree. Figure 3(1,2,3) shows the transition rules
extracted from (Fokkink, 2000) that define the oper-
Figure 2: Navigation tree of an instance of process ational semantics of terms of basic process algebra
P = (a + b).c.(dke). f . and ACP. The operational semantics of a language de-
scribes how a valid sentence must be interpreted in
After creating the navigation tree, the execution sequential steps. Figures 4 to 6 show how the transi-
monitor recovers the current state of the instance exe- tion rules are applied on the navigation tree branches
cution by consulting the execution instances log, that to guarantee that the instance status, after starting or
is stored in the database. This operation only is done finishing the execution of an action, will be consistent
for instances that have already started their execution, with the behavior specified by the algebraic expres-
but have not finished it yet. When an instance execu- sion of the process associated with the instance.
tion is started, all nodes in its navigation tree will have The four rules in Figure 3(1) states that the pro-
the current status not started. If the instance execu- cess defined by the term x + y terminates successfully
tion have already been started in a previous execution after executing x or y. Figure 4 shows how the alter-

326
 BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL DATABASE MODEL

resented by x.y executes first x and, after x have been

finished successfully, it starts the execution of y. Fig-
ure 5 shows the possible state changes in a branch
rooted at a sequential composition (“.”) operator.
The status of a node with a sequential composition
operator is specified by the status of its child nodes in
the following way:
• Started : if the left child has status started or if
Figure 4: State transitions for a branch rooted at an operator the left child has the status finished and the right child
of alternative composition. has status started. The right child will never have sta-
tus started while the left child does not have status
native composition operator (“+”) is treated in a nav- finished ;
igation tree branch. As in figures 7 to 14, the nodes • Finished : if the left and the right child has sta-
delimited by an ellipse with dashed line indicate the tus finished ;
actions that are enabled for execution in the current • Canceled : if the left and/or the right child has
instance state represented by the tree. The labeled ar- status canceled.
rows signalize the start or the end of the execution of The process term xky indicates that the terms x and
an action1 . So, S(a) signalizes the start of the action y will be executed parallelly, i.e., it is possible to ex-
a, while F(a) signalizes the end of the same action. v √ v
ecute an initial transition of x (x −→ or x −→ x0 )
The start or finish of the execution of an action in or a initial transition of y. This behavior is formally
an instance modifies the status of a leaf node in its specified by the rules in Figure 3(3). The set of all
navigation tree. Each modification in the status of a the possible states originated by an expression involv-
leaf node in a navigation tree requires the update of ing the parallel composition (“k”) operator is large, as
status of its predecessor nodes. To enable this opera- Figure 6 shows. The status of a node that represents
tion, each node keeps a pointer to its parent node, be- the parallel composition operator is specified by the
sides the pointers to the left child and the right child status of its child nodes in the following way:
nodes. The status node configuration in a navigation • Started : if the left or the right child has status
tree determines the current state of the instance; thus, started ;
a change in the instance state generates a new set of • Finished : if the left and the right child has sta-
actions currently enabled for execution. The status of tus finished ;
a node that represents an alternative composition op- • Canceled : if the left and/or the right child has
erator is specified by the status of its child nodes in status canceled.
the following way:
The conditional execution, negative conditional
• Started : if the left or the right child has status
execution, unlimited repetition, number limited rep-
started ;
etition and function limited repetition operators are
• Finished : if the left or the right child has status
not treated as nodes in the navigation tree. They are
finished ;
treated as attributes of nodes, since they influence the
• Canceled : if the left and the right child has sta-
execution of the complete branch rooted at the node
tus canceled.
with which they are associated. If a node have a rule
or a function associated with it, then it is necessary
to execute this rule or function before “visiting” the
node in the navigation algorithm. The execution of
the rule or function, like in the case of atomic ac-
tions, is a responsibility of the applications that use
the NPTool. When an action or a process term is
delimited by some of the two conditional execution
operators, the rule that conditionates the execution is
Figure 5: State transitions for a branch rooted at an operator attributed to the node that represents the delimited ac-
of sequential composition. tion or to the root node of the delimited process term.
In the example of Figure 7(a), the execution of ac-
In Figure 3(2), the rules state that the process rep- tion a is conditioned by the return value true of rule
1 In NPTool, the actions of a process may not have an r1 , while the execution of b is conditioned by the re-
instantaneous execution, i.e., the execution can be distin- turn value false. In this example, the return value of r1
guished by a start event and by an ending event. was true; so, after finishing the execution of r1 , only

327
ICE-B 2008 - International Conference on e-Business

Figure 6: State transitions for a branch rooted at an operator of parallel composition.

(a) (b)
Figure 7: Treatment of (a) node associated with rules or functions and (b) nodes with unlimited repetition.

the action a was enabled to be executed. tree should be visited. The execution monitor starts
In the case of repetitive actions or process terms, to visit the navigation tree from its root node. Ac-
the node that represents the action or term is labeled as cordingly the current status of the tree nodes, the nav-
repetitive, and if this repetition is limited by a number igation returns a set of steps currently enabled to be
or a function, this number or function will be associ- executed in the instance.
ated with the node. Figure 7(b) shows how a repeti- Only the basic operators of NPDL are kept in the
tive node is treated by “expanding” it in an equivalent navigation tree. The other operators are treated by
branch. a mapping algorithm that removes them of the nav-
The inspection of a navigation tree in order to get igation tree, replacing them by an equivalent branch
enabled steps to execution in an instance is also con- containing only basic operators. Figures 8 and 9 show
trolled by the transition rules expressed in Figure 3; how nodes rooted at non-basic NPDL operators in the
the rules are used to determine how the nodes of the navigation tree can be replaced by other equivalent

328
 BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL DATABASE MODEL

• a2 gathers the prices of the order items;

• a3 liberates the order for purchasing;
• a4 registers the purchase receiving;
• a5 registers a problem with the purchase.
The acquisition system can be specified by the
process P defined in NPDL by the commands:
SET P1 = a1 |* a2 + (a1 |* a2).P1;
SET P = %r1 P1 . a3 . (a4 + a5?* . a4);
The check step of order data is treated as a rule (in
the case, r1 ). Thus, the execution of the activities in-
volved in the acquisition system will only be enabled
if the order data is valid and there is available bud-
get for the purchase. The subprocess P1 encapsulates
the actions of sorting items (a1 ) and gathering items
price (a2 ) . The actions are composed by the operator
“|∗”, that indicates that they can be executed in any
order, but not in parallel. Moreover, P1 was defined
Figure 8: Treatment of a branch rooted at a multi merge in a recursive way. This indicates that these actions
composition operator. can be repeated an unrestricted number of times be-
fore the liberation of the order for purchasing. After
branches. the liberation of the order (action a3 ), it is possible
Processes that involve in their definition a refer- that problems occur in the purchase. These problems
ence to other processes are treated by a simple substi- are registered in the acquisition process by the action
tution. The navigation tree for the referred process is a5 . The repetition operator “?∗” was used associated
created and “connected” to the navigation tree of the with a5 , to represent the possibility of occurring more
instance, replacing the node that represents the pro- than one problem during the purchase.
cess. After the substitution of the process node by the The initial navigation tree of an instance of acqui-
tree that represents it, the algorithm for obtaining en- sition process is represented in Figure 10(a). Each
abled steps is applied on the root node of the referred node has, at the upper left side, a character that indi-
process. cates one of the following states: “N” (not started),
The execution of an instance is considered suc- “S” (started) and “F” (finished). It is important to no-
cessfully complete when the status of the root node of tice that, associated with the node that represents the
its navigation tree is set to finished. process P1 in the tree, exists the rule r1 . Such as the
tree nodes, a rule has a status too. Associated with the
node representing action a5 , exists the operator “?∗”,
that indicates that the action can have its execution
4 EXAMPLE OF EXECUTION repeated.
CONTROL In the beginning of the execution, the tree is vis-
ited starting on its root node (distinguished by a
This section illustrates a simulated execution of a pro- dashed line in Figure 10(a)). While the navigation
cess instance. It shows how works the navigation that algorithm is visiting the tree nodes, it comes across
gets the enabled steps and updates the state of the nav- the rule r1 and it adds the rule in the set of currently
igation tree. The process used as example is a simpli- enabled steps to execution. Since the execution of the
fied version of a system developed for controlling the other steps depends on the execution of step condi-
acquisition of items to a library collection. tioned by r1 , no other step can be executed at first.
Consider the set of rules and the set of actions Considering now that r1 have been executed and
belonging to the acquisition system that are defined, its return value was true, the algorithm for getting the
respectively, as R = {r1 } and A = {a1 , a2 , a3 , a4 , a5 }, currently enabled steps is applied on the node associ-
such that: ated with r1 in the tree and finds the node x in Fig-
• r1 checks the completeness of data from acqui- ure 10(b). The node x represents the process P1 and,
sition order and the availability of the budget for pur- as explained previously, it needs to be expanded by
chasing; replacing the node by its navigation tree. Figure 11(a)
• a1 sorts the order items according to some prior- shows the tree after the substitution.
ity; Continuing the visitation to the tree nodes, the al-

329
ICE-B 2008 - International Conference on e-Business

(a) (b)
Figure 9: Treatment of a branch rooted at the operators (a) interleaved parallel and (b) discriminator composition.

(a) (b)
Figure 10: (a) Initial state of the tree for an instance of acquisition process. (b) Tree after the execution of the rule r1 .

gorithm detects the existence of the operator “|∗” in children removed.

two different places in the tree (nodes x and y in Fig- The navigation for getting the currently enabled
ure 11(a)). Since the operator “|∗” is not a NPDL steps to execution detects the node y in Figure 12(b),
basic operator, it requires a special treatment. which once more represents the process P1 . The node
In NPDL, a process term like a| ∗ b is equivalent is replaced by the navigation tree of process P1 , result-
to a.b + b.a. Figure 11(b) shows the navigation tree ing the tree in Figure 13(a). In the tree in Figure 13(a),
after mapping the operators “|∗”. After the map- the nodes x and y represents the operators “|∗”, which
ping, the visitation algorithm is able to identify the must be mapped to basic operators, resulting the tree
currently enabled steps to execution. As the distin- in Figure 13(b). In the tree in Figure 13(b), the en-
guished nodes in Figure 11(b) shows, there are two abled steps are the actions a1 (nodes w and y), a2
enabled steps to execution in the navigation tree: ac- (nodes x and z) and a3 (node v). If the chosen step
tions a1 and a2 . for execution is the action a3 , the application of the
algorithm for propagating the status update through
Consider that, after the execution of r1 , the action the tree nodes over the node v in Figure 13(b) results
a2 have been executed, and the algorithm for propa- in the tree in Figure 14(a).
gating the status update through the tree nodes is ap- When the algorithm of navigation finds the node
plied on the nodes x and z of Figure 11(b). In the y in the tree in Figure 14(a), it will apply the map-
resulted tree, only the action a1 is enabled to execu- ping of the operator “?∗” associated with the action
tion; Figure 12(a) shows the tree after the execution a5 , resulting the tree in Figure 14(b). In this tree, the
of a1 . currently enabled steps to execution are the actions a4
In a navigation tree, branches rooted at a node and a5 . The tree in Figure 14(c) is obtained by execut-
whose status have already been defined as finished ing the action a4 and applying algorithm for propagat-
and which do not have enabled steps to execution can ing the status update over the node w in Figure 14(b).
be removed from the tree. The goal of this removal is The root node of the tree in Figure 14(c) has the sta-
the optimization of the main memory occupied by the tus finished, which indicates that the execution of the
storage of the trees. Figure 12(b) illustrates the tree process instance was terminated successfully.
resulted by the removal of the inaccessible branches
of the tree in Figure 12(a); the nodes distinguished by
the background color gray are the nodes that had their

330
 BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL DATABASE MODEL

(a) (b)
Figure 11: (a) Navigation tree after the substitution of P1 and, next, (b) after the substitution of operators “|∗”.

(a) (b)
Figure 12: Navigation tree (a) after the execution of actions a2 and a1 , and (b) after the removal of inaccessible branches.

5 CONCLUSIONS ing impact to the data structure. This model also

allows us to make reuse of the processes definition
NavigationPlanTool is a framework that uses NPDL between different applications that use the NPTool,
and a relational database model to specify business as it can be viewed as a common repository of pro-
processes and control their instantiations and execu- cesses. Moreover, the compositional characteristic of
tions. The NPDL language is based on the concept of processes algebra makes the composition of great pro-
navigation plan of the RiverFish architecture as well cesses from smaller ones possible. This constitutes an
as in the operators of processes algebra. important aspect for business processes, in which the
The simplicity of the relational database model reuse of definitions is a common occurrence.
(used by NPTool for the storage of processes, in- The implementation of NPDL as an extension of
stances and its executions) is consequence of the the SQL language provided by NavigationPlanTool
adoption of processes algebra as formal basis, due allows its easy integration to traditional information
to its algebraic and textual form. Albeit simple, this systems, which already have mechanisms that facili-
model has proved to be quite flexible, since it allows tate the access to RDBMSs. In addition to that, the
the enrichment of the expressiveness of NPDL (by in- storage of the processes definition data, instances and
cluding new operators in the language) without caus- its executions in a relational database makes the Nav-

331
ICE-B 2008 - International Conference on e-Business

(a) (b)
Figure 13: Navigation tree (a) after substituting P1 and, next, (b) after the substitution of the operators “|∗”.

(a) (b) (c)

Figure 14: Navigation tree (a) after executing action a3 , (b) after the removal of inaccessible branches and mapping operator
“?∗”; next, (c) after executing action a4 .

igationPlanTool a tool that can be used by informa- (USP); managing of clinical exams in the Human
tion systems that require scalable and reliable control Genoma Institute at USP, and the identification of
to execute business processes. These reliability and HIV drug resistance. Details about the latter can
scalability are assured by the use of a RDBMS. be seen in (Araújo et al., 2008); the referred paper
Currently NPTool has been used in three im- presents an alternative for mapping of genotypic drug
portant applications: controlling the acquisition and resistance algorithms rules using NPDL expressions.
lending of library items at University of São Paulo Our ongoing research topics include the identifi-

332
 BUSINESS PROCESSES MANAGEMENT USING PROCESS ALGEBRA AND RELATIONAL DATABASE MODEL

cation and reuse of control-flow patterns in business

processes, data flow management for business pro-
cesses and automated generation of NPDL expres-
sions from graphical representations.

REFERENCES
Aalst, W. (1998). The application of petri nets to workflow
management. The Journal of Circuits, Systems and
Computers, 8(1):21–66.
Aalst, W. and Hofstede, A. (2005). YAWL: Yet another
workflow language. Information Systems, 30(4):245–
275.
Aalst, W., Hofstede, A., Kiepuszewski, B., and Barros, A. P.
(2003). Workflow Patterns. Distributed and Parallel
Databases, 14(1):5–51.
Araújo, L. V., Sabino, E. C., and Ferreira, J. E. (2008). HIV
drug resistance analysis tool based on process alge-
bra. In Proceedings of the 2008 ACM Symposium on
Applied Computing, Fortaleza, Brazil, March 16-20,
2007, pages 1358–1364. ACM.
Beaudouin-lafon, M., Mackay, W. E., Andersen, P.,
Janecek, P., Jensen, M., Lassen, M., Lund, K.,
Mortensen, K., Munck, S., Ratzer, A., Ravn, K.,
Christensen, S., and Jensen, K. (2000). CPN/Tools:
A post-WIMP interface for editing and simulating
coloured petri nets.
Braghetto, K. R., Ferreira, J. E., and Pu, C. (2007). Us-
ing control-flow patterns for specifying business pro-
cesses in cooperative environments. In Proceedings
of the 2007 ACM Symposium on Applied Computing,
Seoul, Korea, March 11-15, 2007, pages 1234–1241.
ACM.
Ferreira, J. E., Takai, O. K., Braghetto, K. R., and Pu, C.
(2006). Large scale order processing through navi-
gation plan concept. In Proceedings of the IEEE In-
ternational Conference on Services Computing, pages
297–300. IEEE Computer Society.
Ferreira, J. E., Takai, O. K., and Pu, C. (2005). Integration
of collaborative information system in internet appli-
cations using riverfish architecture. In Collaborate-
Com. IEEE.
Fokkink, W. (2000). Introduction to Process Algebra.
Springer.
Leymann, F., Roller, D., and Schmidt, M.-T. (2002). Web
services and business process management. IBM Sys-
tems Journal, 41(2):198–211.
Puhlmann, F. and Weske, M. (2005). Using the pi-calculus
for formalizing workflow patterns. In Springer, editor,
Third International Conference on Business Process
Management (BPM 2005), pages 153–168.

333
 ON THE USE OF “QUALIFIED” DIGITAL SIGNATURES

Maurizio Adriano Strangio

Department of Mathematics, University of Rome “Roma Tre”, Italy
[email protected]

Keywords: Qualified Digital Signatures, Key Agreement Protocols.

Abstract: The European Commission Directive 1999/99/EC aims to provide a community-wide framework for the use
of electronic signatures and thus for promoting electronic trade and communication among the member states.
The directive introduces the notion of “qualified” digital certificates as a means to maintain legal effects of
digital data that are analogous to those of paper-based signatures. To this end, proofs of (physical) identity and
possession (of the private key) are fundamental requirements that must be fulfilled by the requester during the
public key enrollment process.
Digital signatures are often employed as secure buildings blocks in key agreement protocols that use public
key authentication. The need for the rigorous analysis of such protocols has recently emerged; there are
currently several formal models of distributed computing that may serve for this purpose. However, we point
out these models employ rather trivial or unpractical approaches in the modeling of the procedures and policies
employed by certification authorities.
We believe that usage of qualified certificates not only should represent the standard practice for CAs in order
to sustain secure electronic commerce (and in general all forms of secure communication) but also represents
the first step towards the domain of a global PKI.

1 INTRODUCTION Key Infrastructure (PKI) which is an arrangement that

entails a trusted third party (a.k.a. certification author-
Symmetric key algorithms have a troublesome short- ity — CA) fulfilling a notary function by binding the
coming — the need to generate and distribute a se- public key (and a set of informative attributes) to the
cret key to the parties who wish to privately commu- owner of the corresponding private key. The outcome
nicate. Furthermore, keys must be exchanged prior to of the process is a digital certificate (e.g. X.509v3),
any communication by making use of a (out-of-band) signed by the authority itself, that provides undis-
secret channel and not via a public network. putable evidence of the true identity of the principal
With public key cryptography, in a network of n holding the private key/public key pair.
principals, each party needs only to distribute a public While this approach is technically straightfor-
key to her n − 1 peers so the total number of keys to ward, the management (e.g. operation and funding)
manage is O(n), rather than O(n2 ). of a PKI can become quite intricate (see (Ellison and
However, with asymmetric key algorithms public Schneier, 2000) for an overview of the difficulties that
keys can be easily exposed (since they are delivered may be encountered) so that the public key validation
in unencrypted form) therefore a principal using the process may be almost as difficult as distributing se-
wrong key may be totally unaware that a malicious cret keys (after all symmetric key cryptography has
party is posing as her intended peer. As the result, been profitably employed since the era of Ceasar).
public keys must be certified in a separate process Whether a CA may be regarded as trustworthy
prior to any communication, possibly by a trusted (which ultimately is a decision of the relying party)
third party; for the parties willing to communicate is a major concern and is not just a matter of con-
this is logically equivalent to having an authenticated sidering certain distinguishing elements such as the
channel available to distribute the public keys in order reputation or the nationality of the CA. It turns out
to receive assurance of each others purported identity. that there are other requirements that must be satis-
In practice, this service is provided by a Public fied and often these are not clearly stated in the CA

334
 ON THE USE OF “QUALIFIED” DIGITAL SIGNATURES

policies (e.g. in the Certification Practice Statements corresponding to the target public key; therefore a
— CPS) or, even worse, not even completely under- physical proof-of-identity is essential for the legal
stood. For this reason, in some countries (e.g. Italy) recognition of digital signatures.
there are government agencies entitled to perform the CA policies often refer to the PKCS#10 standard
final accreditation of CAs (which are also subjected (PCKS#10v1.7, 2000) which describes a syntax for
to periodic inspections to verify full compliance with certification requests; an entity applying for a cer-
the law). tificate may be prompted by the RA to digitally sign
In fact, the main thesis of this paper is an electronic “certificate request” that will be subse-
that a “physical proof-of-identity” and “proof-of- quently conveyed to the CA for the final certificate
possession” are both mandatory to maintain the legal handout. Such an arrangement subsumes a proof-of-
status of digital signatures. With a physical proof- possession. Surprisingly, the standard also includes
of-identity the certificate requester must provide cor- the provision for requests using paper forms; this im-
roborative evidence of her identity (e.g. an id-card) plies that CAs should define alternative means in their
in a “face-to-face” registration procedure at the reg- policies for the requester to prove knowledge of the
istration authority (RA). With a proof-of-possession private key.
the subscriber must prove he exerts (exclusive) con-
trol over the private key corresponding to the public
key undergoing the certification.
Although the “physical” identification process is 2 DIGITAL CERTIFICATES IN
often perceived as an unduly burden by most sub- THE EUROPEAN COMMUNITY
scribers (although it is essentially identical to apply-
ing for a driving license) it is beneficial for a number The European Commission Directive 1999/99/EC
of reasons: (European-Parliament, 1999) aims to provide a com-
- it discourages from making use of PKIs for crim- munity framework for the use of electronic signatures
inal activity; and thus for promoting electronic trade and commu-
nication among the member states.
- facilitates law enforcement agencies in the In particular, the legal recognition of electronic
prosecution of the crimes related to signa- signatures entails the use of advanced electronic
ture/encryption schemes (e.g. identity theft, signatures based on “qualified certificates”, which
fraud); are created by secure (signature-creation) devices
- is also valuable for assessing the trustworthiness (e.g. tamper-proof standard-compliant devices such
of individuals. as smart cards holding the private key), to ensure they:
On the one hand, the proof-of-possession require- (a) satisfy the same legal requirements of hand-
ment has been the subject of many discussions (see written signatures on paper-based data (refer to
(Asokan et al., 2003; Lauter and Mityagin, 2006)) the definition of advanced electronic signature in
with the majority of PKI standards firmly encourag- the Directive);
ing its use (observe that when the CA generates the
(b) are admissible as evidence in a court of law.
public/private key pairs this requirement is immedi-
ately fulfilled). On the other hand, many CAs will The majority of European member countries have al-
issue certificates offering different levels of identity ready embodied the directive in their legal systems
assurance; a physical proof-of-identity is required for (civil laws).
the certificates that are used in applications that need Qualified certificates are certificates that com-
to establish the purported identity of the subscriber ply with the rules contained in Annexes I and II of
with certainty (e.g. commercial transactions); this is the aforementioned Directive and appear to satisfy
opposed to certificates issued with email addresses both the proof-of-possession (Annex I, letter (e)) and
used as identification credentials (which are suitable proof-of-identity (Annex I, letter (c)) requirements as
for non critical applications). we discussed in the preceding section.
However, often an explicit indication of how the Under such arrangements we call the resulting
subscriber was identified is omitted from the certifi- certificates “strong qualified certificates”; as opposed
cate attributes thus opening subtle vulnerabilities in to “weak qualified certificates” which may be sup-
its use (the relying party may unwillingly omit to per- plied by CAs that do not strictly comply with both
form any validity checks). the above requirements (e.g., consider a CA perform-
We stress that a proof-of-possession merely im- ing a proof-of-identity by means of an on-line zero-
plies that the applicant has access to the private key knowledge identification protocol).

335
ICE-B 2008 - International Conference on e-Business

3 KEY ESTABLISHMENT any time;

PROTOCOLS THAT USE c. Arbitrary key registration. Parties can register ar-
DIGITAL SIGNATURES bitrary public keys (even the same key as some
other party) without any validity checks. The ad-
Digital signature schemes (and other public key en- versary can register public key for corrupted par-
cryption schemes) are often employed as secure ties at any time.
building blocks in key establishment protocols that We comment that case c. is unrealistic since it is ex-
use public key authentication for the purpose of en- tremely unlikely that a CA will ever accept to enroll
suring the parties involved in the communication that public keys for which the requester has not provided
the session key was established with the intended peer any validity check; the minimum requirement is veri-
and not with an impostor. Once a session key is avail- fying the identity of the applicant.
able, the data exchanged is authenticated and undis- Item a. exemplifies the behavior one would nor-
closed against third parties thus allowing secure com- mally expect from honest principals holding valid cer-
munications over an insecure network. tificates. The case of interest here is the compromise
Traditionally, cryptographic protocol security has of the private key of a principal whether or not he is
been a matter of perceived confidence supported by aware of this fact (the adversary may be able to subtly
heuristic proof arguments and by the protocol surviv- obtain a copy of the private key).
ing many years of public scrutiny. More recently, the The proof-of-possession requirement invoked by
approach has changed and the priority has switched item b. is not sufficient to achieve adequate levels
to developing formal proof frameworks. In this con- of security (as discussed above); we have already
text, a main line of research employs complexity- pointed out that without a proof-of-identity CAs will
theoretic models for distributed network computing not issue certificates. We now elaborate further on this
and is dedicated to key establishment protocols (Bel- point.
lare and Rogaway, 1993; Blake-Wilson and Menezes, Lauter and Mityagin (Lauter and Mityagin, 2006)
1998; Shoup, 1999; Bellare et al., 2000; Canetti and have recently presented protocol KEA+ that, as op-
Krawczyk, 2001; LaMacchia et al., 2006; Diffie et al., posed to the original version (KEA), is resilient to un-
1992). Such models employ both private and public known key share (UKS) attacks; as a countermeasure,
key cryptographic techniques. they suggest to include the identities of the protocol
In the above models most of the fine level details participants as arguments of the key derivation func-
of PKIs are abstracted away; while in principle this tion (KDF) and also claim that the above countermea-
is a reasonable approach (we already mentioned that sure avoids the need for a proof-of-possession.
PKIs are an intricate subject on their own) we often Recall that a UKS attack involves a (man-in-the-
see that important topics such as the key registration middle) adversary, posing as a legitimate party (say
procedures and policies employed by CAs are disre- C) in a protocol run between honest parties (say A
garded by many authors. We believe that such issues and B), that is able to convince one party (e.g. A) to
are not only tightly related to the correct operation accept her identity (C) while the peer (B) is unaware
of the protocol but may also eventually lead to (legal) of this fact (i.e. B thinks he’s interacting with A). This
disputes among the participants and therefore must be vulnerability is regarded of interest whether or not the
appropriately settled. adversary is able to have the attacked party accept a
As a starting point for our discussion, let us recall chosen session key.
three public key registration procedures commonly We believe there are two issues that are over-
encountered in the literature (LaMacchia et al., 2006): looked in their arguments. Firstly, the adversary must
provide a valid proof-of-identity; therefore, unless she
a. Honest key registration. All parties (including is willing to reveal her true identity (otherwise she
those controlled by the adversary) follow the key must be able to perfectly disguise herself and also of-
generation procedures honestly and register the fer a counterfeit id card — indeed a non trivial task),
resulting public keys before engaging in any com- she may be liable of being legally pursued for her ac-
munications. The adversary can corrupt parties tions in a court of law. Observe also that the (physical)
only after key registration has completed; proof-of-identity policy rules out the well known on-
b. Proof-of-possession. An authority performs some line UKS attack described by Kaliski (Kaliski, 2001)
validity check upon public key registration. In against the MQV protocol (Law et al., 2003) (notice
particular, a party is required to prove knowledge that in this attack the adversary is able to have the tar-
of the corresponding secret key. The adversary get party accept a chosen session key).
can register public keys for corrupted parties at Secondly, the lack of a proof-of-possession (albeit

336
 ON THE USE OF “QUALIFIED” DIGITAL SIGNATURES

not being necessary for a protocol to resist UKS at- is cryptographically strong enough — say, at least 128
tacks) implies loss of the non-repudiation property for bits) or as an access token to be included as an argu-
digital signatures; at a later time the signer can even- ment of the registration protocol on the user side (and
tually deny having participated in the protocol run verified by the CA) thus providing an authenticated
(and therefore having established a particular session communication channel.
key with a peer).
We may conclude that the KEA+ protocol (and
many others alike) are not suitable for electronic trade 4 CONCLUSIONS
and commerce and at best can be used within the
realm of a corporate domain for the exchange of data
In this paper we have (re)affirmed the need for CAs
between employees (the proof-of-possession would to require mandatory proofs of identity and posses-
be satisfied anyway since the keys are generated and
sion (of the private key) from subscribers requesting
managed in house). digital certificates for public keys in order to preserve
From the above remarks we see that the primary
the legal status of the associated signature schemes.
sources of concern derive from the legal implica- We have also reviewed the notion of qualified certifi-
tions involved in electronic data processing facilities
cates as introduced by the European community di-
(which constitute the underlying principles of the Eu- rective 1999/99/EC (European-Parliament, 1999). As
ropean Directive).
far as electronic documents are concerned (e.g. con-
The lessons learned are twofold: tracts) the directive says that all member states should
- strong qualified certificates are necessary to pre- ensure that advanced electronic signatures (i.e. signa-
vent legal disputes; tures that are linked to qualified certificates) must be
- if legality is not a concern peers may use weak given the same legal effect as paper-based signatures
qualified certificates. and are thus form valid evidence in court trials.
Therefore, relying parties must be able to verify cer- We have also pointed out that formal models of
tificate policies of the CA to determine whether the distributed computing, where the security of key es-
certificates are suitable and trustworthy for a particu- tablishment protocols using public key authentica-
lar application. tion is evaluated, often adopt a rather trivial approach
As a consequence, CAs should also have clearly when considering public key enrollment procedures.
stated policies for appropriately setting certificate at- Our main thesis is that large-scale CAs (offering
tributes; for example, in strong qualified certificates national or international cross-border services) should
the X.509v3 Key Usage extension (RFC3280, 2000) only offer qualified certificates (with the appropriate
should always specify only one use for the public key proofs established) whether they are relative to digi-
(namely for signature validation) and have the non re- tal signatures used to sign documents or employed as
pudiation bit asserted. building blocks in key establishment or other crypto-
We mention the recent attempt of Boldyreva et al graphic protocols. Other types of certificates, which
(Boldyreva et al., 2007) to set out rigorous models of we have broadly classified as weak qualified certifi-
the public key enrollment process with a CA. In par- cates, only grant a limited degree of security (assur-
ticular, they show that a simple challenge-response ance) with no legal effects and can be eventually used
interactive protocol suffices to fulfill the proof-of- in particular domains (e.g. the PKI used in a home-
possession requirement. Notice that a fundamental banking system).
assumption in their work is the existence of an authen- Although (in principle) one could choose the
tic channel between the user and the CA; although (as appropriate certificate depending on the application
the authors admit) it would be otherwise difficult to (among those offered by a CA) there would be signif-
establish any reasonable security claim this hypothe- icant benefits if all the players involved could assume
sis enormously reduces the practical usefulness of the that certificates delivered an equivalent degree of as-
registration protocols (e.g. running it on an open net- surance rather than having to rely on non-standard
work environment is risky — similar arguments apply policies. To this end, it would be desirable that strong
to the proof-of-possession protocols found in the tech- qualified certificates be adopted as a universal stan-
nical literature for the Internet (RFC2875, 2000)). As dard for digital signatures, thus delivering further im-
a possible enhancement to the registration protocols, pulse to the development of interoperable world-wide
one may consider the CA assigning a unique trans- PKIs.
action key TK to the subscriber when the physical
proof-of-identity is performed; the TK may be used
only once either as a shared session key (provided it

337
ICE-B 2008 - International Conference on e-Business

ACKNOWLEDGEMENTS RFC3280 (2000). Internet X.509 Public Key Infrastructure

Certificate and Certificate Revocation List CRL Pro-
The author is grateful to the reviewers for their helpful file. Network Working Group.
comments and suggestions. Shoup, V. (1999). On Formal Models for Secure Key Ex-
change. Technical Report RZ 3120, IBM Research.

REFERENCES
Asokan, N., Niemi, V., and Laitinen, P. (2003). On the Use-
fulness of Proof-Of-Possession. Proceedings of the
2nd Annual PKI Research Workshop, pages 122–127.
Bellare, M., Pointcheval, D., and Rogaway, P. (2000). Au-
thenticated Key Exchange Secure Against Dictionary
Attack. In Proceedings of EUROCRYPT 2000, LNCS
1807:139–155.
Bellare, M. and Rogaway, P. (1993). Entity Authentica-
tion and Key Distribution. In Proceedings of CRYPTO
1993, LNCS 773:232–249.
Blake-Wilson, S. and Menezes, A. (1998). Entity authen-
tication and Authenticated Key Transport Protocols
Employing Asymmetric Techniques. Security Proto-
cols - 5th International Workshop, LNCS 1361:137–
158.
Boldyreva, A., Fischlin, M., Palacio, A., and Warinschi,
B. (2007). A Closer Look at PKI: Security and Ef-
ficiency. Proceedings of PKC 2007, LNCS 4450.
Canetti, R. and Krawczyk, H. (2001). Analysis of Key Ex-
change Protocols and Their Use for Building Secure
Channels. Advances in Cryptology-EUROCRYPT
2001, LNCS 2045:453–474.
Diffie, W., van Oorschot, P., and Wiener, M. (1992). Au-
thentication and Authenticated Key Exchange. De-
signs, Codes and Cryptography, 2:107–125.
Ellison, G. and Schneier, B. (2000). Ten risks of PKI: What
you’re not being told about Public Key Infrastructure.
Computer Security Journal, 16(1):1–7.
European-Parliament (1999). Directive 1999/99/EC of the
European Parliament and of the Council of 13 Decem-
ber 1999 on a Community Framework for Electronic
Signatures.
Kaliski, B. (2001). An Unknown Key Share Attack on the
MQV Key Agreement Protocol. ACM Transactions
on Information and System Security, pages 36–49.
LaMacchia, B., Lauter, K., and Mityagin, A. (2006).
Stronger Ssecurity of Authenticated Key Exchange.
http://eprint.iacr.org/2006/073.
Lauter, K. and Mityagin, A. (2006). Security Analysis of
KEA Authenticated Key Exchange Protocol. Pro-
ceedings of PKC’06, LNCS 3958:378–394.
Law, L., Menezes, A., Qu, M., Solinas, J., and Vanstone,
S. (2003). An Efficient Protocol for Authenticated
Key Agreement. Designs, Codes and Cryptography,
28:119–134.
PCKS#10v1.7 (2000). Certificate Request Syntax Standard.
RSA Laboratories.
RFC2875 (2000). Diffie-Hellman Proof-of-Possession Al-
gorithms. Network Working Group.

338
 RESEARCHING SEARCH
A Study into Search Engine Marketing Practices in Ireland

Chris Barry and Debbie Charleton

Department of Accountancy and Finance, National University of Ireland, Galway, University Road, Galway, Ireland
[email protected], [email protected]

Keywords: Information search, Search Engines, Search Engine Marketing, Website optimization, ethics.

Abstract: Researchers have identified the Web as the searchers first port of call for locating information. Search
Engine Marketing (SEM) strategies have been noted as a key area for firms to consider when developing,
maintaining and managing their Websites. SEM falls into two categories - ‘Search Engine Optimisation’
(SEO) and ‘Paid Search’. To examine how these SEM methods are reflected in practice, and the issues
inherent in carrying out SEM effectively, a survey was conducted amongst small to medium enterprises
(SMEs) in Ireland. The research reveals that Irish SMEs plan to spend more time and resources on SEM in
the future. Most firms utilize an informal SEM strategy, where SEO is perceived to be the more effective
technique in driving traffic to a company Website. Respondents cite the use of ‘keywords in title and
description tags’ as the most used SEM technique; while ‘Pay for Placement’ was found to be the most
widely used Paid Search technique. In concurrence with the literature, measuring SEM performance remains
a significant challenge with many firms unsure if they measure it effectively. An encouraging finding from
the study is that Irish SMEs adopt a positive ethical posture when undertaking SEM.

1 INTRODUCTION fall into two categories - ‘Search Engine

Optimisation’ (SEO) and ‘Paid Search’ (Boughton,
For many, online search has become an important 2005; Jansen and Resnick, 2005; Feng, Bhargava,
daily activity to navigate the Internet and has and Pennock, 2007). SEO is the process of
triggered fundamental changes in the way people identifying and fine-tuning elements of a Website in
seek and find information (Browne, Pitts and order to achieve the highest possible visibility when
Weatherbe, 2007). Boughton (2005) and Fain and a Search Engine responds to a relevant query (Xing
Pedersen (2006) have identified the Web as the and Lin, 2004; Zhang and Dimitroff, 2005). SEO
searchers first port of call for locating information. leverages the Search Engine algorithm to channel
Search Engines regularly feature amongst the most users to specific Websites, the results of which are
frequented Websites and are a vital mechanism by referred to as organic results. Paid Search uses
which e-commerce sites promote themselves (Jansen Search Engines to trigger a display of
and Molina, 2006). Given the importance of the advertisements based on the topic or keyword
Search Engine as a method for finding information, entered (Laffey, 2007).
and the opportunity it presents to advertisers as a A major factor in choosing one strategy over
marketing medium, Search Engine Marketing another is the searcher’s perception of sponsored
(SEM) strategies have been noted as a key area for listings (Paid Search) over that of organic or
firms to consider when developing, maintaining and editorial listings (SEO). Jansen and Resnick (2005)
managing their Websites (Sen, 2005). suggest that searchers tend to trust the relevance of
organic links over sponsored links, whereas
Boughton (2005) argues that searchers more
commonly following sponsored advertisements. Sen
2 STUDY BACKGROUND (2005) uses an analytical model to assess the use of
Paid Search over SEO. He finds that even if the cost
SEM has emerged as the main method firms use to of Paid Search and SEO were the same, firms would
successfully increase the visibility of their Website. still select Paid Search as the most popular strategy.
SEM consists of a number of methods that broadly

339
ICE-B 2008 - International Conference on e-Business

In contrast, Xing and Lin (2004) find SEO more There is widespread recognition of the
popular based on cost and due to searchers’ importance of using performance metrics for SEO
perceptions of Paid Search results. Jansen and and Paid Search in order to achieve a competitive
Molina (2006) surmise that a dual strategy of SEO advantage (Weischedel, Matear and Deans, 2005;
and Paid Search is the best approach having found Fain and Pedersen, 2006; Laffey, 2007). However
that Paid Search results are just as relevant as non- despite theses calls there is general consensus
paid results (SEO). Laffey (2007) discusses how a throughout the literature regarding the mis-use of
strategy may depend on which Search Engine is measurement tools for SEM (Weischedel et al,
subscribed to. He reports that Google users are more 2005). Some SEO measures put forward are: setting
likely to click on organic results and least likely to a baseline to measure inbound links to a site, where
click on paid results, whereas MSN users are more a company should continue to improve the number
likely to click on Paid Search results. and quality of inbound links to increase the site’s
Several optimizing SEO techniques have been page rank; and site usage statistics (Schultz and
identified. Some of these include: providing Fristedt, 2005) where the likes of Google Analytics
keyword rich Website content throughout all Web can be used. Zhang and Dimitroff (2005) and Seda
pages (Seda, 2004; Schultz and Fristedt, 2005; (2004) recommend tweaking different SEO aspects
Dawson and Hamilton, 2006); using keywords in the of the Website and then measuring the change in
title and description meta tags in the Website code rank of the Website on the SERP on a continual
(Zhang and Dimitroff, 2005); and keeping the likes basis. Paid Search measures put forward include:
of flash, graphics, forms and frames to a minimum Cost Per Click (CPC) where the cost to an advertiser
(Schultz and Fristedt, 2005; Dawson and Hamilton, to generate one click is monitored; Cost Per Mille
2006). Mentz and Whiteside (2003), Oliva (2004), (CPM) where the cost to display an advertisement a
Seda (2004), and Schultz and Fristedt (2005) thousand times is recorded; Cost Per Action (CPA)
recommend a link development strategy whereby which monitors the cost for actions such as signing
other good quality and relevant Websites are used to up for a new account or making a sale; and
develop or implement a link back to the company Conversion Rate which is the ratio between CPC
Website. and CPA (Fain and Pedersen, 2006). Smith (2002)
Paid Search takes a number of different forms. discusses conversion metrics as a key tool for
Paid Placement is where, in response to a particular measuring ROI. Another important and commonly
search term, a firm pays a fee for their link to appear used measure is Click Through Rate (CTR). CTR is
on the search results page, for example Google’s the ratio between the number of times a Web link is
AdWords. Paid Inclusion is where Search Engines displayed against the CPC (Kumar and Shah, 2004;
are paid to index pages from a Website (Thurow, Fain and Pedersen, 2006).
2007). These results can often be mixed into the Ethical practice in SEM is an increasingly
organic/editorial listings on the search engine results important consideration for all stakeholders (Palmer,
page (SERP). For example, Yahoo and other Search 2005), particularly searchers who rely on Search
Engines firms offer Paid Inclusion programmes for Engine to organise and distribute returned results in
this purpose. At present Google does not offer such an ethical way (Zimmer, 2006). Ethics takes two
a service (Mangalindan, 2003). With Paid forms - the responsibility of Search Engine operators
Submission, advertisers pay Search Engines to speed and the onus on Search Engine marketers to conduct
up the process of having their Website reviewed themselves ethically. Conflictingly, it is in the best
more quickly than they would under a free interest of Search Engines operators to push for paid
submission (Sen, 2005). However there is no advertisement (that generate revenue) rather than
guarantee that it is accepted and included in the organic results. This leads searchers to be concerned
SERP (Thurow, 2007). Contextual search is another about SERP results since some Search Engines do
type of Paid Search whereby search results are not disclose the difference between sponsored and
returned based on user behaviour and relevance organic search results (Moxley, Blake and Maze,
rather than matching keywords (Vine, 2004), for 2004) - a clear breach of any assumed ethicality.
example Google’s AdSense programme. Local Other concerns are expressed about: unethical link
search allows advertisers to target local customers in development such as when a site is found with
a geographic region rather than marketing to a identical link development on other sites that are
generic audience (Jones, 2006). Google permits this completely unrelated; bogus blogs generated to
feature through its AdWords programme. include links to increase the Website’s rankings
(Economist, 2006); ‘Black Hat SEO’ where popular

340
 RESEARCHING SEARCH - A Study into Search Engine Marketing Practices in Ireland

links are rented out from quality Websites and technical aspects of SEM should not be a barrier to
hidden links are placed on prestigious sites unknown understanding or developing online marketing
to their owners; and Web spamming where invisible techniques.
or unreadable text, or excessive use of the same
word are used to fool the Search Engine spider into 4.2 Nature of the SEM Strategy
thinking there are certain keywords on the Website
(Esparza, 2007). Click fraud is also discussed at Concurring with recommendations elsewhere
length in the literature (Asdemir and Yaha, 2006; (Ostler, 2001; Porter, 2002), two thirds of those
Laffey, 2007). It occurs when an advertiser falsely surveyed describe their SEM strategy as closely
generates clicks on an ad with the only goal of aligned with their firms overall business strategy.
increasing the payment to the advertiser. This finding suggests firms have moved beyond
experimentation and approach a more mature
relationship between their online and broader
3 RESEARCH METHODOLOGY strategies.
Many respondents (46%) revealed their SEM
The broad objective of this research effort was to strategy was of an informal nature, while (39%)
examine SEM strategies of small to medium sized reported a fairly formal or a formal SEM strategy.
enterprises (SMEs) in Ireland. It was also designed The remainder indicated that they did not carry out
to: identify the specific SEM techniques used; reveal any SEM strategy. This contrasts with the
which strategies are perceived as being more exhortations of writers that recommend a strategy
effective; discover if SMEs use performance formulation to achieve sustainable competitive
metrics; to investigate if firms consider an ethical advantage (Varadarajan and Jayachandran, 1999;
perspective when undertaking SEM activity; and to Novak, Hoffman and Yung, 2000; Schultz and
identify the main challenges faced in undertaking Fristedt, 2005; Sen, 2005). In relation to the
SEM. The research method used was an extensive, effectiveness of their SEM strategy, nearly three
quantitative survey. The population from which the quarters of respondents agreed they would consider
sample was drawn possessed the following their SEM strategy ineffective if it did not return
attributes: an SME operating in Ireland with an in- their company’s Website link on the first page of the
house marketing manager or specialist in a SERP. This finding is reinforced by research that
marketing department and an operating company suggests searchers view at most the first few pages
Website. The sampling frame was selected based on of the SERP (Lempel and Moran, 2000; Zwick,
the research questions, as well as the attributes listed Rapport, Lo and Muthukrishnan, 2003).
above for selection. The sample information was
compiled from Kompass, a business database listing An overwhelming number (91%) listed Google
companies in Ireland. A total of 95 responses were as the Search Engine most subscribed to for
received out of a sample of 623, giving a response indexing by companies. This is followed by Yahoo
rate of 15.24 %. The data collection method chosen at 65%, with MSN coming significantly behind
for this study was a self-administered postal (29%), while AOL, Alta Vista and Ask only
questionnaire and a matching Web-based received a very small percent. These findings are in
questionnaire. line with findings elsewhere by Schultz and Fristedt
(2005) and Taylor (2007). Schultz and Fristedt noted
that Google receives the highest number of keyword
searches followed by Yahoo. The findings here
4 FINDINGS suggest Irish firms are aware of the importance of
subscribing to the most popular search engines to
4.1 Website Development target the largest possible market. It may also
suggest that many of these SME’s are subscribing to
Most firms that participated in the survey develop Google because it is free, whereas subscription to
and manage their Website internally. While one other search engines such as Yahoo and MSN incur
third outsource the development of the Website they cost.
still manage and update their Website in-house. The
significance of this is that most respondents have the
capacity to carefully manage their SEM strategy
with expertise based in-house. For these firms

341
ICE-B 2008 - International Conference on e-Business

4.3 Types of SEM Strategy Used by 4.4 SEO Strategy Techniques

SMEs
Respondents cite the use of ‘keywords in title and
The literature suggests a key factor in a firm’s description tags’ as the most used SEO technique,
choice of SEM has much to do with how searchers followed by use of ‘keywords throughout the whole
perceive organic versus sponsored results (Seda, Website’ (see Figure 2). The first of these findings
2004; Boughton, 2005; Jansen and Resnick, 2005; contrasts with the recommendations of Seda (2004)
Sen, 2005). If firms are reading searchers correctly, and Schultz and Fristedt (2005) who state that
they are choosing to focus on SEO. Nearly half of keywords should be used throughout all pages of the
respondents indicated that they predominantly use Website content. While Dawson and Hamilton
SEO for SEM (see Figure 1). This concurs with a (2006) advise minimizing the use of graphics, forms
European report conducted by SEMPO (2007a), and frames, a significant number (22%) employ this
which found that French, Spanish and Italian technique. The use of these elements often restricts
advertisers focus more on the use of SEO over Paid spiders from reading the Website, thereby affecting
Search. In a similar type report on the US market, its Search Engine ranking.
SEMPO found that over three quarters of About half (46%) reported ‘refreshing page
respondents use SEO as their main form of SEM content’ as a regularly used method - although a
(SEMPO, 2007b). A third of those surveyed here useful ingredient, it seems slighted elevated when
report that they use a combination of SEO and Paid contrasted with the literature. Surprisingly, few cited
Search, while few respondents reported that they use ‘link development’ as an applied method to carry out
Paid Search only. This latter finding would suggest a SEO despite the exhortations of writers. While it has
clear view of Irish SMEs that Paid Search on its own been suggested that techniques should be used in
is not capable of delivering searchers to their tandem in pursuit of optimal SEO, the variability in
Website. The position would seem to be vindicated the logic and operation of each Search Engine’s
by a recent Amarach Consulting report showed a algorithm means differing techniques may be needed
clear majority (82%) of searchers admitting to rarely to achieve effectiveness across each Search Engine.
or never clicking on sponsored results when
conducting a Web search (Amarach, 2007). Use of keywords in title and
Q.6 SEO Techniques used
description tags
Use of Keywords on the home
80 73
page only
Use of keywords throughout 70
N u m b e r o f R e s p o n d e n ts

61
the whole website content 60
Rountinely refreshing web
50
content 42
Use of Cascading Stlye 40
Sheets 28
30
22
Keeping the likes of graphics 19 20
20
Link Development (Affiliate 10 4 2
links with other websites) 0
None of the above 1
Type of Technique
Other (please specify)

Figure 2: SEO Techniques.

Figure 1: Type of SEM Strategy used. 4.5 Paid Search Strategy Techniques
When respondents were also asked which SEM When SMEs were asked about their most used Paid
strategy they perceived to be most effective in Search technique, Paid Placement was ranked first
driving traffic to their company’s Website, most (see Table 1). Given the dominance of Google, this
(61%) indicated SEO while only 10% believed Paid finding echoes the literature’s claim that paid
Search to be so. This view concurs with Xing and placement is Google’s only Paid Search offering.
Lin (2004) who argue SEO is a more cost effective The much smaller numbers for Paid Inclusion (9%)
and viable long-term option. They posit that as Paid and Paid Submission (9%) appears to reflect the less
Search costs rise, SEO costs will remain constant. widespread use of Search Engines like Yahoo and
Only 11% thought Paid Search and SEO were as MSN who offer both types of Paid Search.
effective as each other.

342
 RESEARCHING SEARCH - A Study into Search Engine Marketing Practices in Ireland

Table 1: Paid Search Techniques used by Respondents. Table 2: Measures of SEO.

Total Responses Percent Frequency Percent

Pay for Placement 33 38% Measuring Link popularity 33 37%

Paid Inclusion 8 9% Setting Baselines & Measuring 11 12%

Contextual Search 7 8% Traffic Measurement 58 64%

Paid Submission 8 9% None 18 20%

Local Search 13 15% Other 6 6%

None of the above 25 29%

4.6.2 Paid Search Measurement
Other 3 3%
On measuring Paid Search campaigns, firms
indicated that both Traffic Measurement and CTR
The significant number that does not use any
were jointly the most used methods (see Table 3).
Paid Search technique (29%) most likely reflects the
Fewer use Conversion Rate despite the view by
earlier finding that reported SEO as the more used
Smith (2002) that conversion metrics are a very
and effective technique. The reasonable level of
important tool for estimating ROI. Smith discusses
Local Search is supportive of Laffey's (2007)
how traffic measurement used to be the most
observation that this method assists the smaller
appropriate measure, but the consumer conversion
business compete with the bigger players on a global
rate is now considered to be more important, where
scale and helps target specific geographic markets.
traffic measurement is more useful for analysing
SEO initiatives (Zhang and Dimitroff, 2005; Friesen,
4.6 Measuring SEM 2007).
4.6.1 SEO Measurement Table 3: Measures of Paid Search.
Traffic measurement was listed as the most used Respondents Percent
method for measuring SEO campaigns (see Table 2),
concurring with the views of Zhang and Dimitroff Conversion Rate 23 29%
(2005) and Friesen (2007). Traffic measurement is
the most popular measure for SEO in practice, using Click Through Rate 35 44%
tools such as Google Analytics. This was followed Traffic Measurement 36 45%
by ‘measuring link popularity’, as recommended by
Schultz and Fristedt (2005). Disappointingly, None 30 37%
‘setting baselines and measuring the impact’ did not
Other 3 4%
rate highly, despite its advocacy (Schultz and
Fristedt, 2005). This technique should be more
widely deployed since it is a true indication of a Reviewed at length in the literature are the large
ranking on the SERP at one point in time, versus numbers of Websites that do not measure their Paid
where it is at a future time having carried out SEO Search initiatives (Weischedel et al, 2005; Fain and
improvement measures. Worryingly, nearly a quarter Pedersen, 2006). This corresponds with the
said they do not use any method to assess their SEO significant proportion here (37%) that indicates that
initiatives. Given the current spend on SEM in they do not use any metrics to estimate their Paid
Ireland and the expected growth of this form of Search efforts. This finding may be explained by the
online advertising (Taylor, 2007), this is a matter for recognition throughout the literature that Paid
concern. Search lacks sound measurement tools (Weischedel
et al, 2005). Nonetheless, implementing a Paid
Search campaign and not measuring its outcome is
futile.

343
ICE-B 2008 - International Conference on e-Business

4.7 Effectiveness of SEM Measurement some confidence that firms are acting morally.
Nevertheless, as Search Engine marketers change
Opinions are divided evenly regarding how firm’s their Search Engine provider and develop their
believe they are effectively measuring their SEM strategies (say to less transparent operators like
activity. Nearly half believe that they are effectively MSN), an ethical policy should be carefully
or very effectively measuring it, however the considered.
remainder are not at all happy with their current SMEs were more indifferent to whether their
SEM metrics (see Figure 3). This view compounds users were aware that they were using Paid Search.
the argument in the literature regarding the However, to a great extent it is outside of their
availability of effective tools and justifies why so control since the Search Engine firms design and
many are discontented with their current deliver the SERPs. While Search Engine marketers
arrangement (Fain and Pedersen, 2006). might lobby those Search Engines that do not
disclose the difference between paid and free Search
Engine results, it is really up to consumer watchdogs
and government agencies to ensure consumer rights
are protected. Since there are a significant number of
respondents who were unaware of what constitutes
ethical SEO, these bodies have an important role in
educating firms on the consequences of using third
party marketing practices that are of questionable
ethical merit and of other SEO practices such as
Web spamming and unethical link development.

4.9 Challenges in SEM

Figure 3: Effectiveness of SEM Measurement.
When firms were asked about the main challenges
As argued by Smith (2002), businesses using encountered when undertaking SEM, the greatest
relevant metrics have a greater opportunity to one reported was competing for and achieving a
achieve a competitive advantage. Metrics also high rank on the SERP. While earlier firms felt they
provide valuable information about consumer’s were carrying out their SEO effectively, it still
online behaviour. Clearly much needs to be done to reflects the imperative for any SEM, which is to
improve the satisfaction of firms in assessing the deliver users to their Website. A related challenge
usefulness of their SEM effort. widely cited was the shortage of measurement tools
available for SEM. Closely related to achieving a
4.8 Ethics and SEM high ranking is the respondent’s concerns with the
Search Engines algorithm. This will always be an
Respondents overwhelmingly agree that ethical ongoing challenge for marketers as Search Engines
considerations are a very important part of both their never disclose their ranking algorithms.
SEO and Paid Search campaigns. In concordance A significant number cited knowledge of SEM
with this belief, there was general consensus that practices in general as an obstacle, indicative that
firms would have ethical issues in commissioning much remains to be understood about this form of
‘all forms’ of Paid Search. Respondents strongly marketing for SMEs in Ireland to fully realize the
reported that they would only commission Paid opportunities it offers to their broader marketing
Search that was transparent to users. These are very initiatives. Dedicating time, budget and resources
positive findings and would suggest that SMEs for were listed by many respondents as a barrier to
the most part act ethically when undertaking their adopting SEM. This finding again illustrates SME’s
SEM and have given consideration to ethical issues. lack of knowledge of SEM. Given the projected
Moxley et al (2004) view Paid Search growth of SEM as a marketing medium - dedicating
advertising as misleading for information searchers time, budget and resources is a crucial element in
as Search Engines often do not disclose the carrying out any SEM strategy. Managing keywords
difference between paid and organic results. So remain a major element for both Paid Search and
while the most frequently subscribed to Search SEO campaigns. In contrast, unethical competitor
Engine firm, Google, (who clearly display the practices (click fraud, Web spamming and unethical
difference between sponsored results and organic
results on the SERP) is used, searchers may have

344
 RESEARCHING SEARCH - A Study into Search Engine Marketing Practices in Ireland

link development) are not a major concern for statistics-ireland/#internet-usage-ireland> (Accessed

SMEs. 12th May 2007).
Asdemir, K. & Yahya, M. (2006) Legal and Strategic
Perspectives on Click Measurement [Internet],
Available from: <http://www.sempo.org/
5 CONCLUSIONS learning_center/editorials/clickmeasurement.pdf>
(Accessed 26th December 2006).
SEM is a growing strategy that firms use to improve Boughton, B. (2005) 'Search Engine Marketing
the visibility of their Website so that searchers are '.Perspectives in Business [Internet], Available from:
<http://www.stedwards.edu/business/pdf/Perspectives
delivered to the company Website to achieve their
V0201_06.pdf> (Accessed 12th December 2006).
informational or commercial goals. The findings Browne, G., Pitts, M. & Weatherbe, J. (2007) 'Cognitive
here suggest that Irish SMEs plan to spend more stopping rules for terminating information search in
time and resources on SEM in the future. It is also online tasks'. MIS Quarterly, 31, 1, 89-104.
clear that SEM strategies needs to be more formal in Dawson, A. & Hamilton, V. (2006) 'Optimising metadata
nature, in line with recommendations made to make high-value content more accessible to Google
elsewhere. While it is to be expected that an users'. Journal of Documentation, 62, 3, 307-327.
important business initiative begins in a casual and Economist (2006) 'Dancing with Google Spiders'.
experimental manner, very quickly the importance Economist, 378, 8468, 14.
Esparza, S. (2007) Don't risk your rankings with unethical
of aligning the effort with organisational strategy
SEO [Internet], Available from:
becomes evident. A fundamental mission of a SEM <http://www.searchengineguide.com/susanesparza/20
strategy, and clearly identified by respondents in this 07/05/dont_risk_your_rankings_with_u.html>
study, is to achieve a top position on the first page of (Accessed 22nd May 2007).
the SERP. Fain, D. & Pedersen, J. (2006) 'Sponsored Search: A Brief
The study also reveals that the great majority of History'. Bulletin of the American Society for
respondents use SEO or a combination of SEO and Information Science and Technology, 32, 2, 12-13.
Paid Search. Despite the general lack of Feng, J., Bhargava, H. & Pennock, D. (2007)
comprehensive tools to measure SEM effectiveness, 'Implementing Sponsored Search in Web Search
Engines: Computational Evaluation of Alternative
firms seem to have a clear preference for SEO at this
Mechanisms.' INFORMS Journal on Computing, 19,
time and perceive it to be more effective. To 1, 137-148.
reinforce the perceptions of SMEs it is evident that Friesen, T. (2007) How to Measure SEO Success
SEM initiatives need to be measured with [Internet], Available from:
appropriate tools so that outcomes can be fed back <http://publications.mediapost.com/index.cfm?fuseacti
into the strategic decision making process. What on=Articles.showArticleHomePage&art_aid=53187>
also emerges is that firms need to increasingly (Accessed 14th April 2007).
consider the behaviour and geographic location of Jansen, B. & Resnick, M. (2005) Examining Searcher
searchers as an integral part of their SEM strategy. Perceptions of and Interactions with Sponsored Search
in: Workshop on Sponsored Search Auctions at ACM
An encouraging finding from the study is that
Conference on Electronic Commerce (EC'05).
Irish SMEs adopt a positive ethical posture when Vancouver, BC, Canada.
undertaking SEM. On a cautionary note however, Jansen, B. & Molina, P. (2006) 'The effectiveness of Web
since Paid Search is, for now, less widely employed, search engines for retrieving relevant ecommerce
the more obvious ethical problems with it do not links'. Information Processing and Management, 42,
present themselves. Vigilance to give users 4, 1075-1098.
transparency in recognizing paid versus free search Jones, W. (2006) 'Microsoft and Google Vie for Virtual
results must be part of a firm’s SEM strategy is World Domination'. IEEE Spectrum, 43, 7, 16-18.
searchers are to remain confident in the integrity of Kumar, V. & Shah, D. (2004) 'Pushing and Pulling on the
Internet'. Marketing Research, 16, 1, 28-33.
Search Engine operatives.
Laffey, D. (2007) 'Paid Search: The innovation that
changed the Web'. Business Horizons, 50, 211-218.
Lempel, R. & Moran, S. (2000) 'The stochastic approach
REFERENCES for link structure analysis (SALSA and the TKC
effect).' Computer Networks, 33, (1-6), 387-401
Amarach (2007) Search Engine Usage in Ireland - The Mangalindan, M. (2003) 'Playing the Search Engine
Facts & Figures, Statistical Analysis March 2007. Game'. The Wall Street Journal, June 16th 2003.
[Internet], Available from: <http://www.redcardinal.ie/ Mentz, G. S. & Whiteside, R. (2003) 'The revenue impact
search-engine-optimisation/23-03-2007/search-engine- of Online Search Engine Marketing Technology for

345
ICE-B 2008 - International Conference on e-Business

Medical Clinics'. Journal of Information Technology Electronic Commerce. ACM, Fredericton, New
Impact, 3, 2, 101-110. Brunswick, Canada
Moxley, D., Blake, T. & Maze, S. (2004) 'Web search Zhang, J. & Dimitroff, A. (2005) 'The impact of webpage
engine advertising practices and their effect on library content characteristics on webpage visibility on search
service'. The Bottom Line: Managing Library engine results'. Information Processing and
Finances, 17, 2, 61-65. Management, 41, 3, 665-690.
Novak, T., Hoffman, D. & Yung, Y. (2000) 'Measuring Zimmer, M. (2006) 'The Value Implications of the
the customer experience in online Environments : A Practice of Paid Search'. Bulletin of the American
Structural modeling approach.' Marketing Science, 19, Society for Information Science and Technology, 32,
Winter 1, 22-42. 2, 23-25.
Oliva, R. (2004) 'Playing the Search. So little time, so Zwick, R., Rapport, A., Lo, A. & Muthukrishnan, A.
many sites.' Marketing Management, 13, 2, 48-51. (2003) 'Consumer sequential search: Not Enough or
Ostler, J. (2001) 'Search Engine Marketing Strategy'. NZ Too Much?' Marketing Science, 22, 4, 503-519.
Marketing 20, 6, 48.
Palmer, D. (2005) 'PopUps Cookies and SPAM: Toward a
Deeper Analysis of the Ethical Significance of Internet
Marketing Practices'. Journal of Business Ethics, 51,
1, 271-280.
Porter, M. (2001) 'Strategy and the Internet'. Harvard
Business Review, 79, 3, 62-78.
Schultz, J. & Fristedt, J. (2005) 'Calling All Search
Engines'. Association Management, Jun 2005,
Supplement, 8-13.
Seda, C. (2004) 'In the Click'. Entrepreneur, 32, 8, 60-63.
SEMPO (2007a) Search Marketing Efforts Across France,
Italy, & Spain. [Internet], Available from:
<http://www.sempo.org/learning_center/research/euro
pean_user_study_07> (Accessed April 20th 2007).
SEMPO (2007b) The State of Search Engine Marketing
2006 [Internet], Available from:
<http://www.sempo.org/news/releases/02-08-07>
(Accessed 12th May 2007).
Sen, R. (2005) 'Optimal Search Engine Marketing
Strategy'. International Journal of Electronic
Commerce, 10, 1, 9-25.
Smith, A. (2002) 'Loyalty and E-Marketing Issues'.
Quarterly Journal of Electronic Commerce, 3, 2, 149-
161.
Taylor, C. (2007) Irish firms stuck in search engine limbo
[Internet], Available from:
<http://www.electricnews.net/frontpage/news-
9920283.html > (Accessed 26th May 2007).
Thurow, S. (2007) Search engine marketing: paid
submission, pay-for-inclusion, and pay-for-placement
[Internet], Available from:
<http://www.grantasticdesigns.com/paidservices.html
> (Accessed 2nd February 2007).
Varadarajan, P. & Jayachandran, S. (1999) 'Marketing
Strategy: An Assessment of the State of the Field and
Outlook'. Journal of the Academy of Marketing
Science, 27, 2, 120-143.
Vine, R. (2004) 'The Business of Search Engines'
Information Outlook, 8, 2, 25-31.
Weischedel, B., Matear, S. & Deans, K. (2005) 'The use of
emetrics in strategic marketing decisions: a
preliminary investigation'. International Journal of
Internet Marketing and Advertising, 2, 1/2, 109-125.
Xing, B. & Lin, Z. (2004) The Impact of Search Engine
Optimisation on Online Advertising Market. in:
Proceedings of the Eight International Conference on

346
SHORT PAPERS
 AN OPTIMIZATION METHOD FOR REDEMPTION AND DUE
DATE MATCHING IN ASSIGNMENT OF ELECTRONIC
RECEIVABLES BY USING INTEGER LINEAR PROGRAMMING

Toshiyuki Moritsu
Systems Development Laboratory, Hitachi Ltd. 890, Kashimada, Saiwai-ku, Kawasaki, Kanagawa 212–8567, Japan
[email protected]

Norihisa Komoda
Graduate School of Information Science and Technologies, Osaka University 2–1
Yamadaoka, Suita, Osaka 565-0871, Japan
[email protected]

Keywords: Electronic Receivables, Integer Linear Programming, Finance, Due Date Matching, Optimization.

Abstract: This paper shows an optimization method for redemption and due date matching which assigns the receiv-
ables to the payments date under the pre-defined constraints which related companies specified. Our proposed
method determines the pairs of the receivables and the payment with proper new settlement date which closes
the fluid assets of companies to their target amounts by extending/shortening the redemption/due date. This pa-
per shows that this matching program is categorized in integer linear programming. By applying this matching
mechanism, transferors could utilize credit of issuers and also reduce fluid assets for payments. Effectiveness
of our optimization method is shown by executing simulation which emulates the issuing and receiving status
of receivables in Japanese companies.

1 INTRODUCTION pany B has a poor credit rating and the electronic

receivables issued by company B have not been re-
In recent years, various countries have made progress ceived by the transferee (company C), then if we sup-
in the development of systems for handling credit pose that company A has a high credit rating then
transactions electronically, including EBPP/EIPP company B can pay by transferring credit issued by
(Electronic Bill/Invoice Presentation and Payment) in company A. This supplementation of reliability by a
the United States (Fairchild, 2003) and e-billing sys- third party is particularly useful as a way of utilizing
tems in South Korea (KFTC, 2002). Japan is now also the credit rating of a parent company in the financing
expecting an increase in this type of business, and in of subcontractor corporations in business affiliations
June 2007 the Japanese parliament approved the Elec- that account for approximately 60% of such arrange-
tronically Receivables Legislation (FSA, 2005) which ments in Japan (JSBRI, 2007).
is due to go into force at the end of 2008. Electron- However, it has been pointed out that previous
ically receivables are a new form of credit whereby nominative claims and credits to payment are limited
electronic registrations at organizations that register in terms of the opportunities for utilizing one’s credit
electronic receivables become only requirement for in the assignment of claims. One possible reason for
the generation and transfer of credit, and are expected this is that the payment conditions (amount payable
to utilized as a means for the exchange of nominative and due date) and credit redemption conditions are
claims and credits to payment on which contract doc- not exactly same. Since electronic receivables make
uments and printed bonds are based (Ikeda, 2006). it easy to rewrite the conditions electronically, it is rel-
The assignment of claims allocated to the payment atively simple to accommodate differences in sums by
of electronic receivables owned by a business has the splitting the electronic receivables. This has actually
merit of allowing the transferor to utilize the remit- been implemented in the book entry transfer system
ter’s credit rating (Oogaki, 2006). Specifically, when for short term corporate bonds in Japan. On the other
the transferor (company B) owns electronic receiv- hand, in cases where the due dates do not match, it is
ables issued by a issuer (company A), even if com- necessary to adjust the gains and losses of the inter-

349
ICE-B 2008 - International Conference on e-Business

ested parties (issuer, transferor, transferee) before al- 2 METHOD FOR MATCHING
tering the conditions. This due date modification has THE DUE DATES OF
the side benefits of allowing businesses to make effec-
tive use of surplus floating assets (referred to simply ELECTRONIC RECEIVABLES
as “assets” in the following) and providing a means
for supplementing shortfalls in assets. It is thus nec- In this section we discuss a method for matching elec-
essary to reach an agreement on items such as what tronic receivables and payments based on the prior
the interest rate should be and by how much the term registration of due date alteration conditions which is
should be extended or contracted. a premise of this study. The processing flow is illus-
It is difficult to make an agreement between the trated in Fig. 1. First, the participants register their re-
interested parties regarding the change in due date by spective asset fluctuation targets and due date adjust-
executing work flow between them because it can take ment criteria (1). The payment registrations are then
much time to reach an agreement if consideration is accepted (2), the combination of electronic receiv-
given to factors such as changes of circumstances re- ables and payments is determined based on these in-
sulting from conflicting conditions. formation (3), and approval is obtained for the match-
Therefore in this study it is envisaged that making ing results. The due date adjustment criteria are as-
alterations to the due date of electronic receivables in sumed to consist of a range of possible due dates for
an effective manner requires a mechanism whereby redemption payments, and interest conditions. In the
the participants first register assets fluctuation targets following, we will discuss the asset fluctuation targets
and variation conditions whereby changes in due date specified by the participants, the possible range of re-
can be tolerated, and then combinations of payments demption/payment due dates, and the interest condi-
and electronic receivables are determined to satisfy tions.
these constraints. For this study we therefore propose • Asset Fluctuation Targets.
a matching scheme that optimizes combinations of Each participant specifies when and how much
payments and electronic receivables so as to achieve they would like the current assets to fluctuate. For
the best possible assets fluctuation targets for all par- example, the estimation of future changes in cur-
ticipants. In particular, in this matching of electronic rent assets is prepared from predictions of the pay-
receivables and payments, the level of assets target ment and receipt of electronic receivables (and
achievement of the participants depends not only on other settlements), and by setting the target hold-
which electronic receivables should be combined with ings of current assets at each future timing, the
which payments, but also on how the new due date for fluctuation targets are set according to the differ-
the redemption/payment of electronic receivables and ence between the target holdings and estimated
payments is set and matched. We therefore formula- transitions.
rize the issue of combining electronic receivables and
payments by including the selection of a new due date • Possible Range of Redemption/Payment due
whose selection range is determined by these combi- Dates.
nations, and the results can be treated as a problem If necessary, the participants specify a range for
of integer linear programming. Also, by simulating the extension or contraction of payment due dates
the transaction of electronic receivables under condi- and/or redemption due dates.
tions simulating the issue of corporate bills and ac- • Interest Conditions.
counts receivable in Japan, we first verify the extent The participants specify a lower limit of interest
to which the assignment of claim is promoted by the to be accepted in cases where the redemption due
due date matching function. We also verify the opti- date is extended or the payment period is reduced,
mization effects by comparing the optimized match- and conversely an upper limit of interest to be paid
ing of electronic receivables and payments with some in cases where the redemption due date is brought
other combination methods. forward or the payment due date is extended. Note
that the interest is set according to the debtor’s
credit rating.

350
 AN OPTIMIZATION METHOD FOR REDEMPTION AND DUE DATE MATCHING IN ASSIGNMENT OF
ELECTRONIC RECEIVABLES BY USING INTEGER LINEAR PROGRAMMING

3 FORMULARIZATION OF THE upper limit of interest according her/his floating assets

needs and the lower limit according to credit rating of
PROBLEM OF MATCHING issuer of Ri jm and transferors (who transfer Ri jm before
ELECTRONIC RECEIVABLES transferor j).
AND PAYMENTS IP( j, P jkn , f )/ IR( j, P jkn , f ). Upper/lower limit of interest
to be paid/received by transferor j in cases where the
payment date of payment P jkn is contracted/extended
In this section we will devise a formula for the prob- to date f . Transferor j decides the upper/lower limit
lem of matching electronic receivables and payments, same as in case of IP( j, Ri jm , f )/ IR( j, Ri jm , f ).
and we will show that this problem can be expressed
IP(k, P jkn , f )/ IR(k, P jkn , f ). Upper limit of interest to be
as a problem in integer linear programming (Schri- paid/received by transferee k in cases where the pay-
jver, 1986)(Aardal et al., 2005). ment date of payment P jkn is extended/contracted to
The problem of matching electronic receivables date f . Transferee k decides the upper/lower limit same
and payments is expressed using the following nota- as in case of IP( j, Ri jm , f )/ IR( j, Ri jm , f ).
tion. Cxe . Fluctuation target sum of current assets of participant
Constants. x on date e
Ri jm . The m-th electronic receivable issued by issuer i and Variables.
held by transferor (current holder) j. v(Ri jm , P jkn , f ) . Transferred sum whereby electronic re-
D(Ri jm ). Redemption date of electronic receivable Ri jm ceivable Ri jm is allocated to payment P jkn at a new due
DE(Ri jm ). Upper limit of the extension of the redemption date f
date of electronic receivables Ri jm set by issuer i The Objective function of this matching problem
DS(Ri jm ). Upper limit of the reduction of the redemption can be expressed as shown in Formula 1.
date of electronic receivables Ri jm set by issuer i
V (Ri jm ). Redemption sum of electronic receivable Ri jm Objective Function.
Max(Σi jkmn f
P jkn . n-th payment made to transferee k by transferor j
|D(Ri jm ) − D(P jkn )| · v(Ri jm , P jkn , f ) (1)
D(P jkn ). Payment date of payment P jkn
DE(P jkn ). Upper limit of the extension of the payment date
of payment P jkn set by transferee k The target of this problem is to make the best pos-
sible effort to ensure that the fluctuation targets of all
DS(P jkn ). Upper limit of the reduction of the payment date
participants are satisfied by the fluctuation of current
of payment P jkn set by transferee k
assets resulting from changes to the due dates of elec-
V (P jkn ). Payment sum of electronic receivable Ri jm tronic receivables and payments. To achieve this goal,
IP(i, Ri jm , f )/ IR(i, Ri jm , f ). Upper/lower limit of interest it is preferable to mach the combinations of payments
to be paid/received by issuer i in cases where the and electronic receivables having longer discrepan-
redemption date of electronic receivable Ri jm is ex- cies of due dates with the greatest transferred sums,
tended/contracted to date f . Issuer i decide the up-
per/lower limit of interest according floating assets within the range of the following constraints. As a re-
needs. sult, the target function can be expressed as the sum
total of the values obtained by multiplying the trans-
IP( j, Ri jm , f )/ IR( j, Ri jm , f ). Upper/lower limit of interest
to be paid/received by transferor j in cases where the ferred sums by the discrepancies between the redemp-
redemption date of electronic receivable Ri jm is con- tion due dates and payment due dates, and can be ex-
tracted/extended to date f . Transferor j decides the pressed by formula 1.
Meanwhile, the constraint formulae can be
expressed by Formulae 2 through 15 as follows:

Constraint formulae.
v(Ri jm , P jkn , f ) ≥ 0 (2)
V (Ri jm ) ≥ Σkn f v(Ri jm , P jkn , f ) (3)
V (P jkn ) ≥ Σim f v(Ri jm , P jkn , f ) (4)
Cxe ≥
Σ jkmn f v(Rx jm , P jkn , f ) +
Σi jmn f v(Ri jm , P jxn , f ) (5)
Figure 1: Process of Due Date Adjustment between Elec- where Cxe ≥ 0 && D(Rx jm ) ≤ e < f &&
tronic Receivables and Payments. f < e ≤ D(P jxn )

351
ICE-B 2008 - International Conference on e-Business

Cxe ≤ the payable sum cannot be fully allocated by the trans-

−Σ jkmn f v(Rx jm , P jkn , f ) − fer of electronic receivables alone).
Σi jmn f v(Ri jm , P jxn , f ) (6) Constraint formula 5 shows that when the fluctu-
where Cxe < 0 && f < e ≤ D(Rx jm ) && ation target of the current assets of participant x on
D(P jxn ) ≤ e < f day e is increased, the increase target of participant x
on day e is greater than the sum of the total amount
v(Ri jm , P jkn , f ) = 0 (7) of electronic receivables drawn by participant x for an
extension astride day e and the total amount of pay-
where ( f ≤ D(Ri jm ) || D(P jkm ) ≤ f ) &&
ments in which participant x is the transferee for a re-
D(Ri jm ) < D(P jkm ) (8)
duction astride day e (a participant’s current assets are
where ( f ≤ D(P jkm ) || D(Ri jm ) ≤ f ) && increased by the extension of redemption due dates on
D(P jkm ) < D(Rx jm ) (9) electronic receivables drawn by the same participant,
where f < DS(Ri jm ) || f > DE(Ri jm ) || or by the reduction in payment due date of payments
received by the participant).
f < DS(P jkn ) || f > DE(P jkn ) (10)
Constraint formula 6 is the converse of constraint
where (IR( j, Ri jm , f ) < IP(i, Ri jm , f ) || formula 5, and shows that when the fluctuation target
IR( j, P jkn, f ) < IP(k, P jkn , f )) && of the current assets of participant x on day e is de-
D(Ri jm ) < D(P jkm ) (11) creased, the reduction target of participant x on day
where (IR(i, Ri jm , f ) < IP( j, Ri jm , f ) || e is less than the sum of the total amount of elec-
IR(k, P jkn , f ) < IP( j, P jkn , f )) &&
tronic receivables drawn by participant x for a reduc-
tion astride day e multiplied by the negative value of
D(P jkm ) < D(Rx jm ) (12)
the total amount of payments in which participant x is
where Cie > 0 && f < e ≤ D(Ri jm ) (13) the transferee for an extension astride day e (a partic-
where Cie < 0 && D(Ri jm ) ≤ e < f (14) ipant’s current assets are reduced by the reduction of
redemption due dates on electronic receivables drawn
where Cke > 0 && D(P jkn ) ≤ e < f (15) by the same participant, or by the extension in pay-
where Cke < 0 && f < e ≤ D(R jkn ) (16) ment due date of payments received by the partici-
pant).
The constraint formulae consist of the following three Constraint formulae 7–15 define the possible
types: range of combinations of electronic receivables and
1. Constraints in which the possible range of trans- payments. Specifically, the range for which reverse
ferred sums is predetermined (formulae 2–4) combinations are not possible is defined as a trans-
ferred amount of 0 in formula 7, and the applicable
2. Constraints in which fluctuations of current assets ranges are specified by formulae 8–16.
arising from transfers associated with changes of Range 8 and range 9 specify that electronic receiv-
due dates are kept within the fluctuation target ables and payments cannot be matched unless the new
(formulae 5–6) due date after modification is between the redemp-
3. Constraints in which the electronic receivables tion date of the electronic receivables and the date on
and payments that can be combined are predeter- which the payment is due.
mined (formulae 7–16) Range 10 specifies that electronic receivables and
Constraint formula 2 imposes the condition that payments cannot be matched unless the new due date
the transferred sum must be positive. after modification is within the redemption dates of
Constraint formula 3 imposes the condition that the electronic receivables and the possible range of
the sum total of sums transferred when electronic re- payment due dates of the payment.
ceivables are transferred in separate parts does not Range 11 specifies that in cases where the redemp-
exceed the sum of the original electronic receivables tion due date of electronic receivables is before the
(but not necessarily equal since the splitting of elec- date on which payment is due, matching of the elec-
tronic receivables may leave parts that are not suitable tronic receivables and payments is not possible unless
for transfer). the interest to be paid by the receiver due to extension
Constraint formula 4 imposes the condition that of the redemption due date on the electronic receiv-
the sum total of payments made by the transfer ables is less than the interest required by the transferor
of electronic receivables does not exceed the total due to this extension, and the interest paid by the as-
amount of be paid (but not necessarily equal since signee due to a reduction in the redemption due date
new issues and combinations are made in cases where is smaller than the interest required by the transferor

352
 AN OPTIMIZATION METHOD FOR REDEMPTION AND DUE DATE MATCHING IN ASSIGNMENT OF
ELECTRONIC RECEIVABLES BY USING INTEGER LINEAR PROGRAMMING

due to this reduction. Tax Agency and from settlement trends for 2003 pub-
Range 12 relates to the case where the redemp- lished by the Bank of Japan, including the average
tion due date comes after the date on which the pay- sales figures for Japanese corporations, payable liabil-
ment is due, and is the converse of range 11 where the ities (accounts payable, bills payable), received credit
payer/payee relationships of interest payments by the (accounts receivable, bills receivable), average sum
debtor, transferor and transferee are reversed. of bills cleared, and average sum of accounts receiv-
Range 13 specifies that when there is an increase able. The conditions for the issue of electronic receiv-
in the fluctuation target for current assets on day e, ables were assumed based on this financial informa-
it is not possible to perform matching with electronic tion. Specifically, we made assumptions regarding the
receivables issued by the issuer involving a change in average redemption period of electronic receivables,
the redemption due date so as to reduce the redemp- the average frequency of issue and the average sum.
tion due date astride day e (a change that depletes cur- The respective calculation formulae are shown below.
rent assets). Since this means it is possible to exceed
the target if only one of the combinations is agreed • Average redemption period of electronic receiv-
upon, in this formularization the fact that fluctuation ables = receivable credit / sales × 365
targets cannot be exceeded is added as a premise to
this constraint. • Average number of electronic receivables issued
Range 14 relates to the converse of range 13 in = (bills receivable / average sum of bills cleared
cases where the current formula fluctuation target of + accounts receivable / average sum of accounts
the issuer is reduced. receivable) / 365
Ranges 15 and 16 are the same as ranges 13 and
14 where the issuer imposes limits on the redemption • Average sum of credit = receivable credit / (bills
due data by means of restrictions on changes to the receivable / average sum of bills cleared + ac-
payment due date in transferee k. counts receivable / average sum of accounts re-
With regard to the definition contents of the above ceivable)
target functions and constant formulae, first, target
function 1 is the maximization of a primary function In the simulation, the number of companies was
with v(Ri jm , Pjkn , f ) as a variable, and it takes an inte- taken to be 260 (one thousandth of the actual num-
ger value. Also, the constraint formulae are all first- ber of businesses), and measurements were performed
order inequalities that take v(Ri jm , Pjkn , f ) as a vari- by repeating the transactions over two years. In real
able. This problem can therefore be classified as an situations, not necessarily all the credit is replaced
integer linear programming problem. with electronic receivables, and not necessarily all
the electronic receivables are subject to being trans-
ferred, so the simulation was performed by making a
4 EVALUATION few changes to the ratio of transferable sums with re-
gard to the credit sums of the electronic receivables
The effects of changes in due date and the effects belonging to a business. The simulation environment
of optimizing the combinations of electronic receiv- parameters were as follows: MPU: Xeon1 2.8 GHz,
ables and payments were verified by simulation. This Memory: 3 GByte, Windows XP2 , JDK 1.6.0 01 013 ,
section discusses the preconditions under which the LpSolve 5.5.0.10(Berkelaar et al., 2004). In the evalu-
simulation was conducted, and then presents the mea- ation results shown in the next section, measurements
surement results. were also performed by varying some conditions of
the other parameters (variation in redemption peri-
4.1 Preconditions ods of electronic receivables, variation in frequency
of issue of electronic receivables, variation in mone-
tary value of electronic receivables, number of com-
There is currently no statistical information relating
panies simulated), but the effects of these changes
to the issue of electronic receivables. We therefore
were smaller than those of the parameters shown in
performed the simulation by assuming conditions for
Table 1 and thus these results are omitted.
the issue of electronic receivables based on finan-
cial information from Japanese businesses. Table 1
shows the financial information and the conditions 1 Xeon is a registered trademark of Intel Corporation.
for the issue of electronic receivables assumed in this 2 Windows XP is a registered trademark of Microsoft
simulation. The financial information was sourced Corporation.
from corporate statistics published by the National 3 Java is a trademark of Sun Microsystems, Inc.

353
ICE-B 2008 - International Conference on e-Business

Table 1: Financial Statement of Average Japanese Company

and Assumed Issue Condition of Electronic Receivables.

Total Sales(kY=) 20,483

Total Receivables(kY=) 82,887
⁀
Account Receivables(kY=) 66,287
⁀
Note Receivalbes(kY= 16,600
Average Amount of Account Receivables 2,000
Average Amount of Note Receivables 4,000
Assumed Average Redemption Period 59
of Electronic Receivables(Day)
Assumed Average Issue Cycle of 0.102
Electronic Receivables(Times/Day)
Assumed Average Amount of Electronic 2222
Receivables(kY =) Figure 3: Ratio of Adjusted Asset to Target.

can be assigned in the transfer of electronic receiv-

ables is about 3.3% of the total. On the other hand,
when the redemption due date and payment due date
are set with the ability to be varied forwards or back-
wards by up to 3 days, the ratio of the payment in
the transfer increases to 16.7% (about 5 times larger)
when all the electronic receivables are transferred.
Also, even when the ratio of the electronic receivables
to be transferred is reduced to about 1/32 of the to-
tal electronic receivables, an increase of 12.3% (about
3.7 times) is seen. These results show that by provid-
Figure 2: Ratio of Payments by Transfer. ing a framework for making adjustments of a few days
in the redemption due date or payment due date, each
business can make a relatively large improvement to
4.2 Measurement Results the ratio of payments made by transfers.
The optimization results are shown next. Figure 3
We will first use the due date modification function shows the extent to which it is possible to achieve the
to investigate the extent to which payments are pro- fluctuation targets of current assets set by the partici-
moted by transfers. Next, by optimizing the combi- pants for each method of matching a number of elec-
nations of payments and electronic receivables, we tronic receivables and payments. The specific match-
will verify the extent to which it is possible to achieve ing methods used to make these measurements were
the participants’ fluctuation targets. Finally, we will as follows:
investigate the computational load required for opti-
mization. • Randomly ordered Matching.
The graph in Fig. 2 shows the ratio of all pay- Matching is performed in random order from
ments in which payment was made by assignment of among candidates consisting of electronic receiv-
claim. The horizontal axis shows the number of days ables and payments that are capable of being com-
by which the payment due date and redemption due bined.
date can be varied in either direction. The multiple • Matching in Order of Effects on the Level of
measurement results in this graph correspond to mea- Achievement.
surements made while varying the criteria regarding
Matching is performed in order of the magnitude
the extent to which the electronic receivables owned
of how the achievement of the fluctuation targets
by a business are transferred on a monetary basis.
is affected by candidates consisting of electronic
This evaluation shows the results obtained when the
receivables and payments that are capable of be-
matching of electronic receivables and payments is
ing combined.
optimized.
In these results, the case where the modifiable date • Matching based on Optimal Pattern Searching.
is 0 corresponds to the case where no changes can be Matching is performed by searching for optimal
made to the redemption due date of the electronic re- patterns that are closer overall to the fluctuation
ceivables. In this case, the ratio of the payment that targets.

354
 AN OPTIMIZATION METHOD FOR REDEMPTION AND DUE DATE MATCHING IN ASSIGNMENT OF
ELECTRONIC RECEIVABLES BY USING INTEGER LINEAR PROGRAMMING

Table 2: Fragmented Number of the Electronic Receivables.

Average Standard Max

Deviation
Random Matching 2.09 1.05 10
Impact to Asset Target 1.38 0.57 7
Prior Matching
Optimized Matching 2.06 1.93 9

electronic receivables are fragmented by the match-

ing of due dates. The number of fragments is a value
Figure 4: Effect of Optimization.
that shows on average how many electronic receiv-
ables the original electronic receivables are divided
into at the time of redemption. The average number
of fragments was 2.08 when matching was performed
in random order, 1.38 when matching was performed
in order of the effects on the effects on the level of
target achievement, and 2.06 when matching was per-
formed based on optimal pattern searching. The rea-
son for the small number of fragments obtained when
Figure 5: Calculation Time for Daily Optimization. matching in order of the effects on the effects on the
level of target achievement is thought to be because in
this algorithm, matching is performed preferentially
The horizontal axis in the graph of Fig. 3 shows on groups of electronic receivables and payments in-
the number of days by which the payment due date volving larger sums and greater differences in due
and redemption due date can be varied in either di- dates, and these larger sums suppress the fine frag-
rection. The vertical axis shows the extent to which mentation of electronic receivables.
the fluctuation targets preset by each business could
be achieved. With regard to these measurements,
the fluctuation targets of the businesses were set ran-
domly in the range of the sums of electronic receiv- 5 CONCLUSIONS
ables owned by these businesses. Figure 4 compares
the improvement rate of optimization between ran- In this study, we have proposed a scheme for optimiz-
dom matching and matching based on the effects on ing the matching of redemption due dates and pay-
the level of achievement. These results show that al- ment due dates for electronic receivables to promote
though there is some degree of variation when there the transfer of electronic receivables. Specifically,
are few modifiable days, the average ratio of improve- we have demonstrated an optimization scheme that
ment is approximately 18.1% for matching based on makes every effort to make the fluctuations in the cur-
the effects on the level of achievement, and approxi- rent assets of each business (caused by changes of due
mately 19.6% for random matching. date) approach their fluctuation targets based on cri-
Next, the computation time needed to optimize the teria specified by each business regarding the fluctua-
combinations of electronic receivables and payments tion targets of current assets, the range of alterations
is investigated using the graph shown in Fig. 5. In this to due dates, and the rates of interest. We have also
graph, the number of businesses is shown on the hori- shown that this problem can be classified as a type
zontal axis and the time required for the optimization of integer linear programming problem. By perform-
computations per day is shown on the vertical axis. ing simulations based on the financial circumstances
In a simulation of 260 companies, the computation of average Japanese businesses, we have shown that
time required for optimization was 113 seconds per this technique is capable of promoting transfers in-
day on average. This remains future work to make volving changes of due date and improving the degree
this method in practice because this simulation done to which fluctuation targets are achieved by optimiza-
with 1/1000 of actual number of companies and the tion.
computation time glows exponentially according to One issue for further study is the problem of elec-
increase the number of companies. We discuss this tronic receivables being finely fragmented by the op-
problem in section 5. timization process. In the method of this study, if the
Finally, Table 2 shows the extent to which the fluctuation targets of the participants’ assets are im-

355
ICE-B 2008 - International Conference on e-Business

proved, then the electronic receivables can be arbi- Institute. http://www.chusho.meti.go.jp/sme english/
trarily fragmented and a number of payments will be whitepaper/whitepaper.html.
assigned to a number of new due dates. This fragmen- KFTC (2002). Check clearings. KFTC 2002, pages 18–19.
tation of electronic receivables leads to increased ad- http://www.kftc.or.kr/english/statistics/KFTCAR.pdf.
ministration costs, so to actually put this method into Oogaki, H. (2006). Finance in secondary loan market
practice, it is important to investigate how to control and electronic receivables (in japanese). Banking
the fragmentation of electronic receivables. Law Journal, Kinzai Institute for Financial Affairs,
1781:20–28.
A second issue is that of partitioning the opti-
mization regions. In this study, optimization was per- Schrijver, A. (1986). Theory of Linear and Integer Pro-
formed by using a single target function to represent gramming. Wiley-Interscience.
the criteria of all the companies concerned, but this is Shimamura, A., Moritsu, T., and Someya, H. (2006). Deliv-
inefficient with regard to increasing the scale of busi- ery path length and holding tree minimization method
of securities delivery among the registration agencies
nesses to which the method is applied. In practice, connected as non-tree (in japanese). IEEJ Trans., 126-
there is considered to be some degree of locality in C(4):506–512.
the transaction relationships between businesses, so if
groups of businesses can be split into suitable ranges,
then it should be possible to split the optimization
problem into multiple sub-problems with fewer vari-
ables and constraints. In this way it should be possi-
ble to reduce the computational cost and speed up the
computation time by employing parallel processing.
A third issue is that not necessarily all of the opti-
mal matching results demonstrated by this algorithm
are the best matching results from the viewpoint of
each individual business. For actual operations, an
important issue is therefore to somehow present alter-
native proposals in cases where a business rejects the
matching results.

REFERENCES
Aardal, K., Nemhauser, G., and Weismantel, R. (2005). Op-
timization: Handbooks in Operations Research and
Management Science, volume 12. Elsevier.
Berkelaar, M., Eikland, K., and Notebaert, P. (2004). lp-
Solve: Open Source (Mixed-Integer) Linear Program-
ming System. GNU LGPL (Lesser General Public Li-
cense).
Fairchild, A. (2003). Possible Distinermediation: What
Role for Banks in Electronic Invoicing(EIPP).
In 16th Bled eCommerce Conference eTransfor-
mation, pages 107–118. http://domino.fov.uni-
mb.si/ECOMFrames.nsf/pages/bled2003.
FSA (2005). Summary of Discussion on Electronic
Receivable Legislation from a Financial System
Perspective. Newsletter of Financial Services
Agency, The Japanese Government, pages 2–5.
http://www.fsa.go.jp/en/newsletter.
Ikeda, M. (2006). Electronic receivables - examination
and essay for legislation - (in japanese). Banking
Law Journal, Kinzai Institute for Financial Affairs,
1781:8–19.
JSBRI (2007). White Paper on Small and Medium En-
terprises in Japan. Japan Small Business Research

356
 E-LEARNING FOR NEW GRADUATE EMPLOYEES
Another Function of e-Learning for New Graduate Employees of Japanese
and Korean Companies

Jiro Usugami
Department of Economics, Oita University, 700 Dannoharu, Oita City, 870-1192, Japan
[email protected]

Keywords: New graduate employee, human capital development, employee retention, pre-entry training program.

Abstract: Japanese companies tend to consider that e-learning is important for a reason other than human capital
development in new graduate employee training and development programs, namely for improving
employee retention. Two remarkable features of the Japanese employment system are the seasonally
scheduled hiring of new graduates and company commitment to the T&D programs for new graduate
employees before and after their official hiring. Today, large- and medium-sized Japanese companies make
their offers to candidates nearly one year prior to their graduation from university or college. Many of these
companies also provide various T&D programs, including e-learning, for the potential new graduate
employees until their graduation. We call these “pre-entry training” programs. They are distinguished from
internship programs in that pre-entry training programs are provided only for those students who are due to
be officially hired the next fiscal year. Besides Japanese companies, a number of foreign companies in
Japan and some Korean companies provide pre-entry training programs that utilize e-learning. This study
examines, based on case studies, the hypothesis that e-learning in the pre-entry training programs for new
graduate employees has the function of retention as well as human capital development.

1 INTRODUCTION function of retaining potential new graduate

employees as well as developing human capital,
Many Japanese companies utilize e-learning in the based on case studies.
training and development programs for new In Japan and Korea, there is intense competition
graduate employees, before and after their official among companies hiring new graduates, which
hiring. makes employee retention a critical issue. Japanese
New graduate hiring is done seasonally in Japan. companies need to retain potential new graduate
Large- and medium-sized companies make their employees for about one year prior to official hiring,
offers to candidates nearly one year prior to their and tend to consider that e-learning in the pre-entry
graduation from university or college. During this training programs is an effective retention strategy.
long period until graduation, many of these This study comprises the following chapters.
companies provide various T&D programs, Chapter 2 provides an overview of the new graduate
including e-learning, for the potential new graduate hiring schedule in Japan and typical e-learning
employees, which we call “pre-entry training” menus in the pre-entry training programs. Chapter 3
programs. They are distinguished from internship presents a theoretical model and hypothesis on the
programs in that pre-entry training programs are function of e-learning in the pre-entry training
provided only for those students who are due to be programs. Chapter 4 analyzes the objectives of
officially hired the next fiscal year. companies utilizing e-learning in the pre-entry
Besides Japanese companies, a number of training programs for potential new graduate
foreign companies in Japan and some Korean employees, based on case studies. Chapter 5
companies also provide pre-entry training programs provides conclusions.
that utilize e-learning.
This study examines the hypothesis that e-
learning in the pre-entry training programs has the

357
ICE-B 2008 - International Conference on e-Business

2 HIRING SCHEDULE & development (bookkeeping, marketing, etc.), and

E-LEARNING business etiquette and communication (business
greeting protocol, telephone etiquette, etc.). There
are also orientation programs that include company
Large- and medium-sized Japanese companies tend
overviews and support programs for acquiring
to hire a certain number of new graduates every
business certificates.
April when the new fiscal year begins. By the time
they are officially hired, many of these new graduate
employees have experienced a variety of e-learning
courses provided by the companies. 3 MODEL AND HYPOTHESIS
This chapter presents an overview of the new
graduate hiring schedule in Japan and typical e- What is the function of e-learning in the pre-entry
learning menus in the pre-entry training programs. training programs for new graduate employees?
Many large- and medium-sized Japanese
2.1 New Graduate Hiring Schedule in companies provide Off-JT programs such as class
Japan instruction, role-play training and camp training
after officially hiring new graduates. On the other
The new graduate hiring season starts more than one hand, in the pre-entry training programs, e-learning
year prior to graduation. Students graduate in March tends to be easier to provide since the trainees are
and the Japanese fiscal and academic year begins in still attending university or college at various
April. Hiring tends to be scheduled as follows. locations including overseas. Therefore, companies
First, from October to December, many Japanese recognize the effectiveness of e-learning.
companies announce their new graduate recruiting to This chapter presents a theoretical model and
students in their third year of university or college. hypothesis on the function of e-learning in the pre-
From December to March of the following year, job entry training programs.
fairs are held in several big cities in Japan.
Companies also provide career guidance and 3.1 Theoretical Model
conduct on-campus recruiting. From January to
April, written tests and job interviews are conducted The principal objective of e-learning is human
for candidates in their last year of school. capital development. According to Becker (1975),
From April to June, companies select the new the major idea underlying human capital theory is
graduate employees for the next fiscal year, and that an individual’s performance is determined by
make their offers to the candidates. During a period his/her stock of knowledge and productive skills.
of about six months from October of the students’ Applying this theory to the case of new graduate
last year to March when they graduate, many of employees, those who have accumulated some stock
these companies provide pre-entry training programs of knowledge and skills by the time they are
that include e-learning courses, events, classroom officially hired would demonstrate good work
instruction, camp training, and on-the-job training performance from the outset. In addition, providing
for those students who are due to be officially hired Off-JT programs after their hiring would be efficient
after graduation. According to a survey conducted and performance on the job would be high.
by the Institute of Labor Administration, over 60% Figure 1 shows a basic human capital
of large- and medium-sized companies provided pre- development model. Dotted and solid lines indicate,
entry training programs in 2005. respectively, the performance of new graduate
Then, on the first business day in April of the employees who experienced Off-JT after being hired,
new fiscal year, initiation ceremonies are held and and those who experienced only OJT.
new graduates are officially hired. Figure 2 is a human capital development model
for a case in which pre-entry training programs are
2.2 Typical e-Learning Menus provided. The broken line indicates the performance
of new graduate employees who experienced pre-
E-learning in the pre-entry training programs has a entry training before being hired and OJT afterwards.
variety of menus. Large enterprises tend to develop The movement of the dotted line indicates that pre-
their own e-learning courses. Typical menus of entry training shortens the time for Off-JT programs
outsourced e-learning are language education after hiring and improves its efficiency.
(English, Chinese, etc.), PC skills development This model suggests that companies could
(Word, Excel, PowerPoint, etc.), business skills accelerate the human capital development of new

358
E-LEARNING FOR NEW GRADUATE EMPLOYEES - Another Function of e-Learning for New Graduate Employees of
Japanese and Korean Companies

graduate employees, improve the efficiency of T&D accelerating human capital development
programs for them, and gain high performance and improving the efficiency of T&D
sooner by making e-learning programs available to programs for them.
all trainees wherever they are located. However, the
model does not refer to the relationship between e-
learning and employee retention. 4 ANALYSES
Performance
What are the actual objectives of the companies that
utilize e-learning in the pre-entry training programs?
Off-JT Usugami (2006) conducted a survey in 2005 and
found that Japanese and Korean companies of 137
OJT samples aimed to maintain the motivation of
potential new graduate employees, improve their
Time
u
interpersonal skills and retain them by providing
→ Officially hired pre-entry training programs that include e-learning.
Figure 1: Basic Human Capital Development.
Hanada & GaiaX (2008) conducted a survey for
companies that provided pre-entry training programs
P Pre & Off-JT including SNS (Social Networking Service) in 2008.
They reported that of those that replied, 24
Pre & OJT companies recognized the prime importance of
retaining potential new graduate employees, easing
← their anxiety, and building their interpersonal skills
by providing activities for them until their
graduation. Furthermore, Hanada & GaiaX revealed
T
Pre-entry training → Officially hired
that these companies adopted SNS in the pre-entry
training programs in order to build the interpersonal
Figure 2: Pre-Entry T. & Human Capital Development. skills of potential new graduate employees, retain
them, and maintain their motivation.
3.2 Hypothesis This chapter provides three case studies and
discusses the company’s recognition of the function
With today’s expanded menus and contents for new of e-learning in the pre-entry training programs.
graduate employees, e-learning has become an easy
tool for developing the human capital of students 4.1 Case Study 1
who are due to start working after graduation.
However, we believe that there is another reason Case 1 is a large Japanese insurance company. They
why Japanese and Korean companies are utilizing e- adopted e-learning in their six-month pre-entry
learning in the pre-entry training programs, besides training programs in October 2006. In their e-
human capital development and improving the learning program, potential new graduate employees
efficiency of T&D programs (Figure 3). We provide are given individual pages and provided with
a hypothesis for our analyses. compulsory and optional menus.
Compulsory menus are comprised of certificate
E-learning HRD acquisition courses such as security trading and
financial planning, PC skills development courses
such as Excel and PowerPoint, TOEIC courses,
management overview, and business etiquette.
Retention
Trainees who tend to lag behind are encouraged by
emails from Human Resource Development staff. As
New function of E-learning an optional menu, video courses are provided for
Figure 3: New Function of E-learning. about 300 subjects. The most popular subjects
among trainees are business communication, team
Hypothesis: Companies recognize that e-learning in building, time management, and marketing.
the pre-entry training programs is The company in Case 1 indicated that the
important for retaining potential new objectives of adopting e-learning were to retain
graduate employees as well as potential new graduate employees, average out their

359
ICE-B 2008 - International Conference on e-Business

knowledge and skills, develop human capital, and Table 1: Functions of E-learning (Case 3).
improve the efficiency of T&D programs after
E- Mentoring Events
officially hiring them.
learning
Maintenance of ✓ ✓
4.2 Case Study 2 motivation
Improvement of ✓ ✓ ✓
Case 2 is a large Japanese service company that retention
regards employee retention as an important issue. In Interpersonal ✓ ✓
July 2007, they adopted SNS for potential new skills
graduate employees. Elimination of
job mismatching
The company is building a specific community
Reduction of ✓ ✓
in the SNS, and provides CEO messages, overviews early job hopping
of management and group companies, and a variety Basic business ✓ ✓
of information. In addition, recruiters and young skills
employees join the community as advisers. Professional
This company adopted SNS in order to retain business skills
potential new graduate employees. They are asked to Efficiency of ✓ ✓
post their opinions about events held for them on the T&D programs
community board. The company utilizes their ✓ = Effective
opinions for the next new graduate recruiting and Note: Interview results of Case 3
pre-entry training programs.
4.4 Discussion
4.3 Case Study 3
E-learning is utilized as a tool to develop the human
Case 3 is a large Korean manufacturing company. capital of employees and improve the efficiency of
As in Japan, there is seasonally scheduled hiring of T&D programs for them.
new graduates in Korea also. Large- and medium- In fact, the companies of the three case studies
sized companies tend to make their offers to recognize that e-learning is important for developing
candidates three to six months prior to their the human capital of potential new graduate
graduation from university or college. The period employees and improving the efficiency of T&D
until official hiring is shorter than that in Japan, but programs for them. Besides these known functions,
the competition between Korean companies is just as the case studies show that Japanese and Korean
intense, and potential new graduate employee companies expect e-learning to help improve
retention is a critical issue. employee retention when hiring new graduates. This
According to an interview survey that we feature was demonstrated by the survey of Hanada
conducted in February 2008, this company provides & GaiaX.
e-learning courses, events, and mentoring in the pre- Japanese and Korean companies make their
entry training programs for potential new graduate offers more than a few months prior to students’
employees. Their e-learning menus comprise graduation amid intense competition to hire new
orientation and PC skills development in graduates, and provide pre-entry training programs
programming. The contents of orientation include until officially hiring them. As a result, companies
overviews of management philosophy, company tend to focus on e-learning’s function of retaining
history, business model, and products and services. potential new graduate employees, besides the
The company in Case 3 mentioned that they conventional functions of maintaining their
recognized e-learning as an effective means of motivation, developing their basic business skills,
retaining potential new graduate employees, and improving the efficiency of T&D programs for
developing their basic business skills, and improving them.
the efficiency of T&D programs for them (Table 1).

5 CONCLUSIONS
This study examined the hypothesis that e-learning
in the T&D programs for potential new graduate
employees has the function of retention as well as

360
E-LEARNING FOR NEW GRADUATE EMPLOYEES - Another Function of e-Learning for New Graduate Employees of
Japanese and Korean Companies

human capital development, based on the recent Ministry of Economy, Trade and Industry Edition, 2007.
survey and case studies. e-Learning White Paper 2007/2008, Tokyo Denki
Many Japanese companies and some Korean University Press.
companies provide T&D programs that include e- Usugami, J., 2006. Pre-entry Training Programs as
Human Resource Strategy (in Japanese), Chuo Keizai
learning for potential new graduate employees prior sha.
to official hiring. These are known as pre-entry
training programs. Today, e-learning has numerous
menus and contents and is an easy means of
providing pre-entry training programs to potential
new graduate employees who are located in various
places including overseas.
E-learning has been utilized for human capital
development. We investigated another reason why
companies adopt e-learning in their pre-entry
training programs, based on the recent survey and
three case studies. We found that companies tend to
consider that e-learning in the pre-entry training
programs is important for retaining potential new
graduate employees as well as accelerating human
capital development and improving the efficiency of
T&D programs for them.

REFERENCES
Becker, G.S., 1975. Human Capital: A Theoretical and
Empirical Analysis, with Special Reference to
Education, 2nd Edition, Columbia University.
Brown, K.G., A field Study of Employee e-Learning
Activity and Outcomes, 2005. Human Resource
Development Quarterly, vol.16, No.4, pp.465-480.
Brown, W. and Yoshioka, C.F., Mission Attachment and
Satisfaction as Factors in Employee Retention, 2003.
Nonprofit Management & Leadership, vol.14, no.1,
pp.5-18.
Buck, J. M. and Watson, J. L., Retaining Staff Employees:
The Relationship Between Human Resources
Management Strategies and Organizational
Commitment, 2002. Innovative Higher Education,
vol.26, No.3, pp.175-193.
Hanada, M. and GaiaX, 2008. Joint Research Report on
Naiteisha-follow (in Japanese), GaiaX.
Klein, H.J., Fan, J. and Preacher, K.J., 2006. The effects of
early socialization experiences on content mastery and
outcomes: A meditational approach, Journal of
Vocational Behavior, vol.69, pp.96-115.
Lim, H., Lee, S.-G. and Nam, K., 2007. Validating E-
learning factors affecting training effectiveness,
International Journal of Information Management,
Vol.27(1), pp.22-35.
Mincer, J., 1974. Schooling, Experience, and Earnings,
NBER and Columbia University Press.
Ministry of Economy, Trade and Industry Edition, 2005.
e-Learning White Paper 2005/2006, English version,
Tokyo Denki University Press.

361
 OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF
E-SUPPLY CHAINS AMONG MALAYSIAN’S SMES

Kay Hooi Keoy, Mohamed Zairi

Centre forEnteprenruship and Entepreneurial Management
Bradford University of Bradford School of Management, BD9 4JL U.K.
[email protected],[email protected]

Khalid Hafeez
2
The York Management School, Sally Baldwin Buildings, Block A, University of York
Heslington, York, YO10 5DD, U.K.
[email protected]

Keywords: e-Supply chain, systems engineering, e-Business adoption, network organisation, structural equation
modelling, SMEs.

Abstract: This paper reviews the available literature to identify theoretical and empirical gaps regarding e-Supply
chain adoption among small and medium sized enterprises (SMEs). We argue that a generic e-Supply chain
framework should employ appropriate operational and behavioural perspectives. We propose that e-Supply
chain operation can succeed only when integrated with an efficient supply chain network and a strategic
plan committed to e-Business. Survey data from 208 Malaysian SMEs are collected and Structural Equation
Modelling (SEM) is employed to apply the proposed framework. More specifically, a set of three empirical
models are examined to evaluate the validity and impact of supply chain strategy, e-Business adoption, and
the interaction of these constructs on business performance. Our review suggests that much of the existing
e-Supply chain adoption and implementation literature is not firmly grounded in theory. We have suggested
that the systems engineering tradition of focusing on the interactions of technology, organizational structure,
and personnel provides a useful framework for understanding the business performance of e-Business. The
technology-organisational and people (TOP) dimensions are based on sound systems engineering principles
that are widely recognised and accepted for improving manufacturing organisation. We suggest that these
principles are equally valid for the e-Business oriented and virtual organisations. These expectations are
supported by our empirical results. We find that measures based on the TOP conceptual scheme provide
reliable and valid scales that are equally applicable to both e-Business and non-e-Business firms.

1 INTRODUCTION questioning the value proposition of e-Business

investments (Zhu et al., 2003).
Recent years have witnessed the worldwide adoption Studies indicate that technology diffusion among
of e-Business for achieving cost savings, improving knowledge intensive SMEs have been difficult (see
customer service, and promoting innovation, and for example, McCole and Ramsey; 2005; Ramsey et
taking advantage of new business opportunities al., 2005). Chapman et al. (2000) argue that SMEs
(Wagner et al., 2003). Despite the burst of the dot- are lagging behind their larger counterparts in the
com bubble, companies are continuing to adopt e- use of the Internet. Other studies have found that
Business operations. Martinsons and Martinsons SMEs are only half as likely to be using e-mail; for
(2002) suggest that the fear of lagging behind in micro companies the figure is even smaller
adopting the Internet technology has rushed many (Chapman et al., 2000). Poon and Swatman (1997)
firms to blindly engage in e-Business initiatives. As suggest that cost is the largest barrier for SMEs
a consequence, many have done so without deriving restricting the adoption of new technology. A recent
much benefit. Despite huge investments in e- report by Spectrum (2001) also supports the view
Business initiatives, academics and practitioners are that SMEs need to catch up with their larger

362
 OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF E-SUPPLY CHAINS AMONG MALAYSIAN’S SMES

counterparts in adopting ICTs in the automotive There has been extensive research investigating
components sector. The main issues identified the impact of organisational factors on innovation
include the excessive cost of e-technology and skill and technology adoption (Fjermestad, 2003;
deficiencies in e-Business implementation. Grandon and Pearson, 2004). The factors
This paper provides a critical review of the influencing Internet technology adoption within
available e-Business literature to identify theoretical supply chain strategy can be classified in several
and empirical gaps. Based on our literature review, ways such as internal and external environments,
we identify operational and behavioural firm and individual conditions, and domestic and
perspectives that form the basis of a theoretical international involvement (Moini and Tesar, 2005;
framework for understanding e-Supply chain Lewis and Cockrill, 2002). The perceptions of
adoption and success. We show that these management toward IT adoption are examined in
perspectives relate to the well established “systems many studies (Taylor and Murphy, 2004). .
engineering” principles of technology, organisation,
and people. 2.2 Behavioral Perspective
Technological sophistication of an organisation is
2 E-SUPPLY CHAIN: considered an important factor for businesses’ e-
Business adoption and implementation. There has
LITERATURE REVIEW been extensive research outlining important
determinants of organisational factors on e-Business
Operations management academics have always adoption (Tornatzky and Fleischer, 1990). The
highlighted the strategic importance of operations, majority of organisational factors addressed involve
and its role in corporate success. The consideration such organisational characteristics as size, industry
of operation strategy is relatively as important in e- type and business scope (Zhu et al., 2004, 2006).
Business operations as in operating in traditional However, there is a lack of study addressing the
environments. However, evidence from the literature relationship between information orientation /
suggests that many companies have adopted e- asymmetry and technological innovation /
Business without thinking through their strategic, integration on e-Business adoption (Hsieh et al.,
operational and behavioural impacts (Marshall and 2006).
Mackay, 2002; Gunasekaran et al., 2002; Dutta and From the behavioural perspective, Damodaran
Biren, 2001), which subsequently led to e-Business and Olpher (2000) have identified knowledge
failure. This section considers the impact the transfer, knowledge integration, and practical
Internet has on operational, and behavioural application of knowledge as the main elements for
management perspectives and whether new strategic developing “external” capabilities. According to a
thinking is required in response to the powerful study conducted by Caloghirou et al. (2004), the
external forces that are re-shaping industry. This readiness, and openness towards knowledge sharing
section also aims to sustain the significance of these among business partnerships are important factors in
perspectives by providing supporting evidence from improving business performance and encouraging
the existing e-Business literature. the adoption of e-Business. Establishing knowledge
management mechanisms and advantage knowledge
2.1 Operational Perspective assets is essential for successful technological and
organisational innovation (Bong et al., 2004).
e-Business is important for the supply chain
literature because of the increasing need to integrate 2.3 Performance Measurement
activities and information flows and to optimise the
processes not only at the single company level, but Marshall et al. (1999) define performance
also at the level of inter-company processes measurement as “… the development of indicators
(Landford, 2004; Lattimore, 2001; Stevens, 1989). and collection of data to describe, report on and
The importance and role of web-based technologies analyse performance”. Neely et al. (1995) see
to support company operations (e-Supply chain) is performance measurement as “the process of
widely acknowledged by both practitioners and quantifying the efficiency and effectiveness of
academics (Sanders and Premus, 2005; Porter, 2001; action”. Sanders and Premus (2005) argue that
Skjoett-Larsen, 2000). performance measurement is a complex issue that
incorporates economics, management, and

363
ICE-B 2008 - International Conference on e-Business

accounting disciplines. Zhu et al. (2004) have integration (from a baseline to external; companies
stressed that an appropriate measurement system is needed to focus on people dimensions internally as
essential to support a wide range of performance well as externally. This study argued the
measures. Using Kaplan and Norton’s (2004) applicability of Stevens’ (1989) integration
balance score card concepts, we have identified framework in today’s business environment where
tangible and intangible performance measures to companies want to move from a traditional business
evaluate performance improvements (Hafeez et al. to e-Business. Therefore, the identified dimensions,
2007). Based on the relevant literature (Eikebrokk namely technology, organisation, and people (TOP)
and Olsen, 2005) we identify three domains of are well suited for studying the success of e-Supply
measures to examine the perceived benefits of e- chain adoption.
Business adoption: Financial, Operational efficiency
and Coordination. Within each domain, it is useful Table 1: Incorporation of technology, organisation and
to categorise specific indicators under “operational”, people dimensions within each identified factor.
and “behavioural” perspectives.
Variables examined
Investments for supply chain system
2.4 System Engineering Concept
Integration of operating and planning
database
Systems engineering may be defined as the science

“Operational Perspective”
Standardised and customised

Supply Chain Strategy

of analysing the behaviour of a system (or
information
organisation) by studying the technology, policies
Information sharing and distribution
and management procedures (or organizational
structure) and the behaviour and attitudes of the Organisational structure
people who make up of the organisation (Forrester, Standardised supply chain practices
1961; Parnaby, 1981; Towill, 1993). Many past and and operations
current management initiatives such as Total Quality Integration of individual operations
Management (TQM) (Hafeez et al. 2006, supply channel
chain management (Hafeez, et al. 1996), business Time based logistics solutions
process re-engineering (BPR) (Hammer and Roles and responsibilities
Champy, 1994) are based on systems engineering Developing and maintaining
principles. Systems engineering distinguishes relationships
technology (T) and/or organisation (O) and/or Risk and rewards
people (P) dimensions (or TOP dimensions in short).
Systems engineering emphasizes the inter- Technological innovation and
“Behavioural Perspective”

connectedness of these dimensions, and suggests integration

E-Business Adoption

that change in one is very likely to have implications Information orientation and
requiring changes in others. asymmetry
e-Business operation might best be understood Adoptability of technology
from the perspective of supply chain management. infrastructures
We would particularly draw attention towards
Stevens’ (1989) supply chain management Organisational learning factors
integration framework based on systems engineering Organisational support and value
principles. Stevens’ (1989) model also provides a Organisational knowledge
consistent empirical support, which provides a good management
base for comparisons. Stevens (1989) has Internal and external collaboration
differentiated contributory factors for supply chain Performance measurement
integration into the ‘hard’ issues (such as Readiness mindset of adoption
technology) and the ‘soft’ (e.g. relations, attitudes,
etc). Numerous studies suggest that many companies Following the critique from the literature and
have not yet fully realised the technological gaps identified, it can be seen that the context of
integration of the available office technologies and operational and behavioural management are still fit
software tools such as Material Resource Planning to investigate the success factor of e-Business
(MRP), Distribution Resource Planning (DRP), and adoption. Through a careful content analysis,
Enterprise Resource Planning (ERP). Stevens, as elements have been identified which in the present
early as 1989, advocated that in order to achieve full author’s view contribute to e-Business research.
They can be generally categorised under the well-

364
 OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF E-SUPPLY CHAINS AMONG MALAYSIAN’S SMES

established operations research dimensions of understanding variation in the business success of

technology, organisation and people (see Table 1). Malaysian SMEs.

3.1 Sample Selection

3 THEORETICAL FRAMEWORK
We also describe the results obtained of an empirical
We have argued that a successful e-Supply chain study applying these concepts to data collected from
company needs to take into account “operational” 208 Malaysian SMEs. Using confirmatory factor
and “behavioural” issues. The overarching analysis techniques (i.e. structural equation
theoretical framework is summarized in Figure 1, modelling or SEM), we show the effects of
and includes supply chain strategy and e-Business operational, and behavioural adjustments to e-
adoption constructs. Business on business success. Malaysia has been
developing its information highway capacity since
late 1990s. This is realised by the investment of RM
40 Billion (approximately £ 5.9 billion) to establish
Multimedia Development Corporation (MDC). The
Multi-media Super Corridor (MSC) is one of the key
initiatives of MDC (Low et al., 2000).
We have selected six industrial sectors that had
previously been identified as the leading sectors in
e-Business adoption (UNCTAD, 2001; Daniel et al.,
2002; Daniel, 2003). These include
“Manufacturing”, “Services”, “Information
Technology”, “Finance, Insurance, and Real Estate”,
“Wholesale and Retail Trade”, and “Others”
(agriculture, communication, utility services). Equal
sample sizes (fifty) of firms were selected for each
sector. There are unequal numbers of SMEs in these
six sectors in Malaysia. Stratified sampling with
probabilities not proportional to stratum size
Figure 1: A conceptual framework for e-Supply chain
adoption.
(Dawson, 1998) was used to enable comparisons
between sectors. While such an approach could
restrict the generalisation of the results, it allows for
Figure 1 illustrates that within each of these
a focus on the issues in industries where e-Business
constructs are embedded the three systems
is rapidly becoming institutionalised. Three hundred
engineering principles: “technology”,
questionnaires were emailed across these six
“organisation,” and “people” (or TOP). The
industries. Overall, 208 respondents returned the
framework illustrates that these constructs are inter-
questionnaire for a response rate of 69.3%. Sample
related, and therefore any change in one factor will
sizes and response rates are reported in Table 2
have ramifications for others. We hypothesize that
developments in each dimension (TOP) of each
Table 2: Survey sample characteristics (n= 208).
domain (supply chain strategy and e-Business
adoption) are necessary for satisfactory business
Sample industries Respondents
performance (BP).
Manufacturing 30
Hypothesis H1: Supply chain strategy (SCS) is a
significant determinant of business performance (BP) Services 28
Hypothesis H2 E-Business adoption (EBA) is a significant IT 43
determinant of business performance (BP) Finance, Insurance and Real Estate 35
Hypothesis H3 Business performance (BP) is directly Wholesale and Retails Trade 32
related to the level of mutual dependency (and alignment)
between supply chain strategies (SCS) and e-Business Others 40
adoption (EBA) Total Respondent 208
In the sections that follow, we will illustrate the Response Rate (%) 69.3%
utility of our approach by applying it to

365
ICE-B 2008 - International Conference on e-Business

3.2 Structural Equation Models Table 3: Regression weights for hypotheses H1 to H6 for
the Malaysian sample (n = 208).
We have employed Structural Equation Modelling
(SEM) to test the applicability of our conceptual
framework. SEM is a multivariate statistical
technique that allows for the simultaneous analysis
of the first-order and second-order measurement
factors. In our analysis, the first-order factors
consist of multi-item measures technological,
organizational, and personnel/attitudinal dimensions
of each of the basic constructs of supply chain
strategy, and e-Business adoption. Supply chain
strategy, and e-Business adoption constructs are Table 4: Second factor loadings for sub-hypotheses for the
second-order factors composed of the first-order Malaysian sample (n = 208).
ones. The dependent measure of business
performance is also conceptualized as a “factor of
factors” including financial, efficiency, and
coordination factors, each of which is composed of
multiple items.
The final model provides excellent fit to the data:
χ2 of 588.80, df = 393 with 72 parameters; χ2/df =
1.50; CFI = 0.96; GFI = 0.85; RMSEA = 0.04; TLI
= 0.95). This model fit indices fall in an acceptable
range (> 0.90) and the RMSEA was less than 0.05.
This structural model was nested within the first
order model; in that it had been generated by
imposing restrictions on, the parameters of the first
order model (Figure 2).
Table 3 and Table 4 indicate the hypotheses
results for the Malaysian sample. The path
coefficients of interest in this model were generated
between the independent factors ( ξ , exogenous) of
e-Business constructs and the dependent factor of
business performance (η , endogenous).
Interestingly, the results suggested that e-Business
adoption (H2; γ = 0.53; c.r. = 4.97) was the
strongest stronger predictor of business performance
followed by the supply chain strategy construct (H1;
γ = 0.26; c.r. = 2.70). The correlational paths are
also of key interest when running this model. Results
suggested between correlation between supply chain
strategy and e-Business adoption had strong phi
value of φ = 0.70 at significant value of t >1.96.
The strong correlation was between supply chain Figure 2: Standardised estimates for main and sub-
strategy and e-Business adoption which confirmed hypotheses for Malaysian (n = 208) sample.
that companies in Malaysia regardless of which
sectors they belongs to still treated both of these
factors as a important driver for improvement of 4 DISCUSSION
business performance by treating equally important
and they complement each other when a strategy had
The theoretical model confirms that successful e-
been formulated.
Supply chain requires supply chain strategy, and e-

366
 OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF E-SUPPLY CHAINS AMONG MALAYSIAN’S SMES

Business adoption, which have mutual dependency chain strategy. In e-Businesses, the use of
regardless of geographic and economic differences. technology is a stronger determinant of business
For the Malaysian sample (in the context of a performance than supply chain strategy.
developing country), the formation of e-Business Furthermore, in e-Supply chain, technology use is
adoption is dependent on the implementation of strongly articulated with business and supply chain
supply chain strategy. This is a critical factor for the strategies.
Malaysian e-Business development as most of the Adopting enterprises are not without business
businesses operate in a larger geographical area. performance problems. Our results suggest that for
One explanation of greater relevance of supply e-Business organisations to be successful, supply
chain strategy in the Malaysian sample could be that chain management need to be given a higher level of
some of the Malaysian sample surveyed function as strategic importance (Koh et al., 2007). We would
a role of contractors to core nations, and may be argue that successful business collaboration is the
more focused on primary products. Their success result of human interactions, which can be supported
depends on being able to assemble resources and to by IT, but not to be replaced by IT. This is
deliver products on time. The success for the particularly important in the e-Business context
companies operating in core nations may depend where the traditional business model is usually
more critically on finding new markets for the developed on the backbone of technological
products. Such an explanation may be viewed as infrastructure, and "people" related issues can be
speculation, but the key results are broadly easily buried under the overwhelming emphasis on
consistent with this sort of a “world systems” view. technological details. Technology is not the most
Results also suggest that the operational differences critical factor in improving supply chains. To
in managing a global trade and distribution chain are improve in this area, SMEs must consider relevant
more prominent than any cultural differences in attitudinal issues as identified by Steven (1989) to
explaining the (limited) differences in the surveyed allow for e-technology to be accepted and diffused
samples. in the e-Supply chain.
The results suggested that companies must pay
attention to their technological, organisational, and
human capabilities for improving e-Business 5 CONCLUSIONS
performance. These capabilities are critical when
firms are planning or at the very initial stage of e- Our literature review suggests that the existing e-
Business adoption, where most processes are at low Supply chain and implementation studies lack
integration levels and are full of manual work (Hsin theoretical underpinning. This situation is more
and Shaw, 2005). Companies that intend to venture acute for SMEs as the limited numbers of e-Supply
into e-Business need to acknowledge and identify chain models found in the literature are not tested
barriers caused by “organisation” dimensions by empirically. Systems engineering principles, which
offering training and knowledge for system focus on the interaction of technology, organisation,
integration, standards development, and process and people (TOP), provide a useful conceptual
automation as well as to overcome possible IT scheme for understanding the business performance
resistance. of both e-Supply chain firms and others.
Where some previous studies have identified We have introduced a structural equation
supply chain strategy as key dimensions modelling approach, and used it to examine the
(Wickramatillake et al., 2007; Koh et al., 2006), our sources of good business performance for companies
model extends this by measuring the impact of adopting e-Supply chain. The multi-item constructs
technological, organisational and people related of e-Business adoption and supply chain strategy
issues with e-Business adoption in order to become a relate differently to business performance. The
successful e-Business firm. Both e-Businesses and measures developed here, and the empirical results
conventional businesses use information technology. can be used as a benchmarking tool for the SMEs
Our results suggest, however, that technology plays who wish to embark on e-Supply chain adoption
a much more critical role in the business journey. The study also provides some useful
performance of enterprises that have fully adopted directions for new economy cyber-entrepreneurs,
the e-Supply chain model. In non-adopting guiding them to give due consideration towards
businesses, the use of technology is positively appropriate operational and behavioural factors
related to business performance, but only modestly when considering e-Supply chain adoption.
so; and, technology use is not integrated with supply

367
ICE-B 2008 - International Conference on e-Business

It would be an added value to expand findings Forrester, J.W. (1961) Industrial Dynamics, T. Press,
obtained from the quantitative study by conducting Cambridge, MA and John Wiley & Sons, New York.
qualitative investigations in a case study format. As Grandon, E. and Pearson, J.M. (2003). Strategic Value and
stated by Patton (1987), “case studies are useful Adoption of Electronic Commerce: An Empirical
Study of Chilean Small and Medium Businesses.
where one needs to understand some particular Journal of Global Information Technology
problems in great depth and identify rich Management. Vol. 6, No. 3, pp. 22-43.
information that can be learned from few exemplars Gunasekaran, A., Marri, H.B., McGaughey, R.E. and
of the phenomenon in question”. The future research Nebhwani, M.D. (2002). E-commerce and its impact
could be conducted as a complementary study, to on operations management. International Journal of
further assess and test the applicably of the e-Supply Production Economics. Vol. 75, No. 1, pp. 185-197.
chain factors of e-Business adoption and to identify Hafeez, K. Abdelmeguid, H. and Malak, N.M. (2006). A
and investigate any potential benefits, obstacles or Framework for TQM to Achieve Business Excellence.
Total Quality Management, 17(9), pp. 1213-1229.
emerging themes associate with it. Several
Hafeez, K., Griffiths, M., Griffiths, M. and Naim, M.M.
organisations (minimum three organisations from (1996). Systems Design of a Two-echelon Steel
each industry) that expressed their interests and met Industry Supply Chain. Int. J. Production Economics,
the criteria from both samples could be contacted for Vol. 45, pp. 121-130.
face-to-face interview. It is hoped that this Hafeez, K., Zhang, Y.B. and Malak, N.M. (2007).
combination of quantitative and qualitative study Outsourcing Non-core Assets and competences of a
will further support and verify the applicability and Firm using Analytic Hierarch Process. Computer and
robustness of the proposed conceptual model. Operations Research (C&OR).(in Press
August/September issue); reference 10-
1016/j.cor.2006.
Hammer, M. and Champy, J. (1993). Business Process
REFERENCES Reengineering. London: Nicholas Brealey.
Hsieh, C.T., Lai, F. and Shi, W. (2006). Information
Bong, S.H., Lee, J. and Gil, Y. (2004). Effective team orientation and its impacts on information asymmetry
processes for technology internalization with special and e-Business adoption: Evidence from China's
emphasis on knowledge management: successful late international trading industry. Industrial Management
starter, Samsung case. International Journal of & Data Systems. Vol. 106, No. 6, pp. 825 – 840.
Technology Management. Vol. 27, No. 1, pp. 16-39. Hsin, L.C. and Shaw, M.J. (2005). A Roadmap to
Cagliano, R., Caniato, F. and Spina, G. (2005). Adopting Emerging Technology in E-Business - An
Reconsidering E-Business Strategy and the Impact on Empirical Study. Working Papers. Bureau of
Supply Chains. International Journal of Operations & Economic and Business Research, University of
Production Management. Vol. 25, No. 12, pp. 1328- Illinois at Urbana.
1332. Kaplan, R.S. and Norton, D.P. (2004). Measuring the
Damodaran, L. and Olpher, W. (2000). Barriers and Strategic Readiness of Intangible Assets. Harvard
facilitators to the use of knowledge management Business Review. Vol. 82, No. 2, p 52 - 67.
systems. Behaviour and Information Technology. Vol. Koh, S.C.L. and Tan, K.H., 2006, Translating knowledge
19, No. 6, pp. 405-413. of supply chain uncertainty into business strategy and
Dawson, T.E. (1998). Structural equation modelling actions, Journal of Manufacturing Technology
versus ordinary least squares canonical analysis: some Management, 17, 4, pp. 472-485
heuristic comparisons. San Diego: American Koh, S.C.L., Gunasekaran, A. and Rajkumar, D., 2007,
Education Research Association. ERPII: The involvement, benefits and impediments of
Dutta, S. and Biren, B. (2001). Business Transformation collaborative information sharing, International
on the Internet: Results from the 2000 Study. Journal of Production Economics.
European Management Journal. Vol. 19, No. 5, pp. Lanford, A. (2004).The Current Practice of Organization
449-462. Development within the Firm: A Survey of Large
Eikebrokk, T.R. and Olsen, D.H. (2005). Co-opetition and Industrial corporations. Group and Organization
e-Business success in SMEs: An empirical Studies. No. 17, pp.117–134.
investigation of European SMEs. Proceedings of the Lattimore, O. (2001). Electronic Commerce Tools that
38th Annual Hawaii International Conference on Address Supply Chain Performance. Logistics
System Sciences (HICSS'05).Hawaii, pp. 234-267. Spectrum. Vol. 35, No. 4, pp. 11-12.
Fjermestad, J. (2003). Electronic customer relationship Lewis, R. and Cockrill, A. (2002). Going global –
management: revisiting the general principles of remaining local: the impact of e-commerce on small
usability and resistance – an integrative retail firms in Wales. International Journal of
implementation framework. Business Process Information Management. Vol. 22, No. 3, pp. 195-
Management Journal. Vol. 9, No. 5, pp. 572-591. 209.

368
 OPERATIONAL AND BEHAVIOURAL DIMENSIONS OF E-SUPPLY CHAINS AMONG MALAYSIAN’S SMES

Low L.T., Mazlina, M. and Yang, W.K. (2000). Electronic UNCTAD (2001). E-Finance and Small and Medium-size
Commerce in Malaysia. Akauntan Nasional. Vol. 34, Enterprises (SMEs) in Developing and Transition
No. 5, pp. 48-51. Economies. UNCTAD Expert Meeting. Improving
Marshall, M., Wray, L., Epstein, P. and Grifel, S. (1999). Competitiveness of SMEs in Developing Countries:
21st century community focus: better results by Role of Finance Including E-Finance to Enhance
linking citizens, government and performance Enterprise Development.
measurement. Public Management. Vol. 81 No. 10, Wagner, B.A., Fillis, I. and Johansson, U. (2003). E-
pp. 12-19. Business and E-Supply Strategy in Small and Medium
Marshall, P. and Mackay, J. (2002). An emergent Sized Businesses (SMEs). Supply Chain
framework to support visioning and strategy Management: An International Journal. Vol. 8, No. 4,
formulation for EC. INFOR Journal. Vol. 40, No. 1, pp. 343-354.
pp. 3–22. Wickramatillake, C.D., Koh, S.C.L., Gunasekaran, A. and
Martinsons, M.G. and Martinsons, V. (2002). Rethinking Arunachalam, S., 2007, Measuring performance
the Value of IT, Again. Communications of the ACM. within supply chain, Supply Chain Management: An
Vol. 45, No. 7, pp. 25–26. International Journal, 12, 1, pp. 52-59.
McCole, P. and Ramsey, E. (2005). A Profile of Adopters Zhu, K., Ken, K., Sean, X. and Jason, D. (2004).
and Non-adopters of eCommerce in SME Professional Information Technology Payoff in E-Business
Service Firms. Australasian Marketing Journal. Vol. Environments: An International Perspective on Value
13, No.1, pp. 36-48. Creation of E-Business in the Financial Services
Neely, A., Gregory, M. and Platts, K. (1995). Performance Industry. Journal of Management Information
measurement system design – a literature review and Systems. Vol. 21, No. 1, pp. 17-54.
research agenda. International Journal of Operations & Zhu, K., Kraemer, K.L. and Xu, S. (2003). E-Business
Production Management. Vol. 15 No. 4, pp. 80-116. adoption by European firms: A cross country
Parnaby, J. (1981). Concept of a Manufacturing System. assessment of the facilitators and inhibitors. European
In: Systems Behaviour, 3rd Edn. (Paul Chapman, Journal of Information Systems. Vol. 12, No. 4, pp.
London). pp. 131-141. 251–268.
Patton, M. Q. (1987) How to Use Qualitative Methods in Zhu, K., Kraemer, K.L. and Xu, S. and Vijay G. (2006).
Evaluation. California: Sage Publications, Inc. Migration to Open-Standard Interorganizational
Poon, S. and Swatman, P.M.C. (1997). Determinants of Systems: Network Effects, Switching Costs, and Path
Small Business Internet Usage: A Multi-method Dependency. MIS Quarterly. Vol. 30 (Special Issue),
Investigation of Perceived benefits, Tenth pp. 515-539.
International Bled Electronic Commerce Conference,
Bled, Slovenia, pp. 465-483.
Porter, M.E. (2001). Strategy and the Internet. Harvard
Business Review. Vol. 79, No. 3, pp. 63-78.
Ramsey, E., Bell, P.J. and McCole, P. (2005). Internet-
Based Business Among Knowledge Intensive
Business Services: Some Irish Regional Evidence.
Service Industries Journal (Special Issue - Services in
the Regions). Vol. 25, No. 4, pp. 525-545.
Sanders, N.R. and Premus, R. (2005). Modeling the
relationship between Firm IT capability, collaboration
and performance. Journal of Business Logistics. Vol.
26, No. 1, pp. 1-23.
Skjoett-Larsen, T. (2000). European logistics beyond
2000. International Journal of Physical Distribution &
Logistics Management. Vol. 30, No. 5, pp. 377-387.
Stevens, G.C. (1989). Integrating the Supply Chain.
International Journal of Physical Distribution and
Materials Management. Vol. 19, No. 8, pp. 3-8.
Taylor, M. and Murphy, A. (2004). SMEs and E-Business.
Small Business and Enterprise Development. Vol, 11,
No. 3, pp. 280-289.
Tornatzky, L. G. and Fleischer, M. (1990). The Processes
of Technological Innovation. Lexington: Lexington
Books.
Towill, D.R. (1993). System dynamics- background,
methodology and applications. 1. Background and
methodology, Computing & Control Engineering
Journal; 4(5). pp. 201-208, ISSN: 0956-3385.

369
 A COMPARISON OF WEB SITE ADOPTION IN SMALL AND
LARGE PORTUGUESE FIRMS

Tiago Oliveira and Maria F. O. Martins

Instituto Superior de Estatística e Gestão Informação, Universidade Nova de Lisboa
Campus de Campolide, 1070-312 Lisboa, Portugal
[email protected], [email protected]

Keywords: Web site, adoption, small firms, large firms, information technology.

Abstract: This study compares the impact of different Technology-Organization-Environment (TOE) factors on the
web site adoption decision in small and large firms. A survey that was undertaken by the National Institute
of Statistics on the use of Information Technologies (IT) by firms in Portugal was used as the empirical
basis for this study. We found significantly differences in the factors that determined web site adoption
decision in small and large firms. While large firms are mainly influenced by organizational and
environmental factors, small firms are also concerned about the technological context. Moreover, the results
of our study suggested that, for Portuguese firms, the only factor that is equally important as web site
facilitator is competitive pressure.

1 INTRODUCTION medium-sized enterprises (SMEs) because of the

unique characteristics of SMEs as for example the
New IT, such as Internet enables firms to do lack of business and IT strategy, limited access to
businesses in a different way (Porter, 2001). In order capital resources and poor information skills. While
to strength the potential of the Internet, firms are there exist an interesting and growing literature
establishing their presence on the Web: in 2005, the addressing the determinants of IT adoption in the
overall percentage of enterprises in the EU with a specific context of SMEs (Harindranath et al., 2008,
web site is 61%, but notably higher for larger firms Parker and Castleman, 2007,) and a limited research
(90%) than for small firms (56%). Significantly for microfirms (Clayton, 2000), only a reduced
differences also exist between Member States: while number of studies (Daniel and Grimshaw, 2002)
the leader countries, Sweden and Denmark, are attempt to compare directly the approaches of small
already reaching the saturation level for large firms and large firms to this new domain. Our work seeks
(97%), countries like Portugal (75%) and Latvia to fill this gap in the literature, by analysing the
(65%) are far away from this adoption level. For relative importance of the factors that enable or
small firms, this difference is greater: the web site inhibit web site adoption by small firms compared
adoption level is 80% for Sweden compared with the with large firms. The two main purposes of this
33% level for Portugal (Eurostat, 2006). Do study are the following:
Portuguese small firm managers realize the strategic ƒ To examine the importance of technology-
value of owning a web-site in the same manner as organisational-environmental (TOE) related
large firm managers? Or have they encountered factors as fundamental determinants of web
specific barriers to its implementation? Some studies site adoption;
have been done to understand the differences in IT ƒ To analyze if the relative importance of such
adoption among European Countries (Zhu et al., factors is different for small and large firms.
2003) and much research attempted to comprehend
the relationship between firms size and IT adoption To achieve these research objectives we used a rich
decision (Lee and Xia, 2006). Some authors data set of 637 large firms and 3155 small firms that
(Grandon and Pearson, 2004, Premkumar, 2003) are representative of Portuguese economy. The
suggested that the research findings on large understanding of the determinants of web site
businesses cannot be generalized to small and adoption, at firm level, may be a useful tool in

370
 A COMPARISON OF WEB SITE ADOPTION IN SMALL AND LARGE PORTUGUESE FIRMS

addressing the right type of policy measures to

stimulate the use of internet business solutions, with
the aim of enhancing the competitiveness and
productivity of Portuguese firms (Bertschek et al.,
2006, Black and Lynch, 2001, Bresnahan et al.,
2002, Brynjolfsson and Hitt, 2000, Dedrick et al.,
2003, Konings and Roodhooft, 2002, Martins and
Raposo, 2005, Zhu and Kraemer, 2002). This is
particularly needed in the case of Portugal which, for
several reasons, has been suffering from a serious
lack of competitiveness in comparison to other
industrialized economies. Our work has two
important contributions: the first is related to the
very limited research on comparing the determinants
of IT adoption in small and large firms. Secondly,
we present useful results for Portugal where there
are few published studies on the subject (Parker and
Castleman, 2007). The next section presents the
theoretical framework based on TOE approach.
Then, the proposed hypotheses are tested using an
econometric model. Finally, we present major
findings and conclusions.

2 THEORICAL FRAMEWORK
AND CONCEPTUAL MODEL
In this study we used the TOE framework,
developed by Tornatzky and Fleisher (1990) and
Figure 1: Conceptual framework for web site adoption.
applied in many empirical studies related to IT
innovations. The TOE model identifies three aspects
that influence the adoption and implementation of
2.1 Technology Context
technical innovations by firms: technological
Technology readiness can be defined as technology
characteristics including factors related to internal
infrastructure and IT human resources. Technology
and external technologies of firms; organizational
readiness “is reflected not only by physical assets,
factors relating to firm size and scope,
but also by human resources that are complementary
characteristics of the managerial structure of the
to physical assets” (Mata et al., 1995). Technology
firm, quality of human resources; and environmental
infrastructure establishes a platform on which
factors that incorporate industry competitiveness
internet technologies can be built; IT human
features. This theoretical background is the one used
resources provide the knowledge and skills to
by Iacovou et al. (1995), Kuan and Chau (2001) and
develop web applications (Zhu and Kraemer, 2005).
Premkumar and Ramamurthy (1995) to explain
Theoretical assertions on the impact of Technology
electronic data Interchange (EDI) adoption and by
readiness on IT adoption are supported by several
Thong (1999) to explain information system (IS)
empirical studies, based on data sets representative
adoption and Hong and Zhu (2006) to explain e-
of all sizes of firms (Hong and Zhu, 2006, Zhu et al.,
commerce adoption. Empirical findings from these
2003, Zhu et al., 2006). These results where also
studies confirmed that TOE methodology is a
confirmed within the specific context of SMEs (Al-
valuable framework to understand the IT adoption
Qirim, 2007, Dholakia and Kshetri, 2004, Kuan and
decision. In accordance with TOE theory, we
Chau, 2001, Mehrtens et al., 2001). Therefore, in
developed in the next subsection a conceptual
general we expected that firms with greater
framework for web site adoption (see Figure 1).
technology readiness are in a better position to adopt
web sites. However, as suggested by others authors
(Daniel and Grimshaw, 2002, Parker and Castleman,

371
ICE-B 2008 - International Conference on e-Business

2007, Premkumar, 2003), this factor will probably (Beatty et al., 2001), for SMEs (Iacovou et al., 1995,
affect in a different way small and large firms. Kuan and Chau, 2001) and for all size firms (Gibbs
H1: The level of technology readiness is positively and Kraemer, 2004). However, as suggested by
associated with web site adoption but the impact will Daniel and Grimshaw (2002) small firms and large
vary between large and small firms firms perceived these benefits in a different. We
examine perceived benefits of electronic
Before the internet, firms had been using correspondence and we postulate that:
technologies to support business activities along
their value chain, but many were ‘‘islands of H4: Perceived benefits of electronic correspondence
automation’’— they lacked integration across is positively related with web adoption, but the
applications (Hong and Zhu, 2006). The impact will vary between small and large firms.
characteristics of the internet may help eradicate the The presence of skilled labour in a firm increases its
incompatibilities and rigidities of legacy information ability to absorb and make use of an IT innovation,
systems (IS) and accomplish technology integration and therefore is an important determinant of IT
among various applications and databases. Evidence diffusion (Caselli and Coleman, 2001, Hollenstein,
from the literature suggests that integrated 2004, Kiiski and Pohjola, 2002). Since the
technologies may enhance firm performance by successful implementation of new IT usually
reducing cycle time, improving customer service, requires complex skills, we expect firms with more
and lowering procurement costs (Barua et al., 2004). IT training programs to be more likely to adopt web
We define technology integration as the systems for site. However, there will probably be differences
managing orders that are automatically linked with between firms due to the limited IT budgets of small
other IT systems of the firm. This type of factor firms. We postulate the following:
where also identified by Al-Qirim (2007) for the H5: IT training programs are positively associated
specific case of SMEs. Therefore, we expect firms with web site adoption, but the impact will vary
with a higher level of technology integration to be between small and large firms.
those who adopt web sites sooner. However,
probably there will be significantly differences The fact that workers can have access to the IT
between small and large firms (Daniel and system from outside of the firm reveals that the
Grimshaw, 2002). These reflections lead to the organisation is prepared to integrate its technologies.
following hypothesis: However, this factor is expected to influence in a
different way small firms, where the number of
H2: The level of technology integration is positively employees is small and their presence at the place of
associated with web site adoption, but the impact work is more important than for large firms. We
will vary between small and large firms. postulate that:
The lack of security may slow down technological H6: The level of access to the IT system from outside
progress. For example, for Portugal in 2002 this was of the firm is positively associated with web site
the greatest barrier to internet use (Martins and adoption, but the impact will vary between small and
Oliveira, 2005) and in China it is one of the most large firms.
important barriers to the adoption of e-commerce
(Tan and Ouyang, 2004). We expect firms with a Regulatory environment has been acknowledged as
higher level of internal security applications to be a critical factor influencing innovation diffusion
more probable web site adopters. Within this (Zhu et al., 2003, Zhu et al., 2004, Zhu et al., 2006).
context, there is no empirical evidence suggesting a Firms often refer inadequate legal protection for
same behaviour between small and large firms. online business activities, unclear business laws, and
Therefore we stipulate the following: security and privacy as concerns in using web
technologies (Kraemer et al., 2006). We postulate
H3: Internal security applications are positively that for small firms, this concern will probably be
associated with web site adoption, but the impact different from their large counterparts.
will probably vary between small and large firms.
H7: The presence of internet and e-mail norms is
2.2 Organization Context positively associated with web site adoption, but the
impact will vary between small and large firms.
Empirical studies consistently found that perceived
benefits have a significant impact in IT adoption.
This result is validated for medium to large firms

372
 A COMPARISON OF WEB SITE ADOPTION IN SMALL AND LARGE PORTUGUESE FIRMS

2.3 Environmental Context cumulative distribution. To analyse and compare the

influence of each factor on the probability of being a
Empirical evidence suggests that competitive web site adopter, we need to compute the marginal
pressure is a powerful driver of IT adoption and effect of xj. This effect is obtained, for the
diffusion (Gibbs and Kraemer, 2004, Hollenstein, continuous variables, using the formula given by:
2004, Zhu et al., 2004) and this fact is also verified
∂P ( y = 1 / x )
in small business research (Al-Qirim, 2007, = φ ( xβ ) β j (2)
∂x j
Dholakia and Kshetri, 2004, Grandon and Pearson,
2004, Iacovou et al., 1995, Kuan and Chau, 2001). For the binary explanatory variables it is given by:
Therefore, we expect the probability of adopting a
ΔP ( y = 1/ x )
web site to be positively influenced by the = Φ ( xβ | x, x j = 1) − Φ ( xβ | x, x j = 0 ) (3)
Δx j
proportion of web site adopters in the industry or
sector to which the specific firm is affiliated. where φ(.) is the density standard normal
However, some studies suggested that competitive distribution.
pressure will be more significant in causing small The vector of explanatory variables (x) includes:
firms to adopt an IT than for larger firms, since they
A technology readiness (TR) index that was built by
need to protect their competitive position (Daniel
aggregating 8 items on technologies used by the firm
and Grimshaw, 2002). Therefore, we assume that:
(on a yes/no scale): computers, e-mail, intranet,
H8: The level of web site competitive pressure is extranet, own networks that are not the internet (own
positively associated with web site, but the impact exclusive networks), wired local area network
will vary between small and large firms. (Lange et al.), wireless LAN, wide area network
(WAN), and one item standing for existence of IT
2.4 Controls specific skills in the firm (on a yes/no scale) (Zhu et
al., 2004). The first 8 items represent the penetration
We control, as usual, for industry or economic sector of traditional information technologies, which
effects. We used a dummy variable to control for formed the technological infrastructure (Kwon and
data variation that would not be captured by the Zmud, 1987). The last item represents IT human
explanatory variables mentioned before. resources (Mata et al., 1995). To aggregate the items
we used multiple correspondence analyses (MCA).
The MCA is a method of “multidimensional
3 DATA AND METHODOLOGY exploratory statistic” that is used to reduce the
dimension when the variables are binary. For more
3.1 Data details see (Johnson and Wichern, 1998). The first
dimension explains 50% of inertia. In the negative
The data used in this study were provided by side of the first axis we have variables that represent
National Institute of Statistics (INE) and result from firms that do not use IT infrastructures and do not
the survey On the use of Communication and have workers with IT skills. On the positive side we
Information Technologies in Firms (Iutice) in 2006. have the variables that represent the use of
In our study we defined that small firms have less infrastructures and workers with IT skills.
than 50 employees and large firms have more than Cronbach’s α, the most widely used measure for
250 employees. Our sample consists on 3155 small assessing reliability (Chau, 1999), is equal to 0.8761,
and 637 large firms and is representative of the indicating adequate reliability. Reliability measures
Portuguese private sector excluding the financial one the degree to which items are free from random
error, and therefore yield consistent results.
3.2 Methodology Technology integration (TI) was measured by the
number of IT systems for managing orders that are
We estimated the following Probit Model: automatically linked with other IT systems of the
firm (see appendix). The variable ranges from 0 to 5.
P(y=1/x)=Ф(xβ) (1) This variable reflects how well the IT systems are
Where y=1 if firm decided to adopt a web site, and connected on a common platform.
zero otherwise, x is the vector of explanatory Internal security applications (ISA) was measured
variables, β the vector of unknown parameters to be by the numbers of the use of internal security
estimated, and Φ(.) is the standard normal

373
ICE-B 2008 - International Conference on e-Business

applications in the firms (see appendix). The suggested that all the coefficients have the expected
variable range from 0 to 6. signs and the only independent variable that is not
Perceived benefits of electronic correspondence statistically significant is the access to the IT system
(PBEC) was measured by the shift from traditional of the firm (AITSF). We can identify seven relevant
postal mail to electronic correspondence as the main drivers of web site adoption for small firms:
standard for business communication, in the last 5 technology readiness (TR), technology integration
years (on a yes/no scale). (TI) and internal security application (ISA)
reflecting the technological context; perceived
IT training programs (ITTP) is also a binary variable
benefits of electronic correspondence (PBEC), IT
(yes/no) related to the existence of professional
training programs (ITTP) and internet and e-mail
training in computer/informatics, available to
norms (IEN), representing the organization context;
workers in the firm.
web site competitive pressure (WEBP), concerning
Access to the IT system of the firm (AITSF) was the environmental context. For large firms, we
measured by the number of places from which identify four significant factors influencing web site
workers access the firms information system (see adoption decision: technology readiness (TR), IT
appendix). The variable ranges from 0 to 4. training programs (ITTP), access to the IT system of
Internet and e-mail norms (IEN) was measured by firms (AITSF) and web site competitive pressure
whether firms have defined norms about internet and (WEBP). In both cases, as expected, the economic
e-mail (on a yes/no scale). sector is a relevant factor (SER).
Web site competitive pressure (WEBP) is computed
as the percentage of firms in each of the 9 industries Table 1: Estimated coefficients for web site adoption
model.
that had already adopted a web site two years before
the time of the survey, i.e. in 2004. As in Zhu et al. Small firms Large firms
(2003) the rationality underlying our model is that Technological context
- TR 1.044*** 0.346*
an observation of the firm on the adoption behaviour - TI 0.069*** -0.028
of its competitors influences its own adoption - ISA 0.170*** 0.038
decision. Organizational context
- PBEC 0.293*** -0.039
Services (SER) is a binary variable (yes/no) equal - ITTP 0.235*** 0.644***
one if firm belong to the service sector. - AITSF 0.044 0.278***
- IEN 0.379*** 0.165
Environmental context
- WEBP 0.011*** 0.017***
4 ESTIMATION RESULTS Controls
- SER 0.185*** 0.306**
Constant -1.742*** -1.041***
The web site adoption model is estimated using Sample size 3155 637
maximum likelihood. The estimation results for LL -1038.5 -223.3
small and large firms are presented in Table 1. R2 0.419 0.157
Goodness-of-fit is assessed in three ways. First, we AUC 0.909 0.779
used log likelihood test, which reveals that our Note: * p-value<0.10; ** p-value<0.05; *** p-value<0.01.
models are globally statistic significant. Secondly The estimated marginal effects for the determinants
the discrimination power of the model is evaluated of web site adoption model, for small and large
using the area under the receiver operating firms, are reported in Table 2.
characteristic (ROC) curve, which is equal to 90.9% Their comparison reveals that, as expected, most of
and 78% for small and large firms, respectively. the marginal effects vary between small and large
Finally, the R2 shows that the percentage explained firms. The exception is the web site competitive
by the model is 41.9% for small firms and 15.7% for pressure impact that is the same for small and large
large firms. The three statistical procedures reveal a firms. Therefore hypotheses H1-H7 are validated
substantive model fit, a satisfactory discriminating and H8 is not confirmed.
power and there is evidence to accept an overall There are three additional aspects to be noted here.
significance of the model. Firstly, the technological context is much more
Hypotheses H1-H9 were tested analysing the sign, relevant for small firms than for large firms.
the magnitude, the statistical significance of the Secondly, within organizational context, perceived
coefficients and the marginal effects. As can be seen benefits and internet e-mail norms are more
from Table 1, for small firms, the estimation results important to determine web site adoption for small

374
 A COMPARISON OF WEB SITE ADOPTION IN SMALL AND LARGE PORTUGUESE FIRMS

firms than for their larger counterparts. Finally, the motivations towards the IT adoption are different for
access to the IT system of the firm is relevant only small and large firms. Therefore, government
for large firms. As a whole, our results are in initiatives, such as the Technological Plan, for
accordance with those reported in studies comparing Portugal, must be different for small and large firms,
IT adoption in large and small firms (Daniel and namely those related to procurement incentives.
Grimshaw, 2002). However, the limited number of Secondly, managers should be aware that technology
research in this specific domain difficult the readiness constitutes both physical infrastructure and
generalization of the results. intangible knowledge such as IT skills. This urges
top leaders (mainly in small firms) to foster
Table 2: Estimated marginal effects for web site adoption managerial skills and human resources that possess
model. knowledge of these new information technologies.
Small firms Large firms Therefore, there is a business opportunity for IT
Technological context firms to establish the service that support the small
- TR 0.252*** 0.064* size firms in the technological context. In our
- TI 0.017*** -0.005 opinion this is particularly important in Portugal
- ISA 0.041*** 0.007
Organizational context given the relative importance of small businesses in
- PBEC 0.079*** -0.007 the economy (Vicente and Martins, 2008). Finally,
- ITTP 0.061*** 0.144*** our study sought to help firms become more
- AITSF 0.011 0.051*** effective in moving from a traditional channel to the
- IEN 0.100*** 0.032
Environmental context internet by identifying the profile of early web site
- WEBP 0.003*** 0.003*** adopters.
Controls As in most empirical studies, our work is limited in
- SER 0.044*** 0.056** several ways. The cross-sectional nature of this
Note: * p-value<0.10; ** p-value<0.05; *** p-value<0.01. study does not allow knowing how this relationship
will change over time. To solve this limitation the
future research should involve panel data. Another
5 CONCLUSIONS limitation of our work is that it only investigates web
site adoption decision. To provide a more balanced
Within the context of an increased use of Internet view of firms’ IT adoption decision, other Internet
Business Solutions, such as web sites, this study fills Business Solutions, such as e-commerce should also
a gap in the literature by comparing the relative be examined.
importance of the factors influencing the adoption of
web sites for small and large firms. The theoretical
framework incorporates most of the facilitators and ACKNOWLEDGEMENTS
inhibitor factors identified in other studies. The
research model evaluates, for small and large firms, We would like to acknowledge the National Institute
the impact of three technological factors, four of Statistics (INE) for providing us with the data.
organizational factors and one environmental factor
on the web site adoption decision. Using a
representative sample of Portuguese small and large
firms, the estimation results for this comparative REFERENCES
study reveal that the important determinants of web
site adoption decision vary with size of a firm. Other Al-Qirim, N. (2007) The adoption of eCommerce
communications and applications technologies in
studies in this domain (Daniel and Grimshaw, 2002,
small businesses in New Zealand. Electron. Commer.
Premkumar, 2003) also suggested that the problems, Rec. Appl., 6, 462-473.
opportunities, and management issues encountered Barua, A., Konana, P., Whinston, A. B. & Yin, F. (2004)
by small business in the IT area are different from Assessing internet enabled business value: An
those faced by their larger counterparts. However, exploratory investigation. MIS Quart, 28, 585-620.
our study provides a more in depth analysis since it Beatty, R. C., Shim, J. P. & Jones, M. C. (2001) Factors
identifies those factors that more or less relevant for influencing corporate web site adoption: a time-based
large/small firms and quantifies its impact on web assessment. Information & Management, 38, 337-354.
site adoption decision. These findings have practical Bertschek, I., Fryges, H. & Kaiser, U. (2006) B2B or Not
to Be: Does E-commerce Increase Labor Productivity?
implications for managers and policy makers.
International Journal of the Economics of Business,
Firstly, policy makers should be conscious that the 13, 387-405.

375
ICE-B 2008 - International Conference on e-Business

Black, S. E. & Lynch, L. M. (2001) How to compete: The Johnson, R. & Wichern, D. (1998) Applied Multivariate
impact of workplace practices and information Data Statistical Analysis, New Jersey, Prentice Hall.
technology on productivity. Review of Economics and Kiiski, S. & Pohjola, M. (2002) Cross-country diffusion of
Statistics, 83, 434-445. the Internet. Information Economics and Policy, 14,
Bresnahan, T. F., Brynjolfsson, E. & Hitt, L. M. (2002) 297-310.
Information technology, workplace organization, and Konings, J. & Roodhooft, F. (2002) The effect of e-
the demand for skilled labor: Firm-level evidence. business on corporate performance: Firm level
Quarterly Journal of Economics, 117, 339-376. evidence for Belgium. Economist-Netherlands, 150,
Brynjolfsson, E. & Hitt, L. M. (2000) Beyond 569-581.
Computation: Information Technology, Organizational Kraemer, K. L., Dedrick, J., Melville, N. & Zhu, K. (2006)
Transformation and Business Performance. Journal of Global E-Commerce: Impacts of National
Economic Perspectives, 14, 23-48. Environments and Policy, Cambridge, UK.
Caselli, F. & Coleman, W. J. (2001) Cross-country Kuan, K. K. Y. & Chau, P. Y. K. (2001) A perception-
technology diffusion: The case of computers. based model for EDI adoption in small businesses
American Economic Review, 91, 328-335. using a technology-organization-environment
Chau, P. (1999) On the use of construct reliability in MIS framework. Information & Management, 38, 507-521.
research: a meta-analysis. Information & Kwon, T. H. & Zmud, R. W. (1987) Unifying the
Management. 35(4), 217-227. fragmented models of information systems
Clayton, K. (2000) Microscope on micro businesses. implementation. In critical issues in Information
Australian CPA, 2, 46–7. Systems Research (Boland RJ and Hirschheim RA,
Daniel, E. M. & Grimshaw, D. J. (2002) An exploratory Eds). IN WILEY, J. (Ed.) New York.
comparison of electronic commerce adoption in large Lange, T., Ottens, M. & Taylor, A. (2000) SMEs and
and small enterprises. Journal of Information Barriers to Skills Development: a Scottish Perspective.
Technology, 17, 133-147. Journal of Industrial Training, 24, 5-11.
Dedrick, J., Gurbaxani, V. & Kraemer, K. L. (2003) Lee, G. & Xia, W. D. (2006) Organizational size and IT
Information technology and economic performance: A innovation adoption: A meta-analysis. Information &
critical review of the empirical evidence. Acm Management, 43, 975-985.
Computing Surveys, 35, 1-28. Martins, M. R. F. & Raposo, P. (2005) Measuring the
Dholakia, R. R. & Kshetri, N. (2004) Factors Impacting Productivity of Computers: A Firm Level Analysis for
the Adoption of the Internet among SMEs. Small Portugal. Electronic Journal of Information Systems
Business Economics, 23, 311-322. Evaluation, 8, 197-204.
Eurostat (2006) The internet and other computer networks Martins, M. R. F. & Oliveira, T. (2005) Characterisation
and their use by European enterprises to do eBusiness. of Portuguese Organizations regarding Investment in
Statistics in focus. Information and Communication Technologies –
Gibbs, L. J. & Kraemer, K. L. (2004) A Cross-Country Application of Multivariate Data Analysis Techniques.
Investigation of the Determinants of Scope of E- IN REMENYI, D. (Ed.) The 12th European
commerce Use: An Institutional Approach. Electronic Conference on Information Technology Evaluation.
Markets, 14, 124-137. Turku, Finland, Academic Conferences Limited.
Grandon, E. E. & Pearson, J. M. (2004) Electronic Mata, F., Fuerst, W. & Barney, J. (1995) Information
commerce adoption: an empirical study of small and technology and sustained competitive advantage: A
medium US businesses. Information & Management, resource-based analysis. MIS Quart, 19, 487–505.
42, 197-216. Mehrtens, J., Cragg, P. B. & Mills, A. M. (2001) A model
Harindranath, G., Dyerson, R. & BARNES, D. (2008), ICT of Internet adoption by SMEs. INFORMATION &
Adoption and Use in UK SMEs: a Failure of MANAGEMENT, 39, 165-176.
Initiatives?, Electronic Journal of Information Systems Parker, C. M. & Castleman, T. (2007) New directions for
Evaluation, 11(2), 91-96. research on SME-eBusiness: insights from an analysis
Hollenstein, H. (2004) Determinants of the adoption of of journal articles from 2003 to 2006. Journal of
Information and Communication Technologies (ICT). Information Systems and Small Business, 1, 21-40.
An empirical analysis based on firm-level data for the Porter, M. E. (2001) Strategy and the Internet. Harvard
Swiss business sector. Structural Change and Business Review, 79, 62-78.
Economic Dynamics, 15, 315-342. Premkumar, G. (2003) A meta-analysis of research on
Hong, W. Y. & Zhu, K. (2006) Migrating to internet-based information technology implementation in small
e-commerce: Factors affecting e-commerce adoption business. Journal of Organizational Computing and
and migration at the firm level. Information & Electronic Commerce, 13, 91-121.
Management, 43, 204-221. Premkumar, G. & Ramamurthy, K. (1995) The role of
Iacovou, C. L., Benbasat, I. & Dexter, A. S. (1995) interorganizational and organizational factors on the
Electronic data interchange and small organizations: decision mode for adoption of interorganizational
Adoption and impact of technology. Mis Quarterly, systems. Decision Sciences, 26, 303-336.
19, 465-485.

376
 A COMPARISON OF WEB SITE ADOPTION IN SMALL AND LARGE PORTUGUESE FIRMS

Tan, Z. & Ouyang, W. (2004) Diffusion and Impacts of Did your firm use the following internal security
the Internet and E-commerce in China. Electronic applications, during January 2006? (Yes No)
Markets, 14, 25-35. a) Virus checking or protection software
Thong, J. Y. L. (1999) An Integrated Model of b) Firewalls (software or hardware)
Information Systems Adoption in Small Businesses.
Journal of Management Information Systems 15, 187-
c) Secure servers (support secured protocols such as
214. http)
Tornatsky, L. & Fleischer, M. (1990) The Process of d) Off-site data backup
Technology Innovation, Lexington, MA, Lexington e) Subscription of a security service (e.g. antivirus or
Books. network intrusion alert)
Vicente, M. R. C. & Martins, M. O. (2008) Information f) Anti-spam filters (unsolicited e-mails)
Technology, efficiency and productivity in SMEs:
Evidence for Portugal. paper presented at the Telecom Access to the IT system of the firm
Paris Tech Conference on The Economics of ICT. Did any of those people access the firm's computer
Paris. system from the following places during January
Zhu, K., Kraemer, K. & XU, S. (2003) Electronic business 2006? (Yes No)
adoption by European firms: a cross-country a) From home
assessment of the facilitators and inhibitors. European b) From customers or other external business
Journal of Information Systems, 12, 251-268.
partners’ premises
Zhu, K., Kraemer, K. K. L., XU, S. & DEDRICK, J.
(2004a) Information technology payoff in e-business c) From other geographically dispersed locations of
environments: An international perspective on value the same firm or firm group
creation of e-business in the financial services d) During business travels, e.g. from the hotel,
industry. Journal of Management Information airport etc.
Systems, 21, 17–54.
Zhu, K. & Kraemer, K. L. (2002) e-Commerce metrics for
net-enhanced organizations: Assessing the value of e-
commerce to firm performance in the manufacturing
sector. Information Systems Research, 13, 275-295.
Zhu, K. & Kraemer, K. L. (2005) Post-adoption variations
in usage and value of e-business by organizations:
Cross-country evidence from the retail industry.
Information Systems Research, 16, 61-84.
Zhu, K., Kraemer, K. L. & XU, S. (2006) The Process of
Innovation Assimilation Assimilation by Firms in
Different Countries: A Technology Diffusion
Perspective on E-Business. MANAGEMENT
SCIENCE, 52, 1557-1576.

APPENDIX
Technological integration
Did your firm's IT systems for managing orders link
automatically with any of the following IT systems
during January 2006? (Yes No)
a) Internal system for re-ordering replacement
supplies
b) Invoicing and payment systems
c) Your system for managing production, logistics or
service operations
d) Your suppliers’ business systems (for suppliers
outside your firm group)
e) Your customers’ business systems (for customers
outside your firm group)
Internal security applications

377
 PROTOCOL OF AUTHENTICITY TO PROVIDE LEGAL
SECURITY IN E-CONTRACTS
A Prototype

João Fábio de Oliveira, Cinthia O. de A. Freitas and Altair Santin

Pontifical Catholic University of Parana - PUCPR
R. Imaculada Conceição, 1155, Prado Velho, 80215-901, Curitiba - PR, Brazil
[email protected], [email protected], [email protected]

Keywords: Law and Internet, Consumer Protection, Authenticity.

Abstract: This paper discusses the security problems on the contracts deals over the Internet and describes a protocol
of authenticity that will keep audit trails from the activities during the Web hiring. These audit trails will be
stored digitally, either on the side of the provider or on the consumer. It is understood that it is legal
obligation of the provider to logs and ensure the integrity of the data related to the operations on the
Internet-based commerce. because in situations of dispute can occur a reversal of the burden of proof. Thus,
this prototype ensures confidence on the e-contracts, logging relevant information that help in the
identification of the parties, using a Plug-in software installed on the e-commerce provider and in the
consumer machine, executable in the Web. Moreover, it is important to remind about the security in
contracts agreements over the Internet is an essential feature because it allows the consumer a guarantee of
contract award, since it maintains the integrity of the document, and also can be presented as evidence to the
Judiciary, helping in litigation and satisfying the premises of the legal acceptance of digital documents.

1 INTRODUCTION environment, it introduces a restricting factor and

leads us to a universe of studies on the security,
The increased use of Internet in people's daily lives trustworthy, confidentiality, integrity, and
is already a reality as a basic and essential tool in authenticity aspects. As well as the legality of such
day-to-day of the societies, for things like paying operations given the doubtful facts questionable by
bills, consulting phone catalogs and maps, any of the parties involved in the transaction (e-
relationship between people, electronic messaging contract).
and even to buy objects and consumer services. In face of this, there is a search for technical
According to the Center for Studies of solutions for security infrastructure for e-business
Information and Communications Technology including the security of the information exchanged
(Brazilian Internet Management Committee, 2006), and the storage of this information, especially when
about 14.49% of Brazilian homes already have it comes to making it trusted as far as accuracy of
Internet access, this information are from the last the content stored. In this sense, the cryptography
poll done in 2006, which represents a 1.56% growth and digital signing methods have contributed for the
over the 12.93% of 2005. This growing universe of security of online transactions: e-contract, e-
users and potential consumers of online products commerce, e-business (Behrens, 2007) (Garfinkel,
represents a great concern from the technical and 1997).
legal point of view, due to the yearly increase in the Thus, this paper discusses our prototype for
number of problems that need to be handled by each authentication protocol based on the consumer–
of the related sciences. provider relationship on the contracts over the
On the other hand, Internet brings some Internet, since this protocol defines technical and
information security concerns, since the traceable parameters of the transaction, and provides
documentation is no longer kept in physical form, legal security in e-contracts, both in the provider's
such as paper, but rather stored electronically server and in the consumer's equipment. We discuss
through digital means. While the Internet simplifies our prototype for demonstrating the practical
commercial operations that take place in the digital feasibility of the proposed protocol. This protocol

378
 PROTOCOL OF AUTHENTICITY TO PROVIDE LEGAL SECURITY IN E-CONTRACTS - A Prototype

must be used along with the web server of the problems involving the transmission of data, and
provider and the Web browser of the consumer, and provides a well-defined service to the upper layer
is been developed as a Web Plug-in. protocols based on using services from some lower
This paper is divided into five sections. Section 2 layers. Upper layers are logically closer to the user
highlights the security issues on the web. Section 3 and deal with more abstract data, relying on lower
summarizes the e-contracts since its definition and layer protocols to translate data into forms that can
the relevant aspects for the protocol of authenticity, be physically transmitted. Layer 5, nominated as
which provides legal security in e-contracts. The application layer, specifies and implements the
proposed protocol is presented in Section 4. Some software which interact with the end users. It is at
final considerations are given in Section 5. this level that all concerns related to the information
security should be implemented, in other words, the
applications at the user level must include treatment
2 SECURITY ISSUES ON THE mechanisms that are considered secure enough to,
on one hand, give the end user the assurance that his
WEB network transaction is secure, free of risks of content
modifications, and on the other hand, provide
There are different aspects of security issues on the verified technical conditions of mechanisms
web, such as physical security, personal security, considered secure, such as the use of cryptography
operations security, communications security, and algorithms at the applications level (Schneier, 1996).
network security. Normally, the systems are
concerned about the application of the protection in Table 1: The 5-Layer TCP/IP Model.
e-anything or e-initiative, or specifically e-business,
taking into account the state of the art technological Layer Protocol
infrastructure. The main goal of security for e- 5 – Application HTTP, DNS, SMTP,...
business applications is to protect networks and their 4 – Transport TCP, UDP, DCCP...
applications against attacks, ensuring information 3 – Network IP, IGMP, ICMP...
availability, confidentiality, integrity, auditing, 2 – Data Ethernet, Wi-Fi,
authorization, and authentication (Huang et al., 802.11(WLAN)...
2008) (Meier, 2003). This paper goes beyond to 1 – Physical Modem, PLC, optical fiber...
provide a secure environment. We are here Traffic on the Internet is a client-server software
considering legal security of e-contracts. application, which is executed through a direct user
A secure environment is ensured by combination interface, known as the web browser. In this
of different factors, such as data security, environment, several applications are written in the
networking security, policy, and management of application layer protocol of the TCP/IP known as
information system security (ISS) (Akhter & Kaya, http (hypertext transfer protocol) (Garfinkel, 1997).
2008). Therefore, we need discuss the basis of the Since there are no mechanisms defined in the
information transferred during transactions over the TCP/IP protocol itself, the application is responsible
Internet. to define and implement additional security
The starting point of the information transferred algorithms; minimizing the impact of the
on Internet is the TCP/IP (Transmission Control vulnerabilities that exist on the internet protocol.
Protocol/Internet Protocol) protocol. This protocol, Thus, the security issues on the Web are divided into
in its version 4, has been consolidated for using in three categories (Garfinkel, 1997):
Internet (Comer, 1991). This protocol does not • Security of the Web server, the server's
include security mechanisms for information activities, and stored data: guarantee that
transfer, leaving this for the applications which are the information was not modified or
developed for the end users, which is the reason for distributed without user authorization;
the concern about the related criteria to the • Data security in the computers network:
protection of the content in transit. This means that guarantee that the transmission of
the transferred information between two different information between the server and the
places on Internet, regardless of its physical location, Web browser has a security level based
can be captured by a protocol analyser (WireShark, on well-known standards, such as
CommView, Ghost). So the desired information can cryptography or digital signature (Meier,
be visualised. 2003). Cryptography refers to how your
The TCP/IP protocol specification has a application enforces confidentiality and
conceptual segmentation into five layers, as shown integrity. Digital signature corresponds a
in Table 1 (Comer, 1991). Each layer solves a set of

379
ICE-B 2008 - International Conference on e-Business

countermeasure techniques that can be This is the main problem of cross-border e-

used to reduce risk of tampering with commerce, but this topic is beyond this paper.
data (Meier, 2003); Recognising the value of arbitration as a method
• Data security in the user computer: of settling disputes arising in the context of
guarantee that user’s computer is as international commercial relations and being
protected as possible (firewalls, convinced that the establishment of rules for ad hoc
antivirus, anti-spyware, anti-spam, etc.). arbitration that are acceptable in countries with
A typical firewall helps to restrict traffic different legal, social and economic systems would
to HTTP, but the HTTP traffic can significantly contribute to the development of
contain commands that exploit harmonious international economic relations; the
application vulnerabilities (Meier, 2003). United Nations Commission on International Trade
Therefore, taking into consideration the security Law (UNCITRAL) was established by the United
problems on Internet, these vulnerabilities can be Nations General Assembly in 1966 to promote the
analyzed from the point of view of the contracts progressive harmonization and unification of the law
deals over the Internet. In this case, it is not of international trade. Brazil not adopts these rules
necessary provide information security but is and there are no clear rules or a specific law for
primordial provide legal security for e-contracts. litigation situations involving e-commerce.
Chiu et al. (2003) presents that e-contract
enforcement can be divided into multiple layers and
3 E-CONTRACT perspectives, which has not been adequately
addressed in the literature. This problem is
challenging as it involves monitoring the enactment
A contract is a legally binding agreement.
of business processes in counter parties outside an
Agreement arises as a result of offer and acceptance,
organization's boundary. The authors present an
but a number of other requirements must be satisfied
architecture for e-contract enforcement with three
for an agreement to be legally binding as following
layers: document layer, business layer, and
(Martin, 2003):
implementation layer. In the document layer,
• The parties must have an intention to
contracts are composed of different types of clauses.
create legal relations;
In the business layer, e-contract enforcement
• The parties must have capacity to activities are defined through the realization of
contract; contract clauses as business rules in event-condition-
• The agreement must comply with any action (ECA) form. In the implementation layer,
formal legal requirements (it may be cross-organizational e-contract enforcement
oral, written, partly oral and partly interfaces are implemented with contemporary
written, or even implied from conduct); Enterprise Java Bean and Web services. They
• The agreement must be in accordance of present a methodology for the engineering of e-
the law; contracts enforcement from a high-level document-
• The agreement must not be rendered view down to the implementation layer based on this
void either by some common-law or architecture, using a supply-chain example. As a
statutory rule or by some inherent defect, result, e-contracts can be seamlessly defined and
such as operative mistake. enforced.
However, when such a contract becomes valid, This paper present an authentication protocol to
that is, legally binding and enforceable, is regulated provide legal security in e-contracts implemented in
by contract law that may differ from country to application layer based on document-view. The idea
country, even if subject to International Law. Based is to be able answering some questions such as:
on these considerations, an e-contract is a valid Does the consumer want to buy, is s/he able to buy?
contract deals over the Internet. Simplest example is When is the contract considered to be concluded?
Amazon book store buying. When everything goes There is no clear rule. To avoid problems and
right, it is of no major importance when exactly a protect the consumer (as provided in the Brazilian
contract is concluded and according to which law. Consumer Code) we are developing the protocol to
However, when something goes wrong such as, the capture and save a set of data from consumer and
product does not delivery, payment is not provider. This set of data can be easily visible and
effectuated or is wrongly charged, or the product is accessible by the consumer and accepted by him/her.
damaged; then information about the contract Thus, when constituting the contract elements in a
conclusion as well as governing law may be crucial. virtual environment, the following fundamental

380
 PROTOCOL OF AUTHENTICITY TO PROVIDE LEGAL SECURITY IN E-CONTRACTS - A Prototype

structure items must be established: user-consumer, can be presented in Judicial processes. Thus, the
provider, e-contracts and contract location. proposed protocol of authentication (Plug-in) is
organized in two parts:
• Server: is installed on the Web server
4 PROTOCOL OF provider and follows the standards of the
http protocol and Java-web, and is
AUTHENTICITY configured as an extension of the services of
the server and offered to the consumer as a
In e-contract, such as in the purchase of consumer Plug-in for web browser of provider;
goods from e-commerce Web sites, the consumer • Client: is installed on the consumer’s
will be taking advantage all infra-structure of computer and follows a specific procedure
communication defined in the Internet network. for initial installation of the Plug-in,
When the consumer performs a purchase operation considering the consent of the consumer to
on the providers’ server through her Web browser, accept the installation of this software on her
she lacks the mechanisms to provide physical and computer. The Java programming language
legal evidence of the content accepted during the was chosen due to its flexibility in selecting
transaction at a technical level. In other words, there of information on Internet environment.
are no effective logs on your computer that store or
restore the history of transaction between the
consumer and the provider. Therefore, this Section
presents the theoretical aspects explored in the
proposed protocol and the implementation issues.

4.1 Theoretical Aspects

In practice the hiring conducted on the Internet can
be done by e-mail or directly at the Web site (e.g. in
clicking a button "I accept"). This implies that Figure 1: Plug-in model.
specifically in the context of Web communication,
the level of information security applied in http It is possible, therefore, to present an overview of
protocol guarantees a minimum of security based on the authentication protocol, in which the consumer
mechanisms of encryption. Indeed, the use of SSL possesses in her computer the record of the
(Secure Sockets Layer) in Web applications applies transactions done in the web site of the providers, in
encryption namely, point-to-point, so that the order to extract a report with the information
information is transferred ciphered on the Internet regarding such transaction. On the other hand, the
(Garfinkel, 1997). The encryption point-to-point protocol allows to confront the information
means doing a cryptographic channel between the transferred between both parties, as shown on Figure
provider and consumer applications layers. 1. The request done by the consumer is based on the
The applications developed for the Internet http protocol, then the Plug-in captures the packages
environment, following the client-server on this level of the protocol as well as other
architecture, employ the standard TCP/IP protocol technical information on the TCP/IP level and stores
infra-structure, remaining however, vulnerable to the all data in the log files.
various problems typical to this environment.
Specifically in Web communication, when 4.2 Implementation Issues
communications between the server and the
contractors’ web browser employ the SSL protocol Considering the theoretical aspects presented before,
in the cryptography of the data transferred over the the proposed protocol defines that the consumer,
network, the risk of fraud in the transferred content when accessing the web site of the providers through
is reduced by the cryptography (Garfinkel, 1997). her browser, has an icon with information about the
However, in litigation situations it is necessary authentication protocol available in order to allow
restore the history of transaction between the her with explicit consent and authorization, to
consumer and the provider and there is no protocols download the Plug-in to her computer.
or tools that capture, store and restore this kind of The protocol determines the capture of consumer
information, specially providing legal security. In information, as shown in Table 2, which takes place
this context, legal security represents the consumer’s during the initial interaction process when the
guarantees that the transaction hired on the Internet consumer accepts the installation of the Plug-in.

381
ICE-B 2008 - International Conference on e-Business

This initial registration makes up the official authentication protocol are available in the data
identification of the consumer. package captured in the Ethernet frame, as shown in
Figure 3. The information captured are stored in the
Table 2: Consumer Information – Initial Registration. log file in a sequence and logically structured in data
Field Description files in encrypted text format.
name_contr Consumer’s name
Table 3: e-Contracts Information.
address_contr Consumer’s address
phone_home_contr Consumer’s Home Field Description
Telephone IP_consumer_Internet Consumer’s IP number as
phone_celular_contr Consumer’s Mobile Phone an Internet user
id_contr Consumer’s Legal IP_consumer_real Consumer’s IP number
Identification (consumer machine)
type_id ID, Drivers License, Mask_IP_consumer_re Networking mask
Passport, Working Papers al (consumer machine)
tax_id_contr Consumer’s tax ID Gateway_consumer_re Default gateway
al (consumer machine)
After this initial registration, the algorithm
DNS_consumer_real DNS configured
provides verification of the Plug-in installed on the
(consumer machine)
equipment of the provider, installing it when it is not
IP_provider_Internet Provider’s IP number
already done. From this point on, the operations (provider server)
performed by the consumer on the web site of the IP_ provider_real Provider’s IP number
provider keep all information stored in a log file (LAN server)
recorded on both sides (provider and consumer), as Mask_IP_provider_rea Networking mask
specified in Table 3. l (provider server)
The log file is encrypted through the Triple DES Gateway_provider_real Default gateway
(Triple Data Encryption Standard) algorithm, (provider server)
developed at IBM and published as a standard in DNS_provider_real DNS configured
1977 (Schneier, 1996). Generally TDES uses three (provider server)
different keys and has a key length of 168 bits (3 x date_acess Date: DD/MM/YYYY
56-bit DES), resulting in a symmetric algorithm. time_acess_consumer Hour: HH:MM:SS
Thus, by being symmetric, the key used for time_acess_provider Hour: HH:MM:SS
encryption and decryption of the file is the same, port_acess_consumer Consumer’s TCP or UDP
making it a fast processing algorithm, considering port number
the log file to be processed and ensuring access to porta_acess_provider Provider’s TCP or UDP
the information. port number
In this way, the consumer will be able to URL_provider web address accessed by
generate her report directly through her computer, the consumer
without depending on the information saved in the route_IP Route between
server of the provider, as is the case with the use of consumer’s IP and
other cryptography algorithms that apply provider’s IP
asymmetric keys (which relies on the exchange of
public between both parties).
In the Plug-in installed on the computer of the
consumer is the function to visualize the log file
stored, allowing her extracting the information about
the operations accomplished in printed format as
evidence of the various Web interactions performed
at the web site of the providers. In legal terms, this
report will allow the consumer to provide evidence
as an official document registered at Notary's Office
(Rezende, 1997). Figure 2: Capturing the Data Package.
The capture data follows the structure defined in
Figure 2, which is an example of the data capture The technique for capturing the packages is
using the Wireshark software, keeping the reference based on the use of the libpcap/winpcap libraries,
to the TCP/IP protocol. The information necessary which are low level software libraries available for
to comply with the fields defined in the

382
 PROTOCOL OF AUTHENTICITY TO PROVIDE LEGAL SECURITY IN E-CONTRACTS - A Prototype

programming code of development. These libraries access to the providers, as well as the contracts over
come from network traffic information according to the Internet. In this situation, is important that the
the interface used, for example, Ethernet and consumer can restore, from her own computer, data
Wireless WIFI. and information on the litigated web site. Or,
Global Package Package Package Package Package Package
furthermore, the consumer should be in condition to
Header Header Data Header Data Header Date validate the information given by the provider. Thus,
the use of the authentication protocol will be, in fact,
Figure 3: Format of the Captured Data Package. mapping the operation accomplished through the
Internet by means of the log file, with a full and
This library provides functions that capture secure record of the main elements of the e-contract.
packages in the format of the basic network, on
which there are the header and the individually
separate data. Inside of the TCP/IP protocol in its
didactic classification, it is possible to separate the
ACKNOWLEDGEMENTS
various levels of information (protocol and user
information) allowing for the recording of this The authors wish to thank Dr. Antonio Carlos Efing
(PPGD-PUCPR) which have collaborated to this
information to the log file (Comer, 1991).
work. This work has been supported by CNPq (grant
In future work, we plan to conduct case studies
471627/2006-2 and grant 476637/2006-6).
to evaluate our protocol taking into consideration the
following parameters:
• Performance: computing the impact of
the protocol in the consumer machine REFERENCES
and on the e-commerce provider based
on CPU use. The data collecting will be Atkins, D., Buis, P., Hare, C., Kelley, R., Nachenberg, C.,
done using the SNMP (Simple Network Nelson, A.B., Phillips, P., Ritchey, T., Sheldon, T.,
Management Protocol); Snyder, J. (1997). Internet security professional
• Response time: which is the time a reference, New Riders Publishing, 2nd edition.
generic system or functional unit takes to Behrens, F. (2007). Digital signature and legal business.
react to a given input. In this case, we Jurua, 1st edition.
will verify the response time between the Brazilian Internet Management Committee (Comitê
consumer machine and e-commerce Gestor da Internet Brasil), Center for Studies of
provider. The data collecting will be Information and Communications Technology (2006).
TIC Domicílios e Usuários (ICT Homes and Users.
done using the ICMP (Internet Control
Available: http://www.cetic.br/, [11-march-2008].
Message Protocol) considering the echo- Chiu, D.K.W., Cheung, S.C., Till, S. A. (2003). Three-
request and echo-reply facilities. layer architecture for e-contract enforcement in an e-
This evaluations can demonstrate potential service environment. In 36th Annual Hawaii
scenarios that may benefit from this research. International Conference on System Sciences.
Comer, D. E. (1991). Internetworking with TCP/IP:
principles, protocols, and architecture, Prentice-Hall
International, Inc., 2nd edition, Vol. 1.
5 FINAL CONSIDERATIONS Garfinkel, S., Spafford, G. (1997). Web security &
commerce. O’Reilly & Associates, Inc.
In litigation situations involving e-commerce, the Martin, E.A. (2003). A Dictionary of Law. Oxford
greatest difficulty is in verifying proof of the University Press.
contract established. This difficulty arises due to the Meier, J.D., Mackman, A., Vasireddy, S., Dunner, M.,
fact that the relations are no longer necessarily face- Escamilla, R., Murukan, A. (2003). Improving web
to-face, therefore requiring the use of additional application security: threats and countermeasures,
mechanisms to carry out these contracts as well as Microsoft.
Rezende, A.C.F. (1997). Notary's Office and the perfect
computer tools that allow the registration and notary: property law and notary’s activities. Copola
evidence that the contract was made. Thus, the Livros. 1st edition.
authentication protocol proposed in this article Schneier, B. (1996). Applied cryptography: protocols,
allows both interested parties, consumer and algorithms, and source code in C. John Wiley & Sons,
provider, to keep registry logs with information Inc., 2nd edition.
about the dealing contract. This audit trail is
composed of a variety of information, such as the IP
record of the equipment involved in the transaction.
Therefore, the consumer can issue reports on the

383
USING TECHNOLOGY ACCEPTANCE MODEL TO EVALUATE
USERS’ ATTITUDE AND INTENTION OF USES

Dauw-Song Zhu
Business Administration & Accounting Department, National Dong Hwa University
No. 1, Sec. 2, Da Hsueh Rd. Shou-Feng Hualien, Taiwan 974
[email protected]

Chih-Te Lin
Business Administration Department, National Dong Hwa University &
Lecturer of Food and Beverage Management Department, Taiwan Hospitality & Tourism College
No. 2, Lane 46, Jhongmei 3rd St, Hualien, Taiwan 970
[email protected]

Keywords: Online game, technology acceptance model (TAM), service quality.

Abstract: Many empirical studies have pointed out that the technology acceptance model (TAM) can be used to
explain whether users can accept a new information technology. Therefore, this study has adopted TAM to
investigate external factors that affect gamers’ acceptance of online games. In this study, system quality,
information quality, and service provider’s characteristics were taken as external variables. It was
discovered that in the aspect of perceptions, system quality had positive effects on perceived ease of use and
perceived usefulness. Service provider’s characteristics had positive effects on perceived usefulness and
perceived trust. Besides, in the relationship between user’s perception and attitude and intention, the
research finding was consistent with TAM has suggested; i.e. perceived ease of use had positive effects on
perceived usefulness and user’s attitude, and perceived usefulness had positive effect on user’s attitude.
Finally, user’s perceived trust and attitude would be positively correlated with user’s intention of use.

1 INTRODUCTION M.B., Persinger, E.S., Grunhagen., M., Witte,

C.L., 2007). The TAM model was derived from
With the advancement of information technology, Fishbein and Ajzen’s (1975) Theory of Reasoned
popularity of the Internet, and the gradual prevalence Action and has received considerable support by
of broadband networks, the output value of online many researchers (e.g. Jackson, Chow, & Leitch
games has been increasing in a fast speed. In fact, 1997; Venkatesh & Davis 1996). Further, a number
Internet websites are accessible by consumers from of researchers have examined Internet usage via the
around the globe, and this has contributed to the Tam framework (Gefen & Staub 2000; Venkatesh, &
globalization of markets over the past three decades Massey 2003), and the current research use the TAM
(Yip, 2000), it is very common for researchers to better understand individual attitudes toward
studying the international environment to select one technology within the international Internet shopping
country and use it as a basis for empirical inquiry. environment.
Clearly, managers need to better understand the Moreover, the identified human factors are
growing use of Internet shopping sites and those mainly user perceptions about the utility of this type
consumer characteristics that encourage repeat visits of technology: perceived ease of use and perceived
to these sites. One key consumer characteristic is usefulness (Davis, 1989).Some related studies have
their willingness to use and accept the new also revealed that when one enters the fully
technologies. Since researchers need a framework to immersed state, he will feel totally involved in the
effectively evalue new online phenomena, the activity, his consciousness will be very narrow, his
technology acceptance model (TAM) is applicable in attention is only focused on the activity, and his
this context (Savitskie, K, Marla B Royne awareness is partially lost (Webster et al., 1993;

384
USING TECHNOLOGY ACCEPTANCE MODEL TO EVALUATE USERS’ ATTITUDE AND INTENTION OF USES

Koufaris, 2002). However, the gaming market is indicators.

very competitive, and several new games are
released every month. Thus, when players have a 2.3 Service Provider’s Characteristics
certain degree of ‘flow’ experiences from a new (SPC)
game, they may quit playing it due to awful external
quality of the game. As discussed above, the purpose Saeed et al. (2003) conceived that service provider’s
of this study is to investigate the effects of external characteristics are important because the Internet is a
factors, such as system quality, information quality, virtual channel which creates more sense of
service provider’s characteristics on the perceptions uncertainty to Internet users. Stronger service
of online gamers, and find out whether gamers' provider’s characteristics will enhance users’ trust
perceptions have positive effects on their attitude for and consumers’ perceptions will not be affected.
the game and intention of use, in hope of providing Saeed et al. further proposed the following indicators
the research result as a reference for online game to measure service provider's characteristics,
companies to enhance service quality for customers. including size, reputation, and participation costs. In
Jarvenpaa, Tractinsky, and Vitale (2000), it was
proposed that perceived reputation and perceived
2 LITERATURE REVIEW size of service providers will affect users’ trust for
service providers. Among domestic studies, Chen
2.1 System Quality (SQ) Chuen-Liang (2002) pointed out that online game’s
brand image will positively affect users' intention of
Website quality, system quality, and game quality are use, and brand image can be evaluated by corporate
significantly influential to gamer satisfaction. In the image, word of mouth, popularity, and reputation.
I/S success model proposed by DeLone(1992), 12
empirical studies about system quality were 2.4 Technology Acceptance Model
investigated, and 18 indicators of system quality
were proposed. These indicators include ease of use, Over the past 10 years, Technology Acceptance
usefulness, system accuracy, system flexibility, Model (TAM) has been empirically proven to be an
system reliability, response time, and etc. However, important explanatory model for personal acceptance
Mckinney, Yoon, and Zahedi (2002) measured or use of new information technology. TAM is a
system quality of Internet shops with the following behavior intention model developed based on the
indicators, including access, usability, navigation, Theory of Reasoned Action (TRA) by Davis in 1989.
and interactivity. In Negash (2003) which focused on It was intended to simplify TRA and find out an
online customer service systems, it was argued that effective behavior model that could be widely
some of DeLone’s system quality indicators are applied to explain or predict the factors affecting the
already outdated. Thus, interactivity and access were use of information technology. In TAM, two definite
proposed as system quality indicators. cognitive beliefs were proposed, namely perceived
usefulness and perceived ease of use. The two
2.2 Information Quality (IQ) beliefs determine an individual’s behavior intention
for using technology through attitude. It has been
As online games will be constantly updated and clearly pointed out that external variables will
expanded, game information is very important for directly influence perceived usefulness and
online gamers. In the evaluation of information perceived ease of use and indirectly affect user's
quality, Lee, Strong, Kahn, and Wang (2002) attitude, intention, and practical use. Based on the
developed 15 indicators to assess the information behaviors of information system users, Seddon et al.
quality of an organization. These indicators are (1997) developed a successful information system
accessibility, appropriate amount, believability, model which definitely pointed out that system
completeness, concise representation, consistent quality and information quality would respectively
representation, ease of operation, free of error, affect user’s perceived usefulness and perceived ease
interpretability, objectivity, relevancy, reputation, of use. Lin and Lu (2000) probed into the behavior
security, timeliness, and understandability. In intentions of World Wide Web (WWW) users, using
Mckinney et al. (2002), only 5 indicators, including information system quality to measure user’s
relevance, timeliness, reliability, and scope, were behavior intention for using WWW. It was
adopted. Negash et al. (2003) proposed to use discovered that system and information quality were
informativeness and entertainment as system quality positively correlated with user’s cognition, attitude,

385
ICE-B 2008 - International Conference on e-Business

and willingness. In a review of studies related to data and the gaming process.
online consumer behaviors, Saeed, Hwang, and Yi Based on Mckinny et al. (2002) and Lee et al.
(2003) proposed a set of integrative structure, in (2002), this study proposed 4 dimensions of
which system quality, information quality, service information quality, namely “relevance”,
quality, and service provider’s characteristics are “timeliness”, “reliability", and “scope”. Relevance
influential to user’s perceived ease of use, perceived refers to the applicability of the game information
usefulness, and perceived trust. provided by the website. Timeliness indicates
In recent years, many empirical studies (David, whether the website can provide latest game
1989; Szajna, 1996, Lederer et al., 2000; Lin and Lu, information. Reliability is defined as the correctness
2000; Moon and Kim, 2001; Hsu and Lu, 2003) have of the game information, and scope refers to the
verified that perceived ease of use would affect coverage of the provided game information.
perceived usefulness, perceived usefulness and As suggested in Jarvenpaa et al. (2000), service
perceived ease of use would affect user’s attitude, provider’s characteristics included two dimensions,
and user’s attitude would further affect user’s reputation and perceived size. Reputation is defined
intention of use. as the prestige of the firm, and perceived size refers
to the scale of the company in this industry.
According to TAM and the study of online guys by
3 RESEARCH METHODS Hsu and Lu (2003), perceived ease of use was
defined as the level of easiness that users feel about
In this study, the questionnaires were mainly the functions of an online game. Based on TAM and
distributed to players of a new online game available Hsu and Lu (2003), perceived usefulness was
for public test. The game was selected according to defined as the level to which users feel that the
the observation of popular forum topics on a online game can achieve the gaming objective. In
well-known game website “Bahamut” Hsu and Lu (2003), gaming objectives included fun,
recreation, messaging, information exchange,
(www.gamer.com.tw) during one week (Apr making friends, chatting, team work, fantasy, hobby,
27~May 3 2005). The statistic result revealed that work, and transaction. This study employed TAM
the “RF Online” forum was almost always on the top and Hsu and Lu (2003) to define attitude as the level
3. During that time, “RF Online” was available for of user's preference for an online game. Based on
public test, so it was selected as the research focus. TAM and Hsu and Lu (2003), this study defined the
Later, a web-based questionnaire was formed and intention of use as the intensity of user’s intention to
published on a professional web-based survey use an online game. After all the questionnaires were
website “MY3Q”. The survey link was also posted collected, we analyze the data and verify the
on the online game forum of BBS at Dong Hwa hypotheses with structured equation model (SEM).
University, famous game forums "Bahamut” and The analysis procedure included two parts, basic
“Game Base”, and some game-related communities analysis and overall model analysis. In the basic
on Yahoo for players to connect to the survey system. analysis, descriptive analysis and reliability analysis
The survey period started from May 5 till May 31 would be processed on SPSS 10. In the overall
2005. A total of 319 valid samples were collected. model analysis, confirmatory factor analysis and
structural equation model analysis would be
3.1 Measurement of Variable performed on Amos 4.0.
According to Mckinny et al. (2002), system quality
was divided into three dimensions, including
“access”, “usability”, and “navigation”. Access was 4 RESULTS
defined as the connection response and access speed
of the game and the website. Usability refers to the In the gender distribution among collected samples,
operations of the user interface in the game, and male players accounted for 84.9% and female ones
navigation is defined as the operation of the user only 15%. Players aged between 19-23 took the
interface on the website. Aladwani and Palvia (2002) largest proportion by 37.9%. 45.4% of them had a
pointed out that security mechanism is also an university or college education background, and
important element when users evaluate the quality of 38.8% had a high school or vocational school
a website. Thus, this study also incorporated education background. In terms of occupation,
“security” as dimension of system quality and 66.7% of them were students. This sample structure
defined it as the level of security of gamers’ personal was similar to those observed in other studies of

386
USING TECHNOLOGY ACCEPTANCE MODEL TO EVALUATE USERS’ ATTITUDE AND INTENTION OF USES

online games but not in the studies of behaviors on every dimension had a mean larger than 3, and
the Internet. As to gender distribution, the ratio of attitude and intention of use even had 4.26 and 4.28,
male and female online gamers was 7:3. In the respectively, indicating that players had a high level
aspect of online shopping, female shoppers of preference for the online game.
outnumbered male ones, by a ratio of 6:4. Other In this study, AMOS 4.0 was used to process the
activities, such as the use of portal sites and Internet structural equation model analysis and verify the
phone, had a relatively even ratio of male to female proposed causal path model. After the path was
users. In the age distribution, if age 24 was viewed established and the samples were applied into the
as a barrier, the ratio of those above 24 to those model, it was discovered that many relationships
under 24 was 7:3. among latent variables were not significant. Thus,
the structure was modified to remove the
Table 1: Path coefficients of the proposed model and insignificant path between exogenous and
verification of hypotheses. endogenous variables.
As to basic goodness of fit, the factor loading of
Variables Beta Ha Result
each indicator ranged between 0.5 and 0.95 and
SQ → perceived 1.375** H1 ○
reached the level of significance. Besides, there was
ease of use no negative deviation. Thus, overall, this model was
IQ → perceived 0.303* H2 ○ compliant with the standard of goodness of fit. As to
ease of use the fitness of overall model, the absolute fit
SQ → perceived NA H3 × measures of the overall theoretic model: χ2/d.f=1.90,
usefulness GFI=0.87, RMR=0.04, and RMSEA=0.05, where
IQ → perceived NA H4 × χ2/d.f, RMSEA, and RMR reached the ideal level,
usefulness and GFI also approached the acceptable level of 0.9.
SPC → perceived 0.859** H5 ○ In the aspect of incremental fit measures,
ease of use AGFI=0.84, NFI=0.90, CFI=0.95, NFI and CFI
Perceived ease of 0.424** H6 ○ exceeded the ideal level of 0.9., and AGFI was also
use → perceived close to the acceptable level of 0.9. Among
usefulness parsimonious fit measures, PNFI=0.80 and
Perceived ease of 0.363** H7 ○ PGFI=0.72, all of which were compliant with the
use → user’s standard.
attitude The path coefficients in the path model and the
results of hypotheses verification were organized in
Perceived 0.314** H8 ○
Table 1. In the aspect of the impact of external
usefulness →
factors on user’s perceptions, user’s perceived ease
user’s attitude of use was affected by system quality and
Perceived 0.833** H9 ○ information quality. Thus H1 and H2 were supported.
usefulness → In the aspect of perceived usefulness, the research
intention of use findings revealed that system quality and
User’s attitude → 0.168** H10 ○ information quality had no effect on perceived
intention of use usefulness, so H3 and H4 were not supported.
**p<0.001 However, service provider’s characteristics and
perceived ease of use had positive effect on
However, the ratio was reversed in the aspect of
perceived usefulness. Thus, H5 and H6 were
online shopping, where those above 24 took the
supported. Perceived ease of use and perceived
largest proportion (6:4). In other activities on the
usefulness had positive influence on user’s attitude.
Internet, both groups had an even distribution. In
Thus, H7 and H8 were supported. Finally, perceived
terms of occupation, most of the online gamers were
usefulness and user’s attitude (H9, H10) would have
students. The ratio of student gamers to non-student
positive influence on intention of use at the same
gamers was 6:4. But in the aspect of online shopping,
time.
those of other occupations took a larger proportion,
and students only accounted for 30%. Besides, no
significant difference between the two groups was
observed in other activities. Finally, in terms of 5 CONCLUSIONS
education background, the majority of online gamers
received college education. It can be discovered that System quality and information quality were

387
ICE-B 2008 - International Conference on e-Business

positively related to player’s perceived ease of use. characteristics was empirically proven. Online users
This indicates that when better and more stable cannot directly interact with service distributors or
systems and correct and rich information are providers, so trust becomes an important element in
provided to game players, game players will feel the evaluation of online information systems, and
perceive more ease of use and the entry obstacle of well-known and large-scale firms can usually lower
the game can be reduced. the level of uncertainty.
No significant effect of system quality and The research of online games is seldom across to
information quality on gamer's perceived usefulness foreign nations, and no measurement scale for online
was discovered. Unlike task-oriented information games is available. This study try to use the scales
systems, online games are entertainment-oriented designed for other information systems, and
information systems. Thus, when players are incorporated the opinions of online gamers to
engaged in online games, they do not necessarily develop an integrated scale for online game systems.
play the game to enhance their game performance or The proposed model had compliant validity and
seek higher efficiency but simply kill some time, reliability. Thus, it can be a reference for further
make friends, get rid of the social bindings. This studies
explains why system quality and information quality Besides, in the aspect of brand image, a good
were not significantly related to perceived management mechanism is helpful for the
usefulness. establishment of a brand. If game service providers
Service provider’s characteristics had positive can make use of brand advantage, they can attract
effects on Perceived usefulness. In online games, it more users to participate in their games and enhance
would be time-consuming to accumulate customer loyalty. Besides, a good brand image can
achievements or cultivate relationships, so if the also help promote other games, create popularity,
service provider is in a small scale or it does not and establish a good reputation. Finally, according to
proper manage with players, the service provider the survey of this research, 54% of the gamers
may shut down the game due to improper reported to play other games in addition to the
management. selected game. This reveals that there are numerous
In this study, TAM was adopted to investigate the choices for online games in Taiwan, and the market
factors affecting online game players’ acceptance of competition is very fierce. As a result, if game
games. The research results were consistent with service providers are imprudent in their management,
those suggested in previous studies. This shows that their games may be easily displaced.
if gamers feel that a game is easy to be familiar with
and get involved in, they will have more preferences
for the game and further increase their intention to 7 RESEARCH LIMITATIONS
carry on using it.
In this study, a web-based survey was adopted and
the reference link was posted on some major forums
6 MANAGERIAL IMPLICATIONS and discussion boards for game players to participate
in the survey at will. Thus, random sampling could
Most of the previous studies of TAM focused on the not be conducted, and deviation of the samples in the
second half of the model, i.e. the effect of user’s representation of the population might occur and
perceptions on their attitude and intention. The main possibly lead to a slight bias of the research results.
contribution of this study is that it probed into the
first half of the TAM model to understand the impact
of external factors on user’s perceptions. REFERENCES
In the previous studies about evaluation of
information systems, external factors were mainly Aladwani, A.M, Palvia, P.C., (2002), Developing and
touched upon in the discussion of the impact of validating an instrument for measuring user-perceived
information quality and system quality on web quality., Information & Management, Vol. 39, No.
information systems. However, with the fast 6, pp 467-476.
development of the Internet, many related online Davis, F. D, Bagozzi, R. P, Warshaw, P. R., (1989), User
information systems have been derived. Thus, the Acceptance of Computer Technology: A Comparison
past evaluation models are no longer applicable to of Two Theoretical Models., Management Science,
the evaluation of online information systems. In this Vol.35, No.8, pp. 982-1003.
study, the importance of service provider’s

388
USING TECHNOLOGY ACCEPTANCE MODEL TO EVALUATE USERS’ ATTITUDE AND INTENTION OF USES

DeLone, W. H., McLean, E. R., (1992), Information Szajna, B., (1996), Empirical Evaluation of the Revised
System Success: The Quest for the Dependent Variable, Technology Acceptance Model., Management Science,
Information Systems Research, Vol. 3, No. 1, pp. Vol.42, No.1, pp. 85-92.
61-95. Venkatesh, V. & Davis, F.D. (1996). A model of the
Fishbein, M., & Ajzen, I.(1975). Belief, attitude, intention, antecedents of perceived ease of use: development and
and behavior: An introduction to theory and research, test. Decision Sciences, 27(3),451-481
Reading, MA: Addison-Wesley. Venkatesh, V., Ramesh, V., & Massey, A.P. (2003).
Gefen, D.& Larcker D.F (1981).Evaluating structural Understanding usability in mobile commerce.
equation model with unobservable variables and Communications of the ACM, 45(12),53-56
measurement error. Journal of Marketing Research, Webster, J., Trevino, L., and Ryan, L., (1993), “The
18(1), 1-30. Dimensionality and Correlates of Flow in
Hsu, C.L, Lu, H.P., (2004), Why do people play on-line Human–Computer Interactions,” Computers in Human
games: an extended TAM with social influences and Behavior, Vol. 9, pp. 411–426.
flow experience, Information and Management, Vol. Yip, G.S (2000). Global strategy in the Internet era.
41, No. 7, pp. 853–868. Business Strategy Review, 11(4),1-14.
Jackson, C.M., Chow, S., & Leitch, R.A.(1997).Toward an
understanding of the behavioral intentions to use an
information system. Decision Sciences, 28(2),357-389
Jarvenpaa, S., Tractinsky, N., Vitale, M., (2000),
Consumer trust in an Internet store, Information
Technology and Management, 1, pp. 45-71.
Koufaris, M.( 2002). Applying the technology acceptance
model and flow theory to on-line consumer behavior.
Information System Research, 13(2), 205–223.
Lederer L. A, Maupin J. D., Sena P. M., Zhuang, Y., (2000),
The technology acceptance model and the World Wide
Web., Decision Support Systems, Vol.29, pp. 269-282.
Lee Y. W., Strong D. M., Kahn B. K., Wang R. Y., (2002),
AIMQ: A methodology for information quality
assessment, Information & Management; Vol. 42, No.
2, pp.133-146.
Lin, C-C. , Lu, H., (2000), Towards an understanding of
the behavioral intention to use a web site, International
Journal of Information Management, vol.20, pp.
197-208.
Mckinney, V., Yoon, K., Zahedi, F.M., (2002), The
measurement of web-customer satisfaction: an
expectation and disconfirmation approach, Information
System Research, Vol. 13, No3, 296-315.
Moon, J.-W., Kim, Y-G., (2001), Extending the TAM for
World-Wide-Web context., Information &
Management, Vol. 38, No. 4, pp. 217-230.
Negash, S., Ryan, T., Igbaria, M., (2003), Quality and
effectiveness in Web-based customer support systems,
Information & Management, Vol. 40, pp. 757-768.
Saeed, K. A., Hwang, Y., Yi, M. Y., 2003, Toward an
Integrative Framework for Online Consumer Behavior
Research: A Meta-analysis Approach, Journal of End
User Computing. Vol.15, No.4, pp. 1-26.
Savitskie, K, Marla B Royne M.B., Persinger,
E.S., Grunhagen., M., Witte, C.L., (2007).
Norwegian Internet Shopping Sites: An Application &
Extension of the Technology Acceptance
Model. Journal of Global Information Technology
Management, 10(4), 54-73.
Seddon, P. B., (1997), A Respecification and Extension of
the Delone and Mclean Mode of IS Success.,
Information Systems Research, Vol. 8, No. 3, pp.
240-253.

389
SEMANTIC INTEROPERABILITY FOR E-BUSINESS IN THE ISP
SERVICE DOMAIN

Jane Hall
Fraunhofer FOKUS, Kaiserin-Augusta-Allee 31, 10859 Berlin, Germany
[email protected]

Stefanos Koukoulas
Eworx S.A., 66 Jean Moreas Str, 152 31 Halandri, Athens, Greece
[email protected]

Keywords: Ontology, semantic interoperability, eBusiness, virtual organisation, ISPs.

Abstract: Enterprise interoperability is a challenging goal that has many facets, one of which being the need for
cooperating enterprises to have a precise understanding of the information that they exchange with each
other. Internet Service Providers (ISPs) collaborating in a virtual cluster to market, customize and provision
services to customers require the same understanding of these services and their features. This paper
describes the work being undertaken in the European IST project VISP (IST-FP6-027178) to meet this
requirement. ISP services are being classified and formally specified and an ontology is being developed for
the ISP service domain.

1 INTRODUCTION eBusiness. A brief summary of the work undertaken

concludes the paper.
Enterprises wishing to undertake eBusiness by
cooperating in a virtual organization must agree on
the definitions and meanings of concepts used in 2 ISPS AND VISP
their cooperation, and the software infrastructure
supporting their eBusiness transactions must be Since the advent of the Internet, the traditional ISP
aligned with this understanding so that services can market has been in constant evolution due to the
be offered and provisioned for customers gradual globalization and commoditization of ISP
dynamically, efficiently and effectively. This paper services and to deregulation initiatives aimed at
is concerned with the work being developed in the fostering competition. Small ISPs are best at
European IST project VISP (IST-FP6-027178) to targeting niche markets; they can respond rapidly
support semantic interoperability for a virtual cluster and provide excellent customer service. However,
of small ISPs collaborating to market, customize and small ISPs cannot offer the wide range of services
provision services for business customers. now required by increasingly demanding business
The structure of the paper is as follows. First the customers nor offer the geographical coverage that is
eBusiness context of small ISPs and their necessary in a globalized marketplace.
collaboration in a virtual cluster is briefly outlined. Collaborating in a virtual cluster to undertake
Reference is then made to the significance of eBusiness enables small ISPs to offer a wider range
knowledge and semantics for enterprise of services, particularly when customizing tailored
interoperability in the European research area and to services that are composed from individual services.
research projects here. Work that has been Tailoring composite services from the services of
undertaken in the VISP project on the specification other partners in the cluster in order to offer
and classification of ISP services is introduced. Then innovative and higher value solutions to customers is
the ontology that has been developed for the services indeed one of the principle advantages and reasons
in the ISP domain is presented, including an
application of its use in composing services for

390
 SEMANTIC INTEROPERABILITY FOR E-BUSINESS IN THE ISP SERVICE DOMAIN

behind the idea of collaborating in a virtual VISP The DIP IST project (http://dip.semanticweb.org)
cluster with other small SMEs. objective was to develop and extend Semantic Web
However, when composing services it is and Web service technologies, and it undertook an
necessary to have a precise understanding of the exhaustive state of the art analysis in many of the
features of each service element so that the fields relevant to VISP (DIP, 2004). It developed the
composition of individual services into a service DIP Ontology Management Suite, which is an
bundle is consistent and the composite service can integrated set of tools for efficiently and effectively
be validated. In order to meet these requirements, managing ontologies. VISP was able to input this
work was undertaken within the VISP project to work into its own state of the art analysis.
develop a service decomposition and The Semantically-Enabled Knowledge
characterization methodology and to establish an Technologies (SEKT) IST project (www.sekt-
ontology for semantic interoperability within the project.com) objective was to develop and exploit
cluster. the knowledge techniques underlying Next
Generation Knowledge Management. It also
performed a state of the art analysis on ontology
3 KNOWLEDGE MANAGEMENT evolution which was used as input to the work
undertaken by VISP (SEKT, 2004).
AND ONTOLOGY FOR Other projects undertaking work in this area
ENTERPRISE include INTEROP, where ontology was one of its
INTEROPERABILITY three key thematic components and research was
undertaken to investigate ontology in conjunction
The relevance of knowledge management and with interoperability (http://interop-vlab.eu);
semantic interoperability for cooperating enterprises FUSION, which is developing technologies for the
is a topic of current significance. One of the four semantic integration of a heterogeneous set of
Grand Challenges in the Enterprise Interoperability business applications within SMEs
Research Roadmap (Li et al., 2006) is Knowledge- (www.fusionweb.org/fusion); SUPER, which is
Oriented Collaboration, now renamed Knowledge- aiming to integrate Semantic Web Services and
Oriented Collaboration and Semantic business processes into one consolidated technology
Interoperability to emphasize the significance of (www.ip-super.org); and SYNERGY, one of the
semantics in enterprise interoperability (Charalabidis aims of which is to provide semantic ontology-based
et al., 2008). Knowledge-based collaboration allows modelling of knowledge structures on collaborative
the sharing of knowledge within virtual working (Popplewell et al., 2008).
organisations to the mutual benefit of all partners. The work carried out in these and other projects
The advances being undertaken in the area of provided generic results that could be evaluated and
knowledge management and ontology development used in VISP as a basis for developing a specific
are intended to be of direct benefit to enterprise ontology for services in the ISP domain.
collaboration. Those sharing the knowledge need
have the same understanding of it, hence the
requirement for ontology definitions in this area. 4 SPECIFYING ISP SERVICES IN
Despite a rapidly changing research area, work VISP
already undertaken was studied in the VISP project
to establish the state of the art in ontology A VISP cluster can only work efficiently if all
specification methods and ontology management partners have the same view not only of what a
techniques. Existing approaches were examined and service is and what it offers but also of relationships
their strengths and weaknesses evaluated in and constraints between services, service
conjunction with the specific requirements of the characteristics and characteristic values. In a
VISP project for the use of an ontology. federation of independent autonomous entities, there
The On-to-Knowledge IST project is no centralized control but there must be a common
(www.ontoknowledge.org) was investigated and understanding between the partners of services and
although the standards used in the project have their properties in order to collaborate dynamically
evolved, the experiences gained were found to be in real time. Properly specifying the properties of the
useful input (Davies et al., 2002). The OTS elementary service components is “a key aspect of
methodology and architecture in particular were supporting the proper specification of (composite)
used as input to the VISP ontology work.

391
ICE-B 2008 - International Conference on e-Business

services’ semantics” (Leymann, 2007). The 5 ONTOLOGY FOR THE ISP

composite service must behave correctly when DOMAIN
provisioned for the customer.
A primary objective of the VISP project is to
An ontology is being developed in the VISP project
build an ISP service knowledge base. A service has
for VISP purposes, i.e. to enable partners in the
been defined in VISP as representing an atomic set
cluster to collaborate in providing tailored services
of capabilities or activities that can be provisioned
to customers as dynamically and in as automated a
and used alone, or that can be grouped with other
manner as possible. An ontology can be regarded as
services to form a composite service. A set of
a “set of shared conceptualisations of entities within
elementary services for the virtual cluster has been
an application domain” (Wilson et al., 2006). In
investigated, classified and specified. The result is a
order to carry out eBusiness dynamically and in
knowledge base of ISP services formally represented
close cooperation with other partners in a virtual
using standard representations.
enterprise, all partners in the cluster need to adopt a
The VISP Service Knowledge Base (SKB) is a
common semantic model for their interactions, a
service catalogue that contains the technical
shared understanding that can be automated by using
information relating to the individual services that
a software infrastructure supporting the eBusiness of
can be used to offer services to customers either
the ISPs. The VISP ontology thus constitutes one of
standalone or as part of a composite service. The
the innovative results of the project.
SKB thus provides a common technical
Although problems are associated with the
understanding of services in the cluster, i.e. a
development of ontologies (Hepp, 2007), it was felt
common “language” shared by all partners that
that the requirement for automating what can be a
ensures consistency within the cluster.
complex task meant that an ontology was essential
Services described in the SKB are organized in
for a VISP cluster. Work undertaken on developing
categories and sub-categories. The nine categories
an ontology for service components is often of a
comprise access services, bandwidth services,
quite generic nature, for example (Sheth et al., 2006)
hosting services, application services, network
and (Wang and Xu, 2008). However, the ontology
services, security services, support services, and also
here is being developed for VISP cluster purposes,
software supplies and material supplies. Each of the
to support the validation of ISP services composed
nine categories defined is divided into sub-
automatically and dynamically from other services
categories. The SKB can be centralized, distributed
in an eBusiness context. It therefore has a restricted
or replicated at each partner, depending on the
purpose and a limited circle of users and it should
functional architecture adopted.
not be the bottleneck mentioned by Hepp when
A basic service modelling approach was adopted.
ontologies that are created by a small community are
A service is specified by a list of characteristics and
intended for much wider use.
each characteristic comprises a list of values. The
The main use of an ontology in VISP is in the
service, each characteristic and each value has an
area of service composition and tailoring of
associated description plus other parameters. This
innovative and often complex customized services.
approach is very simple, generic and reusable. It
When new services are composed of individual
does not depend on any particular technical or
component services, it must be ensured that the
business environment and is applicable to many
features of each service comprising the composite
fields apart from the ISP/telecom domain.
service do not conflict or result in inconsistency in
Each service is thus characterized by a general
the tailored service. The use of ontology is intended
description, a set of mandatory and optional
to assist the representation of the complex semantic
parameters with their possible values, usage
and the relationships between the services offered by
limitations and exclusions together with information
the partners in a VISP cluster. Validation of a
on how they can be composed. All services offered
service bundle requires a validation of the rules and
by partners in the cluster are described using this
restrictions among all services that form the service
specification. This information constitutes the VISP
bundle. Semantic information such as restrictions,
service knowledge base that, as the reference for ISP
relationships between services, rules for the values
services in a VISP cluster, is being used to specify
of characteristics, etc. is static. It was therefore
the VISP ontology.
decided to model the content of the SKB using an
ontology whose instance data are the service
descriptions of all VISP services.

392
 SEMANTIC INTEROPERABILITY FOR E-BUSINESS IN THE ISP SERVICE DOMAIN

An analysis was performed within the VISP recommended. Given the increasingly complex
project to compare ontology tools and languages and requirements of the VISP eBusiness environment
to decide on the approach to take given the VISP and the need for rich, consistent and reusable
requirements. Languages such as Topic Maps semantics, ontologies represent an optimal answer to
(www.isotopicmaps.org/sam/sam-model), RDF the demand for an intelligent system that operates
Schema specification (www.w3.org/TR/rdf-schema), more closely to the human conceptual level.
OWL Web Ontology Language,
(www.w3.org/TR/owl-ref/), OWL-S 5.1 Description of the SKB Ontology
(www.w3.org/Submission/OWL-S) and F-Logic
(Kifer et al., 1995) were investigated. Ontology The ontology has been developed in VISP to
editors were evaluated based on (Gomez-Perez et al., represent the semantic information included in the
2002) and work performed in SEKT (SEKT, 2004). VISP domain, i.e. ISP services. This ontology is
Repositories for storing and retrieving ontologies intended to model the SKB but would also describe
and ontology change management systems were also the classification of service descriptions, the
evaluated. attributes of service descriptions as well as the rules,
Based on the evaluation results and the constraints and relationships between them.
requirements of the VISP cluster in doing eBusiness, The service descriptions were classified as
Protégé (http://protege.stanford.edu) was selected as presented in section 4 above. The goal of the SKB
the ontology editing tool, providing not only ontology definition is to support this classification of
ontology editing capabilities but also constituting service descriptions, to include the attributes of
ontology library systems and offering various service descriptions and the relationships between
functions for managing, adapting and standardizing them as well as to demonstrate the rules and the
groups of ontologies while enabling ontology reuse. constraints for these attributes in a formal way that
No tool fulfilled all the requirements for a reliable, can be used to validate a composite service, or
secure, interoperable ontology management service set as it is termed in VISP.
environment although most seemed to provide
adequate solutions for most of the main issues in the
VISP eBusiness environment. Protégé was one of
the first editors available, it is open source and
supports a variety of plugins and import formats,
such as RDF Schema and OWL. It is also widely
used; 68.2% of respondents in a survey were using
Protégé as their ontology editor (Cardoso, 2007).
The principle requirement for an ontology
specification language in VISP is expressive power
and OWL was selected here. It supports the RDF
data model, is used in the Semantic Web world and
was felt to be the most appropriate choice for the
eBusiness domain to be modelled in VISP. It too is
widely used; 75.9% of respondents in the survey
mentioned above are using OWL to develop their
ontologies (Cardoso, 2007).
A two-phase methodology was developed for
designing, developing and using the ontology-based
knowledge management system for the VISP
software environment. In the first phase, the Figure 1: Class hierarchy for the SKB ontology in Protégé.
application area and use of the ontology were
determined. In the second phase, the ontology was The Protégé OWL editor provides three views
built based on the information sources identified in for the definition of an ontology, namely description
the first phase and the concepts used there. Although of classes (see Figure 1), description of properties
the area of tools and technologies enabling the (see Figure 2) and description of individuals. For the
application of ontologies on a business level is not description of classes the SKB ontology is primarily
considered mature and much work is still in based on the schema of the SKB repository.
progress, the use of an ontology in VISP was However, the schema includes several details that

393
ICE-B 2008 - International Conference on e-Business

are not required in the definition of an ontology, the definition of these restrictions is probably the most
primary purpose of which is to enable the validation important part of the SKB ontology in the sense that
of a service set from within the VISP software it differentiates it from a flat database schema such
infrastructure. Such information includes mainly as the SKB itself. Such rules and constraints can be
fields that are used to describe static attributes inside defined in OWL provided that it is possible to
the SKB and which are of no special interest inside express them as logical expressions.
the SKB ontology, for example, Description, Therefore, the definition of the SKB ontology is
GenericDeploymentInformation, References. a two-step process:
The second view is the description of properties. 1. First, the “skeleton” ontology model has to be
OWL properties model all relationships between specified. The class hierarchy, the listing of
individuals of different classes. Apart from this pure properties and the definition of individuals form
listing of the ontology properties, this model also the necessary schema elements of this model and
includes a description of the restrictions that each will be utilized to describe the domain
property of each class should satisfy. This aspect is restrictions in the next step.
further discussed in the following section. 2. Then the rules and constraints governing the ISP
services domain have to be transformed into
logical expressions that can then be expressed
using the expressive syntax of OWL.
Although the work on the classification of ISP
services and their specification using the service
description template (as described in section 4
above) has been almost completed and fully
documented, the work on the documentation of the
exact rules that each service will have to comply
with has not started yet. Since it is expected that this
task will involve a great amount of work which will
be continued even after the end of the project, a
workaround has been used to ensure that OWL is
adequate to express all possible rules and restrictions
in the ISP world. A demo service was described and
the following possible types of restrictions that the
modeller could be asked to represent were identified:
1. The set of selectable enumerated values of a
characteristic depends on the value of another
characteristic.
2. The existence of a characteristic depends on the
Figure 2: List of OWL properties in the SKB ontology. value of another characteristic.
3. The range of values of a characteristic depends
The third view is the description of individuals. on the values of another characteristic (very
Individuals are instances of classes such as in object- similar to case 1 but more complex to represent
oriented programming languages. Thus the because deals with ranges).
descriptions of classes mentioned above also provide 4. Characteristics are mutually exclusive, but at
general descriptions for the individuals of these least one must be present although they are
classes. indicated as optional.
The first type of restriction essentially constitutes
5.2 Restrictions to be Represented an “if” statement: if the value of characteristic A is
greater than 5, characteristic B can only take values
The use of an ontology cannot only be justified by 1, 2 and 3. The representation of an “if” statement
the need to model the domain of interest in a fairly can be easily achieved with the help of the logical
simple way by using OWL classes and properties operators AND, OR and NOT. If X then Y is
that resemble capabilities of UML diagrams and equivalent to NOT X OR Y. Of course, OWL allows
objected-oriented design. The need for an ontology the definition of such a logical restriction.
has evolved due to the need to define complicated As the same analysis has been performed for all
rules and constraints governing the domain of the above types of restrictions, it is clear that OWL
interest (in this case the ISP service domain). The and the definition of the SKB ontology model satisfy

394
 SEMANTIC INTEROPERABILITY FOR E-BUSINESS IN THE ISP SERVICE DOMAIN

the requirements of the ISP domain as well as of the one or more services from VISP, a sales
VISP use case called “validation of a service set”. representative creates a service set that comprises
A technical restriction imposed by the current various services offered by partners in the cluster.
OWL specification is the lack of support for numeric After a service set has been created, it needs to be
ranges. While OWL has cardinality restrictions, validated to ensure it can be instantiated. The sales
datatype values cannot be further restricted. One of representative uses the VISP GUI to click on the
the solutions evaluated was to extend OWL files “validate” button to validate this service set.
with URI references into an (external) XML Schema The use of the SKB ontology allows for a more
file. This means that a separate XML Schema file sophisticated validation strategy than the rather
together with the OWL file will have to be simple and incomplete validation undertaken in the
maintained. Another solution is to define an first two releases of the VISP infrastructure
extension ontology to be imported by the SKB software. The ontology is being used to enhance the
ontology, which defines RDF properties that can be first implementation, e.g. checking intra-service
used to represent XML Schema facets. Other constraints between characteristics and values,
solutions are also being evaluated and hopefully the checking inter-service constraints between services,
next OWL specification will be enriched with the characteristics and values, etc.
capability of expressing restricted datatypes. Compared with the current implementation, the
additional necessary architectural element will be a
5.3 Use of the SKB Ontology framework enabling the manipulation and the usage
of the ontology. Such a framework can be Jena
The use of the ontology comprises two concepts: (http://jena.sourceforge.net/index.html). Jena is a
how to manipulate the ontology and how to exploit it Java framework for building applications related to
to enable the validation of a service set. the Semantic Web in general. It includes:
When manipulating the SKP ontology, Protégé ƒ RDF API
allows for a series of actions regarding a definition ƒ OWL API
of an ontology, such as to: ƒ Rule-based inference engine
ƒ create the ontology definition ƒ In-memory and persistent storage
ƒ browse the ontology ƒ SPARQL query engine (www.w3.org/TR/rdf-
ƒ edit/update the ontology definition sparql-query/)
ƒ query the ontology Its most useful feature for VISP is that it allows an
ƒ store the ontology persistently ontology or a data set to be represented with an
ƒ apply various ontology checks object model, as well as incorporating reasoning
ƒ compute the inferred taxonomy capabilities.
The last two actions are performed with the A main prerequisite before the validation can
assistance of an OWL reasoner, which can be start is the existence of a consistent SKB ontology.
combined with Protégé if it supports the DIG A service description is added to the SKB ontology
interface. An OWL reasoner provides the following when it is approved in the VISP cluster with the
standard inference services: appropriate restrictions. The next steps are then
ƒ Consistency checking: determines whether the followed inside the implementation code of the
ontology contains any contradictory facts. validation of a service set:
ƒ Concept satisfiability: determines whether it is 1. The system reads the service set that is to be
possible for all classes to have any instances. validated either through the repository
ƒ Classification: computes the inferred class containing information about the service sets and
hierarchy. their component services or through an XML
In general, Protégé supports all features expected file.
of an OWL editor. The SKB ontology will be 2. It creates a model representing the data set of
manipulated through the use of Protégé, at least until this service set.
the end of release 3 of the VISP software 3. It loads the ontology model of the SKB
infrastructure. In the future it may be desirable to be ontology. This can be read and created each time
from the respective OWL file or it can be stored
able to perform certain actions on the ontology from
persistently inside a database.
the VISP platform itself.
4. It validates the data model against the existing
The primary reason for the definition of the SKB ontology model, i.e. it performs a global check
ontology was to enable the implementation of the across the schema and instance data looking for
validation of a service set. When a customer requests inconsistencies.

395
ICE-B 2008 - International Conference on e-Business

5. The result will report whether the validation REFERENCES

check has passed and detail any detected
inconsistencies. Alani, H., Hall, W., O’Hara, K, Shadbolt, N., Szomszor,
This approach is quite simple and can be seamlessly M, Chandler, P., 2008. “Building a Pragmatic
integrated with the current implementation of the Semantic Web”, IEEE Intelligent Systems, 23(3),
service set validation since Jena is a Java framework pp.61-68.
that can used by the existing classes. Cardoso, J., 2007. “The Semantic Web Vision: Where Are
We?”, IEEE Intelligent Systems, 22(5), pp.84-88.
Charalabidis, Y., Gionis, G., Hermann, K.M., Martinez,
C., 2008. Enterprise Interoperability Research
6 CONCLUSIONS Roadmap, Update (Version 5.0), European
Commission.
The VISP project has developed both a methodology Davies, J., Fensel, D., van Harmelen, F., 2002. Towards
for categorizing and describing ISP services and an the Semantic Web: Ontology-driven Knowledge
Management, Wiley, Chichester.
ontology for the ISP service domain. The need to
DIP, 2004. D2.1 Report on Requirements Analysis and
add semantic content to ISP service descriptions is State of the Art (WP2 – Ontology Management), DIP
fundamental for ISPs carrying out eBusiness for project IST-FP6-507483, Deliverable D2.1.
innovative, complex and composite ISP services. Gomez-Perez, A., Angele, J., Fernandez-Lopez, M.,
The VISP ontology has been produced for a Christophides, V., Stutt, A., Sure, Y., 2002. A survey
specific context and is being applied in this context. on ontology tools, OntoWeb deliverable 1.3,
This is indeed a recommended approach when the Universidad Politecnia de Madrid.
aim is a lightweight ontology that can be cost Hepp, M., 2007. “Possible Ontologies. How Reality
effective to design, build and maintain (Alani et al., Constrains the Development of Relevant Ontologies”,
IEEE Intelligent Systems, 22(1), pp.90-96.
2008). Ontologies do not need to be large and
Kifer, M., Lausen, G., Wu, J., 1995. “Logical Foundations
complex and the intention of the VISP project was of Object-Oriented and Frame-Based Languages”,
innovative in that it has designed an ontology that Journal of the ACM, 42(4), pp.741-843.
could be part of the VISP platform in a running Leymann, F., 2007. “Semantics for the Cloud”, IEEE
environment. Decisions taken in developing the Intelligent Systems, 22(5), pp.15-17.
ontology tended therefore to be pragmatic with the Li, M-S., Cabral, R., Doumeingts, G., Popplewell, K.,
context, stakeholders and potential users in mind. 2006. Enterprise Interoperability Research Roadmap,
The intention is to further develop the results of Final Version (Version 4.0), European Commission.
the VISP project for commercial use in a virtual Popplewell, K., Stojanovic, N., Abecker, A., Apostolou,
D., Mentzas, G., Harding, J., 2008. “Supporting
cluster of small ISPs. The ontology is an essential
Adaptive Enterprise Collaboration though Semantic
part of this aim and has been designed for such use. Knowledge Services”. In Enterprise Interoperability
The work depicted here has built on existing III. New Challenges and Industrial Approaches, ed. K.
research work in the area and has applied it to a Mertins, R. Ruggaber, K. Popplewell, X. Xu,
specific context. This has required a lightweight Springer-Verlag, London, pp.381-393.
approach avoiding any undue complexity and the SEKT, 2004. D3.1.1 Ontology Management and Evolution
development of a small and highly focussed – Survey, Methods and Prototypes, SEKT project IST-
ontology that fits in with the specific requirements 2003-506826, Deliverable D3.1.1.
of the VISP cluster and its partners. First results Sheth, A., Verma, K., Gomadom, K., 2006. “Semantics to
energize the full services spectrum”, Communications
have shown that the ontology has met these
of the ACM, 49(7), pp.55-61.
requirements and that ISP composite services can be Wang, Z., Xu, X., 2008. “Ontology-based Service
automatically validated with the ontology during Component Model for Interoperability of Service
service provisioning. Systems”. In Enterprise Interoperability III. New
Challenges and Industrial Approaches, ed. K. Mertins,
R. Ruggaber, K. Popplewell, X. Xu, Springer-Verlag,
London, pp.367-380.
ACKNOWLEDGEMENTS Wilson, M., Li, M-S., Jaekel, F-W., Ruggaber, R., Chen,
D., Mehandjiev, N., Zelm, M, 2006. Enterprise
This work was carried out with partial funding from Interoperability Roadmap. Annex I – Indicative
the EU through the IST project VISP (IST-FP6- Research Challenges, Final Version (Version 4.0),
027178). Further information on VISP can be European Commission.
obtained from www.visp-project.org.

396
 ON-DEMAND MOBILE CRM APPLICATIONS FOR SOCIAL
MARKETING
Business and Technology Perspective

Viktor Kaufman
SAP Research, Karlsruhe, Germany
[email protected]

Yuri Natchetoi
SAP Research, Montreal, Canada
[email protected]

Vasily Ponomarev
NPP Rumb, Moscow, Russia
[email protected]

Keywords: CRM, mobile, social marketing, collaboration, customer loyalty, mobile application, application design.

Abstract: Solutions for Customer Relationship Management (CRM) tend to evolve from the traditional passive
recording of transactions between the Company and the Customer to agile optimized strategies for
interaction and cooperation with the Customer. Despite considerable industry and research interest to the
new ”social” CRM 2.0, the gap between the vision and the reality is still quite large. Use of handheld
devices deserves special attention in this respect. Although it can boost sales, customer satisfaction and
customer loyalty, the amount of research work and the number of mobile applications in this area is low. In
this paper, we focus on design and implementation of mobile social CRM solutions, which make extensive
use of collaboration technologies. By means of simple business use-case scenarios, we argue that good
understanding of business objectives, business processes, and technology issues, together, is crucial for
development of compelling social CRM applications.

1 MOBILE SOCIAL CRM As the efficiency of traditional mass-marketing,

such as TV, radio and direct mail, constantly
Nearly every business needs some kind of CRM decreases, companies face the need to review their
solution, even if it is not a specialized CRM CRM strategy. One promising cost-effective and
software system. Banks, Insurance Companies, e- otherwise strategically interesting approach is called
Businesses, non-profit and Public Sector Social CRM or CRM 2.0 (Band, 2008). Social CRM
organizations - all focus on serving their customers. recognizes the fact that mass advertising is not
A large CRM component is usually integral part appealing enough to customers and advocates more
of enterprise business-software landscapes. SAP, attentive and interactive approaches to foster
Oracle, Salesforce.com, and others are the customer relations. Due to the rise of the Internet
established proprietary software vendors. Smaller and increased level of public knowledge, customers
companies also process information that is typical will rather actively search for information they need.
for CRM, such as leads, contacts, accounts, They will leverage their social networks and trust to
promotions, sales orders. In any case basic data their friends and other customers like them. When
management, activities tracking, and reporting making a choice to buy a product or service, they
functionality implies high maintenance costs. will ask opinions of members of their social network
Especially costly is customer-acquisition support. (Webber, 2007). Social CRM extends the
collaborative CRM and focuses more on support of

397
ICE-B 2008 - International Conference on e-Business

co-operation and customer-involvement aspects of and medium-sized businesses (Darrow, 2008). The
customer relations. influencers can get incentives for registering and
With regard to CRM objectives, mobility brings closing leads in the company’s CRM system.
a number of advantages. Due to the ubiquitous Recently (Oracle, 2008), Oracle announced
nature of the mobile communication (Virki, 2007), it Mobile Sales Assistant for company’s CRM on
is especially suitable for social marketing. Most Demand solution that”changes the face of mobile
often, consumers and sales agents wish to make CRM”. It features user collaboration with colleagues
decisions when they are away from their computers and customers, push-based architecture for
and have limited access to information directories. Blackberry®, one-click-away account information
Leveraging broad availability of mobile devices can and customer contact information, and more for $30
facilitate ease of relevant information access, per user per month.
advanced customer service, and field workers During the iPhone™ Software Development Kit
support (Sudan et al., 2007). In Section 4 we discuss announcement, March 6th 2008, Salesforce.com
how this eventually leads to CRM-effectiveness demonstrated its ability to bring its innovative CRM
gains. on-demand services onto the iPhone platform and to
Mobile social marketing is a new concept, but its provide new level user experience. Social mobile
role rapidly becomes perceived as significant for CRM applications are to be expected soon.
CRM promotion and delivery strategies (Webber, Microsoft has only recently entered into the
2007). We consider mobile social marketing a “must CRM market with Microsoft Dynamics CRM 3.0.
have” element of the future integrated CRM Their Mobile CRM solution works only on the
solutions. Possible concrete applications include Windows Mobile Pocket PC devices. For now, it
managing social links on the mobile phone, provides salespeople with up-to-date information
receiving recommendations, consulting community about their accounts and contacts, and helps them
opinions about products, tagging, rating, promoting, manage sales opportunities and track sales-related
and ranking goods and services on the move. activities.
Retailers could get additional means to gather Arvato Mobile offers a set of advanced building
information about consumer preferences and blocks for mobile CRM. They provide for engaging
forecast consumer needs. Advanced applications can user experience including games, send-a-friend,
take care of mobile access to a marketplace, where sponsored pop ups, communication via SMS, MMS
retailers and consumers search for products and and more. The tool set both aims at “customer-club”
services with the best quality-price ratio, place their for customers and supports statistical data
orders directly in the back-end CRM system etc. The aggregation for CRM.
vast popularity of content-generation modes in the Kintera Inc. recently launched Kintera Sphere™
context of social networking web sites (Vaske, v8.0, company’s new social CRM system for non-
2008) suggests support for mobile exchange of profit organizations, providing a total view of the
subjective information related to products and constituent's relationship with the organization
services. Software vendors already support some of (CRM Today, 2008). The company claims to have
these ideas, although the major business-acceptance considered factors such as enhanced trust, sense of
breakthrough requires more time. belonging, instant gratification, emotional release,
and sense of social impact. These features are
1.1 Market Overview definitely interesting for mobile applications too.
In the light of described mobile social CRM
Jim Balsillie, co-CEO of RIM, Canadian wireless developments, the question remains – when and how
device company, says (Lomas, 2008) that the need will new solutions actually enable sales
to seamlessly integrate Web services and desktop professionals to better understand and address their
applications onto handsets “is not a concept. This is customers' needs, anytime and anywhere. As
a reality”. He further mentions: "Once social companies move from products to solutions, the
networking becomes a B2B phenomenon – not technology imperative is to enhance the business
unlike IM and texting – I believe every single social- acumen and insight of the front line, which has not
networking user will want a data plan". been the traditional goal of CRM. The business
SAP and some other major vendors of CRM condition precedents to business-acceptance of
solutions closely follow the new developments. innovative CRM solutions have to be understood in
Recently, social-networking B2B approach has been the first place see Sections 2 and 4. The challenges
chosen to drive adoption of SAP software by small for "technology enablement" leading to effective

398
 ON-DEMAND MOBILE CRM APPLICATIONS FOR SOCIAL MARKETING - Business and Technology Perspective

selling supported by CRM will be presented in interactions. The first approach needs dedicated
Section 3. Our view is based on own applied control mechanisms to canalize and manage
research in the mobile CRM area described in customer relations. The second approach contradicts
Section 4 and summarized in Section 5. with the idea of lightweight intuitive support for
customer and partners interactions. Due to the
formalization requirement, the processes can easily
2 YET ANOTHER CRM WITH become incomprehensible or inflexible. We have
chosen a mixed approach, which avoids undue
MOBILITY SUPPORT? quantification and specification of desired
interactions, but puts strong emphasis on thought-
Mobile access to back-end CRM systems is not new out, informal, early and iteratively adaptable
(Sadeh, 2002). In what respect are mobile social specification of the desired mobile social CRM
applications different? In our opinion, the answer solution.
can be found between the lines of the previous We recognize that most important aspects of the
Section – it is the challenge to better meet business social CRM are: 1) efficient cooperation with
objectives by means of supporting and influencing customers and partners; 2) efficient collaboration
complex human interactions. The idea is that between CRM and non-CRM employees within the
customer and partners feedback, or ability to company; 3) supporting and enlarging existing
collaborate, and not so much stiff internal processes, relationships among customers and 4) shifting focus
drive customer relations, marketing and sales. from the sales volume to better customer experience,
Dynamically changing customer relations and the compare also (Paterson, 2005). In fact, most mobile
need to often review company’s self-perception in CRM solutions fail to address some or all of these
the market add to the overall complexity. issues. In our opinion, this is because of the
There is no one good solution, but poorly mentioned complexity and practical difficulty of
serviceable solutions are easily made. As the involving business analysts, technical experts,
businesses remain sceptical, solution designers look scientists, field workers, and customers early enough
for ways to control the complexity. In a nutshell, two in the solution design.
approaches are under consideration by researchers We believe that transparent and flexible design
and practitioners. The first becomes increasingly principles constitute the best way to gradually
accepted and advocates use of general-purpose accumulate expertise of the many people
collaboration-support tools like Email, push-alerts, contributing to a compelling CRM solution. We
Instant Messaging, chat, mashups etc. alongside the follow an approach, where the following three steps
company’s CRM processes. While there is nothing are being largely addressed in parallel: 1) define and
wrong with this approach, care is needed to decide understand business objectives (market research,
whether it is appropriate. The costs of putting a defining specific CRM measures, customer group
solution into operation and maintenance costs can focus etc. can be part of this step, but general
easily exceed the expected benefits (Galdy, 2008). understanding of implications of the business
The second approach is to adapt and extend objectives is more important); 2) choose business
existing business processes. In (Band, 2007), the processes (in the first place, we identify use-case
authors suggest the following four steps to build scenarios to focus on); 3) address technology issues
compelling CRM applications: 1) define and This step is indispensable especially for mobile
quantify business goals; 2) formulate CRM solutions. The architectural framework described in
strategies and tactics; 3) establish appropriate CRM the next Section facilitates reuse of components, but
measures; 4) link CRM goals, strategies, and some redesign and customisation are still necessary,
metrics. Forrester advocates another four-step depending on the desired business processes. Next,
strategy called POST (Bernoff, 2007): P) review the we proceed with step 3) and then exemplify steps 1)
Social Technographics Profile (decide what is and 2).
possible, customers are divided into groups like
Creators and Inactives); O) pick an objective; S)
choose a strategy; T) select and deploy appropriate
technologies and measure results.
We argue that both approaches have advantages
for mobile social CRM, but undervalue the
intrinsically unstructured nature of customer

399
ICE-B 2008 - International Conference on e-Business

3 TECHNOLOGY PERSPECTIVE: To lighten the application, only subsets of the

APPLICATION DESIGN mentioned standards are used. For example, a very
limited version of the full BPEL interpreter is
implemented on the mobile device, which enables
When designing mobile extensions to the Enterprise
basic composition of the workflow scenarios using
CRM applications, one must be careful to
local and remote services (Hirsh et al., 2006).
distinguish traditional software applications and
The components overview of our architecture is
specific functionality in a mobile-enterprise software
presented in Figure 1. In our Framework, the CRM
context. It is impossible to deploy the entire
business-data objects are being serialized,
business-software system onto a mobile device. The
compressed and transmitted to the client side in the
challenge is to transmit only the relevant business
form of a compressed RDF messages. The
information and implement software functionality
information is stored in the local Persistent Data
required for selected processes.
Store, also in the compressed RDF format, making it
possible to store a significantly larger number of
business objects as compared to a traditional file
system or relational database. The Framework
enables Web service calls and pro-active download
of data required later on in the asynchronous
business process. The client application uses the
locally stored data to support off-line work.
Efficient connection to and data synchronisation
with the back-end Enterprise system is very
important for field workers. We use asynchronous,
message-based communications for this, as they are
a better fit in the mobile environment. The Smart
Figure 1: Architecture overview. Asynchronous Connection Manager (SACM) is a
unified manager of asynchronous communications
The known limitations of mobile devices are: between and with mobile clients. The standard
connectivity, processing power, usability, security communication protocols supported by now are
concerns, and memory requirements, but mobile TCP/IP and SMS. We designed suitable API to
devices also have some advantages, such as multi- asynchronously send messages from the client to the
modal input and high level of availability. All these server, from the server to the client, and between the
considerations have to be taken into account. For clients. The SACM does not explicitly distinguish
social mobile CRM applications, there are additional clients and servers so that it also suitable for mobile
requirements due to on-demand support of often Peer to Peer scenarios.
changing customer interaction patterns.
With our business application framework
(Natchetoi et al., 2008), we have implemented
Mobile CRM client as a Java midlet and native
Objective-C application for iPhone. In our solution,
neither the business logic nor the user interface
forms are hard-coded in the client application.
Instead, the client application partially implements
interpreters of open industry standards like SOAP,
RDF, OWL and XForms. The application logic and
user interface can be easily modified or augmented
at low cost, since we are using standard formats and
collaboration concepts such as mashups from the Figure 2: Sample screenshots.
Web 2.0 tool set, see (Natchetoi et al., 2007) for
details on our mobile business-oriented browser. For Our Mobile CRM prototype based on the
an example, user experience can be enhanced by introduced Framework enables sales agents to
adding XForms accessing back-end Business establish interactions using comprehensive persistent
Objects exposed through SOA Web Services. knowledge about the constituents and provides a
centralized way to persist and manage customer

400
 ON-DEMAND MOBILE CRM APPLICATIONS FOR SOCIAL MARKETING - Business and Technology Perspective

relationships information. Mobile CRM workers can enough, though, in the world of ever more rapidly
update business information, browse reports and changing requirements and increasing customer
invoke Web Services remotely. They can access the expectations. As described in Section 2, we have to
CRM application functionality both online and focus on specific business objectives, using selected
offline. The mobile CRM has been developed for business processes.
iPhone (Finkle, 2007) and for J2ME-enabled phones The corresponding decisions include
using the same Framework. understanding of company's actual or desired core
business strengths and deriving strategic objectives
for interaction with customers. The objectives have
to be detailed by taking into account customer
profiles, relevant market researches, anticipated
trends etc. They can be focused on a particular
customer segment, technology, measurable
performance indicators etc. In any case, the
decisions should lead to a carefully selected set of
desired customer-interaction-support scenarios,
which have to be continuously validated against the
business objectives and provide for differentiation
from the competitors.
Figure 3: Sample screenshots of the mobile CRM To test and further develop our mobile CRM
application. application and our mobile Framework, described in
Section 2, we were looking for some innovative
The Framework enables simplified adaptation or scenarios, which we could implement in the near
extension of existing applications. Beyond the basic future. It became clear that we needed to define a
Contacts, Leads, Opportunities, and Sales Orders compelling set of objectives first. Let us consider an
management, we implemented some collaboration example.
Services supporting multi-channel communication Imagine a Service provider company. It has to
of sales agents, in a way inspired by ECOSPACE face harsh competition. To survive the next drop in
project (http://www.ip-ecospace.org). For now, we prices, the company decides to save on traditional
support Email and Fax access, and Instant marketing and to leverage the knowledge of loyal
Messaging integrated into CRM-processes context. customers about the company's quality services. The
In the next Section we describe further functionality management wants customer retention to be
relevant to mobile social CRM, which we could addressed more; hopes to be able to provide added
support in the future. value especially for new customers; and strives to
differentiate the company from the competition. To
achieve their plans, the management considers the
4 BUSINESS PERSPECTIVE: following use-case scenarios.
ENABLEMENT OF SALES 4.1 Set of Scenarios 1
PEOPLE AND CUSTOMERS
The company needs ways to communicate to its
Social CRM is essentially about interaction with loyal customers. Those customers are presumably
customers. The sales people and the customers are rather busy and sceptical about advertisements.
the prominent interacting parties. It is therefore Therefore, for most scenarios described below,
natural to support them in the first place. This simple traditional communication channels through mail,
assumption already sets priorities for the email, and telephone would be considered annoying
functionality and the needed technology. We and would be too inefficient. The company decides
herewith assume that interactions drive the to go for frank short conversations with rather
development and customization of CRM obvious added value for the customer, and sees
applications. handheld devices to be best suited for this.
There are still different kinds of customers, One option would be to cooperate with a mobile
sales people and interactions. Traditional CRM carrier and provide mobile services, including the
solutions would establish order by suggesting "best ability to send SMS or other kinds of messages to
practice" processes out of the box. This is not the customers. Another option would be to provide

401
ICE-B 2008 - International Conference on e-Business

mobile applications for download. Also it is possible 4.4 Set of Scenarios 4

to cooperate with a handheld devices vendor. The
last two options are much more flexible, compare Selected “premium” users can earn gifts, reputation,
below, and the corresponding applications would or even money by helping out in the scope of their
make use of the technical architecture described in experience. The company provides them with a
the previous Section. mobile tool, which effectively counts them to the
product support team. They receive and answer
4.2 Set of Scenarios 2 questions, concerns and ideas of other users, answer
and forward them. They can exclusively send their
The company is interested in most active, own requests to the company's support team
experienced, or otherwise influential persons among members. The application architecture described
the customers. Determining such customers is to before would even allow certain direct changes in
some extent similar to representing the customer the back-end systems on their behalf. Multimodal
base as a “social network”. The company can make input would facilitate simple upload of a picture,
an educated guess that loyal customers or customers video, or voice explanation to the designated online
with high amount of orders are potentially valuable space, possibly, after review of company's
collaborators. To facilitate their retention and to test responsible employees. The interested other
the guess, it offers the selected customers to forward customers can be notified directly of this event.
a small gift to a person of their choice (but not Next, the chat functionality facilitates
themselves). The gift can be limited Internet domain communication. To achieve even more customer
contract, free SMS pack, credit points for some loyalty, additional services of social value like
services, etc. The choice of gifts can be made more exchange of contact details or introduction of
intelligent later, as the “social network” is growing. customers with similar profile to one another can be
The customers have the choice not to store the offered.
delivery address at the company, since the mere act
of forwarding is enough for the company's purposes. 4.5 Set of Scenarios 5
As soon as the influential customers are known,
one can offer them certain incentives and ask to Sales agents use back-end services directly on their
recommend the services. The company offers to mobile phones. They also benefit from the emerging
conduct a counselling interview. The influential social space. They can tap the experience of
customer suggests time and enters contact details of “premium” customers or ask them for references.
her friend. The company can reassure the influencer The sales agents can exchange thoughts and business
of its quality counselling, including comparisons content through chat and other functionality, which
with similar services offered by competitors. is deployed on their mobile phones. As a side effect,
the company can perpetuate discipline in using back-
4.3 Set of Scenarios 3 end CRM system and, through collaborative style of
work it can uphold the sense of mission for sales
The company might need some user feedback on agents. Similarly to customer feedback, sales agents
provided services. One idea would be to offer a can communicate their concerns directly to the sales-
downloadable game, which becomes activated after support team.
the customer has completed a short satisfaction
survey. Playing the game, the customer may have 4.6 Advanced Scenarios
the option to share her view on possible
improvements for company's services. For specific To strengthen its market position by means of key
services, it may be possible to review other users' differentiators, the company plans to provide special
feedback, and even to discuss them in a kind of kinds of services, based on its customer-base
forum, adapted for mobile use. Considering the information. The company's customer profiles
technical reality of mobile users is facilitated by the include information on their mutual relationships
appropriate technical design. and joint activities. It is now possible to derive or
Especially new customers would benefit from extract information, which can be used for strategic
sharing their problems-SMS in the forum. This decisions. The customers can be divided into
service can be supplemented by an Internet based segments; the adoption level of selected scenarios
online community. can be measured etc. Some of the corresponding
analysis results should be available on the mobile

402
 ON-DEMAND MOBILE CRM APPLICATIONS FOR SOCIAL MARKETING - Business and Technology Perspective

phones for sales agents. Our framework described find job, plan entertainment and receive consulting
before allows to rather easily extending applications support from the social aid experts. The portal is a
to include reporting functionality. complex proprietary CRM system, treating citizens
Given the metadata about joined CRM activities, as clients, served by the municipal Government. It
mathematical analysis of gathered data can be includes multiple elements of collaboration, smart
performed. For example, business classifications can content classification and auto-actualization. The
be tested and improved by means of probabilistic content addressed to the different categories of the
methods like dynamic Bayesian networks (DBN) users passes through the actualization engine that is
and dynamic conditional random fields (CRF) based on linguistics structure-based classification
(Lafferty et al., 2001). Promotion scenarios and approach to direct content to the appropriate group
playing with what-if scenarios becomes possible. of the users. (Ponomarev 2004) The feedback,
The “social network” data can be enriched with collected from the users is processed in order to
data shared on the social sites like MySpace or improve auto-actualization procedure in iterative
Facebook. Then, also cross-selling opportunities and way.
more broad marketing can be pursued. Mathematical The results of three-year long experiments of
analysis can identify potential new customers. using Social approach for this portal have been
collected in order to evaluate the efficiency of the
4.7 Implications of Scenarios Use taken approach. However, the efficiency of such
solution is difficult to express in numbers, because
The general message of this Section is to focus on the goal of such Social portal is not the sales
customer and sales-agent needs as much as possible, volumes but rather the trust and satisfaction of the
and at the same time to achieve the company's users. The results of the customer satisfaction
objectives. Through more focused approach, surveys, taken by the users during the period of two
flooding customers with irrelevant advertisements years indicate a strong positive shift in the “brand”
becomes superfluous. The described scenarios show loyalty and customer satisfaction. (Ponomarev 2005)
how customer retention, market penetration, However, we are still looking into the different
decision support, and powerful reporting can be approaches to quantify user’s satisfaction and trust.
achieved. Establishing relevant social processes We are also looking forward to implement mobile
would push the limitations of the traditional CRM access to the Social portal for youth and make this
applications far into the area of more educated feature available to all users.
Social CRM, with Mobility providing added value.

4.8 Experimental Evaluation 5 SUMMARY

We are in the evaluation phase of the approach, Mobile social CRM is a new concept that promises
presented in the paper. It is difficult to evaluate such more focused, rewarding, engaging and powerful
characteristics as user’s trust quantitatively. We have way to build long-lasting customer relationships.
tested our Mobile CRM client prototype with users Use of Web 2.0 collaboration technologies can
from 11 different industries. Most of them have provide major benefits. These benefits come at a
ranked the Social Marketing and Collaboration cost of more responsible application design. The
elements of the prototype as “somehow useful” and CRM applications have to become agile with regards
“very useful”. to the needed functionality. Mobility support is
In order to verify our ideas on Social Web-based especially technically demanding.
Marketing approach we have also implemented a Furthermore, regardless of the available technical
Web portal targeting the young people, students and expertise, the usefulness of applications has to be
young families living in Moscow. The portal, funded derived from thought-out business requirements. A
by the Moscow Government with highly dynamic formalisation of suitable business requirements for
content provides various information services to the less structured collaboration and communication
citizens. The advantage of taking Social approach processes is difficult by definition. To circumvent
for building this portal is obvious because it is this issue, we suggest following loosely defined
targeting different categories of users, living in the steps suitable for unstructured nature of pervasive
same city and often linked to each other. interactions in social CRM. In short, these steps
The portal informs citizens about municipal define business objectives, business processes and
programs as well as helps them to get education, enabling technology.

403
ICE-B 2008 - International Conference on e-Business

Grounded decisions on the overall business [Online]. PR Newswire Association LLC. Available:
objectives, in the first place lead to requirements for http://www.prnewswire.com/cgi-
a draft technical architecture with flexibility for the bin/stories.pl?ACCT=104&STORY=/www/story/03-
unstructured organic growth processes in mind. One 11-2008/0004771502.
Darrow, B., 2008. SAP wants to influence the influencers.
has to consider expertise from relevant vertical [Online]. SearchITChannel.com. Available:
divisions of the company, identify use-case http://searchitchannel.techtarget.com/news/article/0,28
scenarios in as many details as possible, consult 9142,sid96_gci1308041,00.html?track=sy191&asrc=R
potential users, and particularly take care of SS_RSS-21_191 .
community support, make decisions on the scope of CRM Today, 2008. Kintera Launches Revolutionary
the scenarios to support, adjust the technical design Social CRM System, Uniting Online and Offline
to ensure adaptability with respect to the scenarios Constituent Experiences. [Online]. CRM Today.
and provide the software lifecycle support. Many Available:http://www.crm2day.com/news/crm/118016
.php.
steps can be taken in parallel, but they all require
Sadeh, N., 2002. M-Commerce: Technologies, Services,
continuous exchange between the stakeholders, and and Business Models, Wiley & Sons.
are best elaborated gradually. Transparent simple Galdy, A., 2008. Mobile-Messaging – hoher Aufwand und
design principles constitute the best way to gradually sonst nichts. [Online]. IDG Business Verlag GmbH.
accumulate expertise of the many people Available:http://www.cio.de/knowledgecenter/mobile_
contributing to a compelling CRM solution. it/847929/index.html .
In this paper, we exemplified the mentioned Band, W., 2007. The Right CRM Metrics for your
points. We considered scenarios to support Organization, Forrester Research, Inc.
proliferation of business for a Service provider. Due Bernoff, G., 2007. Objectives: The Key To Creating A
Social Strategy, Introducing the POST Method:
to flexible technical design, our prototype mobile
People, Objectives, Strategy, and Technology,
CRM application can be extended to support many Forrester Research, Inc.
of the business scenarios at a very low cost. We Paterson, J., 2005. 10 Critical factors When Implementing
outlined the important technical decisions. CRM. [Online]. Really Simple Systems, white paper.
We also discussed provision of advanced back- Available:http://www.reallysimplesystems.com/white
end functionality, such as analysing community paper.asp .
structure and producing powerful reports. Such Natchetoi, Y., Kaufman, V., Shapiro, A., 2008. Service-
advanced functionalities can serve as a key Oriented Architecture for Mobile Applications. To be
differentiator from the competition. published in ICSE companion workshop proceedings.
Natchetoi, Y., Kaufman, V., Hamdi, L., Shapiro, A., 2007.
Mobile Web 2.0 Browser for Collaborative Social
Networking. [Online]. ECSCW 2007 Workshop.
REFERENCES Available: http://cscwlab1.informatik.unibw-
muenchen.de/Main/Ecscw2007Ws .
Band, W., 2008. The CRM 2.0 Imperative: Look To New Hirsh, F., Kemp, J., Ilkka, J., 2006. Mobile web Services,
Solutions To Keep Pace With Emerging Social Wiley.
Customer, Forrester Research, Inc. Finkle, J., 2007. SAP unveils Google-inspired software,
Webber, L., 2007. Marketing to the Social Web: How iPhone program. [Online]. Reuters. Available:
Digital Customer Communities Build Your Business, http://www.reuters.com/article/technologyNews/idUS
Wiley. N0452282620071204 .
Virki, T., 2007. Global cell phone penetration reaches 50 Lafferty J., McCallum A., Pereira F. (2001) Conditional
pct, Reuters. [Online]. random fields: probabilistic models for segmenting
Sudan, S. K., Drake, S. D., Ryan, S., Bhattacharyya, R., and labeling sequence data. In Proceedings of the
2007. Worldwide Mobile Enterprise 2007. Top 10 Eighteenth International Conference on Machine
Predictions. IDC. [Online]. Learning
Vaske, H., 2008. Wie Internet und Web 2.0 das Marketing Ponomarev V.V. 2004, Conceptual Model of the linguistic
veraendern. In CW executive briefing: CRM-Software. software complex of the expert search system with
[Online]. IDG Business Media GmbH. Available: elements of the socio-psycho linguistic determination.
http://whitepaper.cio.de/index.cfm?pid=1&pk=2118 . Dialog MEPhI, p. 176
Lomas, N., 2008. RIM: BlackBerry's future lies in social Ponomarev V.V.2005, Implementation of the new
networking. [Online]. CNET News.com. Available: approach to automated reporting from the
http://www.news.com/RIM-BlackBerrys-future-lies- terminal stations of the municipal information
in-social-networking/2100-1039_3-6230542.html. [20 system “Youth”. New software technologies,
April 2008] . Moscow, MGIEM, p. 44-51.
Oracle, 2008. Oracle Introduces Oracle(R) Mobile Sales
Assistant for Oracle CRM On Demand for BlackBerry.

404
 E-COMPLEMENTARITY
The Link to e-Business Value

Pedro Soto-Acosta and Angel L. Meroño-Cerdan

Department of Management & Finance, University of Murcia, Campus de Espinardo, Murcia, Spain
[email protected], [email protected]

Keywords: e-Business, Resource-based theory, Internet, business value, Information technology.

Abstract: In recent years, much debate about the value of e-Business and information technology (IT) has been raised.
Although the macro-level effect of IT and e-Business is undisputed, a question remains on whether e-
Business can provide differential benefits to individual firms. In this sense, there is a need to further
investigate whether and how e-Business creates value. To respond to this challenge, this paper develops a
conceptual model, grounded in the resource-based theory, which analyzes the complementarity of Internet
resources and e-Business capabilities as source of business value. This model posits three relationships:
Internet resources and business value, internal e-Business capabilities and business value, and the
complementarity of Internet resources and internal e-Business capabilities. To test hypotheses, a sample
comprising 1,010 Spanish firms is employed. The results show that, as hypothesized, Internet resources per
se are not positively related to business value and that internal e-Business capabilities have a positive
significant impact on business value. In addition, the results offer support for the complementarity of
Internet resources and internal e-Business capabilities as source of business value.

1 INTRODUCTION Amazon show e-Business can create business value,

but there is a question as to whether the lessons
The relationship between information technology learned from these “Internet giants” are more widely
(IT) and business value has been the subject of much applicable. At the same time, Carr’s assertions
research over the past decade. The results of these (2003), in his article “IT Doesn’t Matter”, have
studies were varied and the term “productivity raised the discussion about the value of IT. Carr’s
paradox” was coined to describe such findings. argument, in a few words, is that because every firm
Nonetheless, recent studies have found positive and can purchase IT in the marketplace and because IT is
stronger linkages, and have attributed the now a commodity based on standards that all
productivity paradox to variation in methods and companies can freely use, IT is no longer a
measures (Devaraj and Kohli, 2003) differentiating factor in organizational performance.
Today IT is surpassing its traditional “back What makes a resource truly strategic – what gives it
office” role and is evolving toward a “strategic” role the capacity to be the basis for a sustained
with the potential not only to support chosen competitive advantage - is not ubiquity but scarcity.
business strategies, but also to shape new business Carr argues that no firm can use IT to achieve a
strategies (Henderson and Venkatraman, 1999). competitive advantage over its competitors.
However, much debate about the value of IT and e- Therefore, Carr concludes, firms should reduce
Business has been raised, due to the gap between e- spending on IT, follow rather than lead IT in their
Business investment and the lack of empirical industry, and avoid deploying IT in new ways.
evidence on e-Business value. Although showing Most management information systems experts
recent signs of advance, much of the existing e- disagree with Carr’s assertions. However, his
Business literature still relies, to a great extent, on argument is appropriate when he points out that not
case studies, anecdotes, and conceptual frameworks, all IT investments have strategic value. Some IT
with little empirical research directed to assessing investments only allow firms to stay in business. The
the impact of IT on firm performance – especially in technology itself will rarely create superiority. For
traditional companies (Brynjolfsson and Kahin, that reason, some research studies found that IT
2002). Case studies on firms such as eBay and spending rarely correlates to superior financial

405
ICE-B 2008 - International Conference on e-Business

results (Hoffman, 2002). However, even though are what it can do as a result of teams of resources
competitors may copy an IT innovation, relative working together. Teece et al. (1997) argued that
advantage can be created and sustained where the capabilities cannot easily be bought; they must be
technology leverages some other critical resource. built. Thus, building capabilities is not only a matter
Kettinger et al. (1994) draw a number of such of combining resources; capabilities are rooted in
complementary resources, such as size, structure, processes and business routines. Also capabilities
culture, and so on, that could make it difficult for involve complex patterns of coordination between
competitors to copy the total effect of the people and between people and other resources
technology. This complementarity of resources is a (Grant, 1991), and between an organization and
corner stone of the resource-based theory and has other organizations. In this respect, Day (1994)
been offered as an explanation of how IT has largely describes capabilities as complex bundles of skills
overcome its paradoxical nature and is contributing and accumulated knowledge, exercised through
to business value (Bhatt and Grover, 2005; Clemons organizational processes, which enable firms to
and Row, 1991). coordinate activities and make use of their assets.
Consequently, to respond to these challenges, Day argues that capabilities and organizational
this paper develops a conceptual model, grounded in processes are closely entwined, because capabilities
the resource-based view (RBV) firms, to analyze the enable the activities in a business process to be
complementarity of Internet resources and carried out. More recently, Makadok (2001)
capabilities as source of business value at the level considers capability as a special type of resource.
of an individual firm. The analysis employs a large More specifically, he defines capability as an
sample of companies from different industries for organizationally embedded non-transferable firm-
hypothesis testing. Moreover, although recent specific resource whose purpose is to improve the
studies (Zhu, 2004; Zhu and Kraemer, 2005) have productivity of the other resources possessed by the
analyzed the relationship between e-Business firm.
capabilities and firm performance, very little work For the purposes of the present study, the above
has been undertaken to identify Internet resources definitions of capability permit the identification of
and capabilities. Similarly, the complementarity of three important characteristics:
Internet resources and capabilities has not been
ƒ Capabilities are rooted in processes and
studied. The present study attempts to cover these
business routines, because it is capability that
gaps in the research.
enables the activities in a business process to
The paper consists of six sections and is
be carried out.
structured as follows: The next section reviews the
ƒ Capabilities are firm-specific, while an ordinary
relevant literature. In Section 3, hypotheses and
resource is not. Because of this
research models are specified. Following that, the
embeddedness, ownership of a capability
methodology used for sample selection and data
cannot easily be transferred from one
collection is discussed. Then, data analysis and
organization to another.
results are examined. Finally, the paper ends with a
ƒ The primary purpose of a capability is to
discussion of research findings, limitations and
enhance the productivity of the other
concluding remarks.
resources that the firm possesses.

2.2 e-Business Resources and

2 LITERATURE REVIEW Capabilities
2.1 The RBV and e-Business The RBV provides a solid foundation to differentiate
between IT resources and IT capabilities and to
The RBV suggest that the effects of individual, firm- study their separate influences on performance
specific resources on performance can be significant (Santhanam and Hartono, 2003). Based on this
(Mahoney and Pandian, 1992). The RBV generally analysis, Bharadwaj (2000) suggested that if firms
tends to define resources broadly and include assets, can combine IT related resources to create unique IT
infrastructure, skills, and so on. While resources capabilities, they can improve their performance. IS
serve as the basic units of analysis, firms create researchers have followed this consideration of IT
competitive advantage by assembling resources that capability because competition may easily result in
work together to create organizational capabilities. the duplication of investment in IT resources, and
Grant (1991) suggests that the capabilities of a firm companies can purchase the same hardware and

406
 E-COMPLEMENTARITY - The Link to e-Business Value

software to remove competitive advantage online procurement to measure e-Business value.

(Santhanam and Hartono, 2003). In this respect, IS The business value of this process is discussed
research offers a useful distinction between IT below.
resources and IT capabilities. The former is asset- e-Procurement, or buying online, can potentially
based, while the latter comprises a mixture of assets provide distinct value propositions to the firm. These
formed around the productive use of IT. come from the reduction of procurement and
In general, IT resources are not difficult to inventory costs, as well as strategic networks with
imitate; physical technology is by itself typically suppliers that allow effective and efficient supply
imitable. However, firms may obtain competitive chain management (SCM). With regard to
advantages from exploiting their physical procurement costs, Kaplan and Sawhney (2002)
technology in a better (and/or different) way than indicated that buying in e-marketplaces considerably
other firms, even though competing firms do not reduces transaction costs. With regard to strategic
vary in terms of the physical technology they links and SCM, Internet technologies can enhance
possess. IT resources are necessary, but not a SCM decision making by enabling the collection of
sufficient condition, for competitive advantages real-time information, and access to and analysis of
(Clemons and Row, 1991). IT resources rarely this data in order to facilitate collaboration between
contribute directly to competitive advantage. trading partners in a supply chain. In this sense,
Instead, they form part of a complex chain of assets Frohlich and Westbrook (2002) showed the
(IS capabilities) that may lead to better performance. importance of linking customers and suppliers
Thus, some researchers have described this in terms together in tightly integrated networks. As a result of
of IT capabilities and argue that IT capabilities can e-Procurement, the collection of real-time
create uniqueness and provide organizations a information on demand is possible and, more
competitive advantage (Bhardwaj, 2000, Bhatt and importantly, products and services are delivered
Grover, 2005; Mata et al., 1995; Ross et al., 1996; quickly and reliably when and where they are
Santhanam and Hartono, 2003). needed (Frohlich, 2002).
Consequently, the present study seeks to In sum, e-Business value may lead to improved
demonstrate that although Internet resources performance on the part of the firm in procurement.
(considered as physical IT) are not responsible for Although it could be argued that customers,
the creation business value, their complementarity suppliers and/or the firm’s wider value network can
with e-Business capabilities is critical to firm value. benefit from online procurement, this study focuses
on analyzing business value at the level of an
2.3 Business Value from a Process individual firm.
Perspective
Although much research using the RBV has focused 3 DEVELOPMENT OF
on an aggregated dependent variable, namely, firm
performance, this may not be the best way to test the
HYPOTHESES
RBV (Ray et al., 2004). For example, because firms
can have competitive advantage in some business This section develops hypotheses for the present
activities and competitive disadvantage in others, study, drawing on the existing information systems
and e-Business literature. Three relationships will be
examining the relationship between resources and
capabilities associated with different processes explored: Internet resources and business value,
within a firm and its overall performance can lead to internal e-Business capabilities and business value,
and the complementarity of Internet resources and
misleading conclusions. Ray et al. (2004) proposed
examining the effectiveness of business processes as internal e-Business capabilities (see Figure 1).
a way to test the RBV logic. Another issue is that
some IT investments may provide benefits after a
INTERNET
RESOURCES
H1

certain period but increase operating costs in the

short term. Thus, using firm performance at the H3 BUSINESS VALUE
e-Procurement effectiveness

macro level is meaningless and can again lead to

misleading conclusions. These arguments lead to the
H2
conclusion that a process approach should be used to INTERNAL
E-BUSINESS
CAPABILITIES
explain the generation of e-Business value within the
RBV, and this is the approach adopted in the present
study. The present research uses the effectiveness of Figure 1: Research model.

407
ICE-B 2008 - International Conference on e-Business

3.1 Internet Resources and Business this capability logic of resources by arguing that
Value competitors may easily duplicate investments in IT
resources by purchasing the same hardware and
Firms obtain competitive advantages on the basis of software and, hence, IT resources per se do not
corporate resources that are firm specific, valuable, provide competitive advantages. Rather, it is the
rare, imperfectly imitable, and not strategically manner in which firms leverage their IT investments
substitutable by other resources (Barney, 1991). IT to create unique capabilities that impact firm
resources are easy to duplicate, and, hence, IT performance (Clemons and Row, 1991; Mata et al,
resources per se do not provide competitive 1995). Thus, it is expected that internal e-Business
advantages (Santhanam and Hartono, 2003). capabilities are positively associated with business
Although IT infrastructure is argued to be valuable, value. The following hypothesis incorporates these
it is not a source of competitive advantage (Bhatt y expectations:
Grover, 2005). Thus, IT infrastructure will rarely Hypothesis 2: There is a positive relationship
lead to superior performance. Similarly, Internet between internal e-Business capabilities and
resources – as defined above – are not difficult to business value
imitate. In general, Internet technology is by itself
imitable. If one firm can purchase certain Internet 3.3 The Complementarity of Internet
technologies and thereby implement some strategies, Resources and Internal e-Business
then other firms should also be able to purchase Capabilities
these technologies, and thus such tools should not be
a source of competitive advantage. Furthermore, as Although there is research that posit a direct
the diffusion of the Internet continues, the ability of relationship between IS resources/capabilities and
proprietary IT to be a source of competitive firm performance (Bharadwaj, 2000; Feeny and
advantage continues to be eroded. These arguments Willcoks, 1998; Santhanam and Hartono, 2003),
suggest that Internet resources may not have a others have questioned the direct-effect argument
significant impact on business value. Thus, the and emphasized that IS resources/capabilities are
following hypothesis is proposed: likely to affect firm performance only when they are
Hypothesis 1: There is no relationship between deployed to create unique complementarities with
Internet resources and business value other firm resources (Clemons and Row, 1991;
Powell and Dent-Micallef, 1997).
3.2 Internal e-Business Capabilities Firm resources are considered complementary
and Business Value when the presence of one resource enhances the
value or effect of another resource (Ravichandran y
Investing in IT is not a necessary nor sufficient Lertwongsatien, 2005; Zhu, 2004). For example, the
condition for improving firm performance, since IT complementarity between online offerings and
investments might be misused (Tallon et al., 2000). offline assets is the essence of “clicks-and-mortar”
In this sense, IT assets cannot improve companies. Customers who buy products over the
organizational performance if they are not used Internet value the possibility of getting support and
appropriately. However, when used appropriately IT service offered through bricks-and-mortar retail
is expected to create intermediary effects, such as IT outlets, including the convenience of in-store pickup
being embedded in products and services, and return (Zhu, 2004). Hence the RBV highlights
streamlined business processes, and improved the role of complementarity as a source of value
decisions, which can be expected to have an creation in e-Business, though is not the only source
influence on the performance of the firm as suggested by Amit and Zott (2001). As mentioned
(Ravichandran and Lertwongsatien, 2005). earlier, Internet resources are not difficult to imitate
Grant (1991) and Makadok (1991) emphasize and per se do not provide competitive advantages.
that while resources by themselves can serve as However, having a proper Web infrastructure may
basic units of analysis, firms create competitive facilitate the internal processing of online operations
advantage by assembling these resources to create and this way influence positively firm performance.
organizational capabilities. Makadok states that That is, the fact of possessing an adequate Web
these firm-specific capabilities, embedded in infrastructure can be critical for the influence of
organizational processes, provide economic returns internal e-Business capabilities on business value.
because that firm is more effective than its rivals in Thus, the following hypothesis is proposed:
deploying resources. IS researchers have adopted

408
 E-COMPLEMENTARITY - The Link to e-Business Value

Hypothesis 3: The complementarity between analysis (CFA) was used to test the constructs.
Internet resources and internal e-Business Based on the CFA assessment, the constructs were
capabilities explains variations in business value further refined and then fitted again. Constructs and
associated indicators are listed in the Appendix and
discussed below.
4 METHODOLOGY ƒ Internet resources construct. This construct
represents the adoption of physical Internet
4.1 Data technologies. In this sense, respondents were
required to assess the presence of four Internet
The data source for the present study is the e- tools: website, Intranet, Extranet and LAN
Business W@tch survey 2004, an initiative launched (local area network).
by the European Commission for monitoring the ƒ Internal e-Business capabilities. This construct
adoption of IT and e-Business activity. The represents the use of online technologies for
decision-maker targeted by the survey was normally supporting internal business processes.
the person responsible for IT within the company, ƒ Business value. As discussed earlier in section
typically the IT manager. Alternatively, particularly 2.3, the present research uses the effectiveness
in small enterprises without a separate IT unit, the of e-Procurement for measuring business
managing director or owner was interviewed. value. That is, business value is assessed
The population considered in this study was the set through the business impact of purchasing
of all enterprises which are active at the national online.
territory of Spain and which have their primary
business activity in one of ten sectors considered. 4.3 Instrument Validation
The sample drawn was a random sample of
companies from the respective sector population CFA using AMOS 4.0 was conducted to assess
with the objective of fulfilling strata with respect to empirically the constructs theorized. Multiple tests
business size. A share of 10% of large companies on construct validity and reliability were performed.
(250+ employees), 30% of medium sized enterprises Model fit was evaluated using the maximum
(50-249 employees) and 25% of small enterprises likelihood (ML) method. The measurement
(10-49 employees) was intended. The number of properties are reported below.
firms totalled 1 010. 91.1% of firms were small and Construct reliability. All constructs had a
medium-sized enterprises (less than 250 employees) composite reliability over the cut-off of 0.70 (Straub,
and each sector considered had a share of around 1989), and also the average variance extracted for all
10% of the total sample. exceeded the preferred level of 0.5 (Churchill,
With regard to respondents’ titles, 54.4% were IS 1979).
managers, nearly 20% were managing directors, and Content and construct validity. Content validity
12.1% were owners. The dataset was examined for was verified by checking the meanings of indicators
potential bias in terms of the respondents’ titles. and by a careful literature review. Construct validity
Since respondents included both IT managers and is the extent to which a construct measures the
non-IT managers, one could argue that IT managers concepts that it purports to measure (Straub, 1989).
may overestimate e-Business value. To test this It has two components: convergent and discriminant
possible bias, the sample was divided into two validity. After dropping insignificant items, all
groups: IS managers (head of IT/DP and other IT estimated standard loadings were significant,
senior managers) versus non-IS managers (owner, suggesting good convergent validity. To assess the
managing director, strategy development and discriminant validity Forell and Larcker’s (1981)
others). One-way ANOVA was used to compare the criterion was used. All constructs met this criterion.
means of factor scores between the two groups. No Table 1 lists several goodness-of-fit statistics to
significant differences were found, suggesting that assess how well specified models explain the
the role of the respondents did not cause any survey observed data. The insignificant p-value (p = 0.187)
biases. for the chi-square statistics implied good absolute
fit. The root mean square error of approximation
4.2 Measures of Variables (RMSEA) is was below the cut-off value 0.08
suggested by Browne and Cudeck (1993). Five
Measurement items were introduced on the basis of incremental fit indices were all above the preferred
a careful literature review. Confirmatory factor level of 0.9 (Gefen et al., 2000).

409
ICE-B 2008 - International Conference on e-Business

Table 1: Measurement Model Fit indices. 6 DISCUSSION

Goodnesss-of-Fit Indices
Chi-Square 66.054 The results showed that Internet resources are not
p-value 0.246 positively related to business value. This finding is
RMSEA 0.032 not surprising, since competitors may easily
Normed Fit Index (NFI) 0.971 duplicate investments in IT resources by purchasing
Relative Fit Index (RFI) 0.955 the same hardware and software, and hence IT
Incremental Fit Index (IFI) 0.997 resources per se do not provide better performance
Tucker-Lewis Index (TLI) 0.995 (Santhanam and Hartono, 2003). This can be
Comparative Fit Index (CFI) 0.997 explained through the RBV, because IT is not
considered a resource that is difficult to imitate; IT is
by itself typically imitable. This result supports the
findings of recent research (Batt and Grover, 2005)
5 EMPIRICAL RESULTS that did no find evidence of a positive link between
IT quality and firm performance. Similarly, Powell
Hypotheses were tested using hierarchical regression and Dent-Micallef (1997) showed that IT by itself
analysis. Table 2 shows Internet resources construct cannot be a source of competitive advantage. Thus,
is not statistically significant, whereas e-Business our results confirm that Internet technology by itself
internal capabilities construct is positive and will rarely create business value.
significant (regression 1), as predicted. Regression 2 Furthermore, results demonstrate that there is a
includes both main and the interaction effect. The positive relationship between internal e-Business
interaction effect between Internet resources and capabilities and business value. Our findings
internal e-Business capabilities was found confirm the existing empirical literature. Bharadwaj
significant, thus, supporting the complementarity of (2000) and Santhanam and Hartono (2003) found
Internet resources and internal e-Business that firms with superior IT capability do indeed
capabilities. To further test the significance of the exhibit superior firm performance. Ravichandran
interaction effect, the incremental R2 between the and Lertwongsatien (2005) showed that an
full model (with interaction term) and the partial organization’s ability to use IT to support its core
model (without the interaction terms) was compared. competences depends on IS capabilities. Thus, even
The result is reported in the lower rows of table 7. In though competing firms do not vary in terms of the
regression 2, the incremental R2 was 0.026, meaning IT they possess, IS capabilities are rooted in
that approximately an additional 3 percent of processes and business routines and provide
explained variance has resulted from the inclusion of competitive advantage. In this sense, the results of
the interaction effect with respect to regression 1. To the present study support the proposition that
compare the partial model against the full models, a internal e-Business capabilities are positively
Wald test was performed and the differences were associated with business value.
found to be statistically significant. Based on this, Finally, the empirical results offer support for the
the partial model was rejected in favour of the full complementarity of Internet resources and internal
model (Greene, 2000). e-Business capabilities. The RBV highlights the role
of complementarities between resources as a source
Table 2: Complementarity results: Impact on procurement. of business value. Researchers such as Steinfield et
Independent variables Regr. 1 Regr. 2 al. (1999) suggest that e-Business value can come
from synergies between online and offline presence.
Internet resources (IR) 0.083 0.099 In this sense, using case studies, they showed the
Internal EB capabilities (IEBC) 0.196*** 0.151** lack of exploitation of these synergies in SMEs. Zhu
IR * IEBC - 0.167** (2004) developed a study which evaluates the impact
F 6.595*** 6.633*** of e-commerce and IT on firm performance
(financial measures), studying both the main effects
R2 0.047 0.069
and the interaction effect of e-commerce and IT on
∆R2 0.026** firm performance. Our results support the RBV and
p<0.1*; p<0.05**; p<0.01*** e-Business literature, therefore, it can be concluded
that having an adequate Internet infrastructure can
Through this analysis, hypotheses H1, H2 and H3 be critical for the impact of internal e-Business
found support. capabilities on business value.

410
 E-COMPLEMENTARITY - The Link to e-Business Value

7 CONCLUSIONS, LIMITATIONS are created over years. A longitudinal study could

AND FUTURE RESEARCH enrich the findings.

In recent years, much debate about the value of IT

and e-Business has been created, due to the gap REFERENCES
between e-Business investment and the lack of
empirical evidence on e-Business value. Thus, today Amit, R. and Zott, C. (2001) Value creation in e-Business.
IS researchers face pressure to answer the question Strategic Management Journal, 22, 493-520.
of whether and how e-Business creates value. Barney, J.B. (1991) Firm Resources and Sustained
Competitive Advantage. Journal of Management, 7,
The complementarity of resources is a corner stone
99-120
of the resource-based theory and has been offered as Bharadwaj, A.S. (2000) A resource-based perspective on
an explanation of how IT has largely overcome its information technology capability and firm
paradoxical nature and is contributing to business performance: an empirical investigation. MIS
value (Bhatt and Grover, 2005; Clemons and Row, Quarterly, 24, 169-196.
1991). Thus, to respond to these challenges, this Bhatt, G.D. & Grover, V. (2005) Types of information
study developed a conceptual model, grounded in technology capabilities and their role in competitive
the resource-based view (RBV) firms, to analyze the advantage: an empirical study. Journal of Management
complementarity of Internet resources and e- Information Systems, 22, 253-277.
Boudreau, M., Gefen, D. & Straub D. (2001) Validation in
Business capabilities as source of business value at
IS research: A state-of-the-art assessment. MIS
the level of an individual firm. The analysis Quarterly, 25, 1-24.
employed a large sample of companies from Browne, M.W. & Cudeck, R. (1993) Alternative ways of
different industries for hypothesis testing. Broadly, assessing model fit. In: Testing structural equation
this research offers several contributions: (1) it models, Bollen, K.A. & Long, J.S. (ed.), pp. 136-162.
identifies Internet resources and internal e-Business Beverly Hills: Sage.
capabilities; (2) it shows that Internet technology by Brynjolfsson, E. & Kahin, B. (2002) Understanding the
itself will rarely create business value; (3) it sheds digital economy. Cambridge, MA: MIT Press.
light on the complementarity of Internet resources Carr, N. (2003) IT doesn’t matter. Harvard Business
Review, May 2003, 41-49.
and internal e-Business capabilities as source of
Churchill, G. A. (1979) A paradigm for developing better
business value. measures of marketing constructs. Journal of
While the contributions of the present study are Marketing Research, 16, 64-73.
significant, it has some aspects which can be Clemons, E.K. & Row, M.C. (1991) Sustaining IT
addressed in future research. First, the sample used advantage: the role of structural differences. MIS
was from Spain. It may be possible that the findings Quarterly, 15, 275-292.
could be extrapolated to other countries, since Day, G.S. (1994) The capabilities of market-driven
economic and technological development in Spain is organizations. Journal of Marketing, 58, 37-52.
similar to other OECD Member countries. However, Devaraj, S. & Kholi, R. (2003) Performance Impacts of
Information Technology: Is Actual Usage the Missing
in future research, a sampling frame that combines
Link?. Management Science, 49, 273-289.
firms from different countries could be used in order Feeny, D.F. and Willcocks, L.P. (1998) Core IS
to provide a more international perspective on the capabilities for exploiting information technology,
subject. Second, the business value measure is Sloan Management Review, 39, 9-21.
subjective in the sense that it was based on Likert- Fornell, C. & Larcker, F.D. (1981) Evaluating structural
scale responses provided by managers. Thus, it equation models with unobservable variables and
could also be interesting to include objective measurement error. Journal of Marketing Research,
performance data for measuring business value. 18, 39-50.
Third, the key informant method was used for data Frohlich, M.T. (2002) e-Integration in the supply chain:
barriers and performance. Decision Sciences, 33, 537-
collection. This method, while having its
555.
advantages, also suffers from the limitation that the Frohlich, M.T. & Westbrook, R. (2002) Demand chain
data reflects the opinions of one person. Future management in manufacturing and services: web-
studies could consider research designs that allow based integration, drivers and performance. Journal of
data collection from multiple respondents within an Operations Management, 20, 729-745.
organization. Fourth, this research takes a static, Gefen, D., Straub, D.W. & Boudreau, M.C. (2000)
cross-sectional picture of capabilities, which makes Structural equation modeling and regression:
it difficult to address the issue of how capabilities Guidelines for research practice. Communications of
the AIS, 4, 1-78.

411
ICE-B 2008 - International Conference on e-Business

Grant, R.M. (1991) The resource-based theory of Journal of Management Information Systems, 16, 137-
competitive advantage: implications for strategy 165.
formulation. California Management Review, 33, 114- Teece D.J, Pisano G. & Shuen A. (1997) Dynamic
135. capabilities and strategic management. Strategic
Greene, W. (2000) Econometric analysis, Upper Saddle Management Journal, 18, 509-533.
River, 4th edition, NJ: Prentice Hall. Zhu, K. (2004) The complementarity of information
Gunasekaran, A., Love, P.E.D., Rahimi, F. & Miele, R. technology infrastructure and e-commerce capability:
(2001) A model for investment justification in a resource-based assessment of their business value.
information technology projects. International Journal Journal of Management Information Systems, 21, 167-
of Information Management, 21, 349-364. 202.
Henderson, J. & Venkatraman, N. (1999) Strategic Zhu, K. & Kraemer, K.L. (2005) Post-adoption variations
alignment: Leveraging transforming organizations”, in usage and value of e-Business by organizations:
IBM Systems Journal, 31, 472-484. cross-country evidence from the retail industry.
Hoffman, T. (2002) Frugal IT investors top best-performer Information Systems Research, 16, 61-84.
list. Computerworld, December 6.
Kaplan, S. & Sawhney, M. (2000) E-hubs: the new B2B
marketplaces. Harvard Business Review, 70, 71-79.
Kettinger, W.J., Grover, V., Guha, S. and Segars, A.H. APPENDIX: MEASURES
(1994) Strategic information systems revisited: a study
insustainability and performance, MIS Quarterly, 18, Internet Resources:
31-58. Does your company have a website? (Y/N)
Mahoney, J.T, Pandian, J.R. (1992) The resource-based Does your company use an Intranet? (Y/N)
view of the firm within the conversation of strategic Does your company use an Extranet? (Y/N)
management. Strategic Management Journal, 13, 363-
Does your company use a LAN? (Y/N)
380.
Makadok, R. (2001) Toward a synthesis of the resource- Internal e-Business capabilities:
based and dynamic-capability views of rent creation. Do you use online technologies to share documents
Strategic Management Journal, 22, 387-402. between colleagues or to perform collaborative work
Mata, F.J., Fuerst, W.L. & Barney, J.B. (1995)
in an online environment? (Y/N)
Information technology and sustained competitive
advantage: a resource-based analysis. MIS Quarterly, Do you use online technologies to track working
19, 487-505. ours and production time? (Y/N)
Powell, T.C. & Dent-micallef, A. (1997) Information Do you use online technologies to support human
technology as competitive advantage: the role of resources management? (Y/N)
human, business, and technology resources. Strategic When an online order comes, is the order fully
Management Journal, 18, 375-405. integrated with the back-end system? (Y/N)
Ravichandran, T. & Lertwongsatien, C. (2005) Effect of
Information Systems Resources and Capabilities on Business value: e-Procurement effectiveness
Firm Performance: A Resource-Based Perspective. What effect has online procurement on the
Journal of Management Information Systems, 21, 237- procurement costs? (1-5)
276. What effect has online procurement on your
Ray, G., Barney, J.B. & Muhanna, W.A. (2004) relations to suppliers? (1-5)
Capabilities, business processes and competitive What effect has online procurement on the costs of
advantage: choosing the dependent variable in
empirical tests of the resource-based view. Strategic
logistics and inventory? (1-5)
Management Journal, 25, 23-37. Note. (Y/N), dummy variable; (1-5), five-point
Ross, J.W., Beath, C.M. & Goodhue, D.L. (1996) Develop Likert-type scale.
long-term competitiveness through IT assets. Sloan
Management Review, 38, 31-42.
Santhanam, R. & Hartono, E. (2003) Issues in linking
information technology capability to firm
performance. MIS Quarterly, 27, 125-153.
Steinfield, C., Mahler, A. & Bauer, J. (1999) Electronic
commerce and the local merchant: opportunities for
synergy between physical and Web presence,
Electronic Markets, 9, 51-57.
Straub, D.W. (1989) Validating Instruments in MIS
Research. MIS Quarterly, 13, 147-169.
Tallon, P., Kraemer, K. & Gurbaxani, V. (2000)
Executives’ perceptions of the business value of
information technology: a process-oriented approach.

412
 A RESEARCH MODEL OF CUSTOMER RELATIONSHIP
MANAGEMENT SYSTEMS FOR MOBILE DEVICES
Description of a Research Model about Customer Relationship
Management Projects

Rebecca Bulander
Institute AIFB, University of Karlsruhe, Englerstr. 11, 76 128 Karlsruhe, Germany
Email: [email protected]

Keywords: Customer Relationship Management, CRM, IT, CRM systems, CRM projects, mobile devices, mobile
actors, business to business.

Abstract: In literature we can find evidence about a high number of failed CRM projects. However concerning CRM
projects in combination with the implementation of mobile devices for field staff in sales and services there
aren’t many results published. So there are missing information about the success of such projects, the
success factors and the objectives. This paper presents a research model about such CRM projects. A CRM
project in this paper is defined as an implementation of a CRM system together with mobile devices. The
research model will give further insight into the success factors of such projects and the achieved objectives.

1 INTRODUCTION and the correct means by which to motivate valuable

customers to remain loyal – indeed, to buy again”
Several studies (e. g. Alt 2004; Thomson et. al 2002; (Dyché 2002, p. 4).
Ryals 2000) document the failure of CRM projects The information technology respective the CRM
in the past. But concerning CRM projects related systems help to enhance the customer relation and to
with the implementation of mobile devices for field standardise the CRM activities and processes in
staff in sales and services we find nearly no fields of marketing, sales and service. This paper is
information so we can make no statement about how focussing on CRM systems integrating actors using
many of these projects failed. To gain more infor- mobile devices. In terms of our research mobile
mation about this and to prevent such CRM projects devices are mobile handheld devices like cellular
from failing it seems reasonable to inspect such phones, personal digital assistants (PDA) and
projects. Therefore it is important to know what are Smartphones but also on-board vehicle computers or
the success factors and the achieved objectives of notebooks.
such projects. For this reason a research model will A CRM project is a project which includes the
be given in this paper. The research model focuses implementation of a CRM system together with
on companies which sell their products and services mobile devices but also the adoption of a customer
to other companies – this is the so called business to oriented strategy. A project is a timely restricted
business (B2B) sector. The B2B sector is more endeavour to archive stated objectives; it has an
appropriate because in this sector we find a higher assigned budget.
number of field staff working with mobile devices in CRM systems together with mobile devices
combination with a CRM solutions than in the B2C expand the channels of CRM to the customer
sector. (Sundararajan 2002) and assist both actors of CRM
For a better understanding there will be given – employees of the suppliers and of the customers
some definitions. Customer Relationship Manage- using mobile devices via public and private com-
ment (CRM) is a customer oriented strategy which is munication networks. Because the employees of the
supported by information technology (IT) (Hippner supplier (e. g. field manager gets mobile access to
2004). CRM “[…] is the infrastructure that enables customer information) and/or of the customer (e. g.
the delineation of and increase in customer value, customer receives flight information on his cellular
phone) can be in the working state mobile or

413
ICE-B 2008 - International Conference on e-Business

stationary we can differentiate between four CRM The last point is important because this infor-
scenarios like shown in Figure 1 (Hampe 2002). In mation gives more insights about the result of an
the mobile state the mobile actor is not located implementation of mobile devices. The last column
within its homelike infrastructure. The research in Table 1 shows if the research about the
model will enclose the three scenarios “classical relationship between the success factors and the
CRM system”, “field staff” and “mobile link”. The objectives is part of the study. This aspect is
scenario “mobile services” covers other aspects of described with “connection between the variables”.
CRM like mobile marketing and needs therefore All studies concerning this research topic are listed
other IT infrastructure. in Table 1.
As can be seen in Table 1 there is no study which
Customer
covers all listed aspects. Most of the studies focus on
Working state the sector of B2C or B2B and B2C. There is only
of an stationary mobile
employee
one study that focuses only on B2B. No study covers
the aspect of mobile information technology and the
Classical CRM
achieved objectives in field staff. Some studies are
stationary Mobile Services
system influenced by a company which delivered the CRM
Supplier

systems or which did the consulting during the CRM

projects.
mobile Field staff Mobile Link According to the described research gaps there is
a need of more research concerning mobile infor-
mation technology in combination with CRM
Figure 1: CRM scenarios using mobile devices. systems and CRM strategy. It is also necessary to
analyse the objectives which are achieved by
implementing CRM systems together with mobile
devices by looking at the field staff. Therefore a
2 LITERATURE REVIEW conceptual development will be given and a research
model will be presented.
Up to now, little research has been conducted in this
described field of research (Salomann et al. 2005).
To get a better overview about the research that has
already been done, they will be grouped by the 3 CONCEPTUAL
following aspects: DEVELOPMENT
• Author and year
• Objective, research item and research In the next paragraphs the research model in
focus business to business or business Figure 2 will be explained. The ellipses represent
to customer sector latent variables; these are the constructs of the
• Method of data analysis model. The relationships between the variables are
• Database (sample size, kind of ques- shown by the arrows between the ellipses. Each
tionning, kind of investigation and arrow is marked with ‘H’ and a number, which
countries in which the enterprises of stands for the number of the hypothesis. The whole
the study are) model contains 20 hypotheses. As shown in Figure 2
• Success factors of the CRM project the success factors are on the left side and the
• Achieved objectives objectives on the right side of the figure. According
to Table 1 the success factors are divided into “CRM
The success factors are grouped by:
as strategy” and “implementation of CRM together
• CRM as strategy with mobile IT”. The objectives are categorized into
• Implementation of CRM system three groups: The first group is the over all
• Implementation of mobile information satisfaction with the CRM project. The remaining
technology (mobile IT) groups are the objectives achieved in the whole
Also the achieved objectives of the CRM pro- enterprise. They are measured in four dimensions.
jects are sub-classified after the following aspects: The objectives referring to the field staff and the use
• related to the enterprise of mobile devices are measured in three variables,
• containing information about the also called dimensions.
objectives of the field staff

414
 A RESEARCH MODEL OF CUSTOMER RELATIONSHIP MANAGEMENT SYSTEMS FOR MOBILE DEVICES -
Description of a Research Model about Customer Relationship Management Projects

Table 1: Selected studies on CRM systems.

Author and objective, research item and method of data database success factors achieved
year research focus business to analysis - Sample size object-

Connections between variables

business (B2B) or business to - Kind of questioning tives
customer (B2C) - Kind of investigation
- Country in which the

Mobile information
CRM as strategy
survey did take place

CRM system

technology

enterprise

field staff
Alt, - Success factors of CRM Explorative -N = 6
Pushmann, projects analysis and - Personal interviews
Österle - Research item: enterprises benchmarking - Half standardized
2004 - B2C questionnaire
- Countries: Germany,
Switzerland, USA
Day, Van - Positional advantages and Qualitative - N = 299
den Bulte relative performance through analysis, - Written and online
2002 customer relating capability factor analysis questioning
and competitive strategy - Standardized questionnaire
- Research item: middle and - Country: USA
big enterprises
- B2B and B2C
Greve - Success factors of the Qualitative - N = 90
2006 implementation of CRM analysis, - Written questioning
systems descriptive - Standardized questionnaire
- Research item: enterprises statistics, PLS - Countries: Europe, mostly
which are consulted by the analysis Germany
company Accenture
- B2C
Jayachan- - Influence of the CRM Qualitative - N = 172
dran et al. information process ands the analysis, - Written and online
2004 use of CRM technology to factor analysis questioning
the customer relationship - Standardized questionnaire
performance - Country: no information
- Research item: enterprises
- B2B and B2C
Reinartz, - Success factors and Qualitative - N = 211 and N = 95
Krafft, objectives of CRM analysis, - Written and online
Hoyer implementations descriptive questioning
2003 - Research item: enterprises statistics, PLS - Standardized questionnaire
- B2C analysis - Countries: Germany,
Austria, Switzerland
Selchert - Measurement of success of Descriptive - N = 32
2005 CRM projects in companies statistics, - Telephone interviews
which have chosen CRM Key performance - Standardized questionnaire
systems of SAG AG indicators - Countries: Germany,
- Research item: enterprises Austria, Switzerland
- B2B and B2C
Wilson et al. - Success factors in CRM Explorative -N = 5
2002 - Research item: enterprises analysis - Personal interviews
- B2B and B2C - Half standardized
questionnaire
- Country: no information
Zahay, - Investigation about the Quantitative - N = 206
Griffin customer based performance analysis, - Telephone interview
2002 and the business growth descriptive - Standardized questionnaire
- Research item: enterprises statistics, - Countries: USA
- B2B factor analysis

415
ICE-B 2008 - International Conference on e-Business

Complete satisfaction with the

CRM project

Satisfaction

Success factors
H5
Achieved objectives in the
enterpsie
Actions
Actions for
for
acquisition
acquisition H1 H6 Dimension
of
of customers
customers finances
CRM
CRM performance
performance H7-H10
Actions Dimension
Actions for
for H2
customer customer
customer retention
retention H11-H13

Dimension
employee

Dimension
processes

Implementation
Implementation of
of Achieved objectives in the
mobile
mobile information
information H3 H14-H17 field staff
technology
technology Dimension
Implementation
Implementation ofof
information
information technology
technology time and costs
Implementation
Implementation of
of - mobile
H4
CRM
CRM system
system H18-H20 Dimension
quality - mobile

Dimension
processes - mobile

Figure 2: Research model.

3.1 Success Factor CRM Performance 3.2 Success Factor Implementation of

Information Technology
The variable CRM performance stands for the ability
of an enterprise to adapt the CRM strategy in its The variable implementation of information
enterprise strategy and the CRM concepts and technology stands for the support of all customers
activities in its marketing, sales and service con- related and oriented activities by IT. Especially in
cepts. To measure the CRM performance three con- CRM are many activities, which wouldn’t be
structs are needed. Some aspects of these constructs possible in an efficient way without IT – e. g. per-
are transferred from the study of Reinartz 2004. But sonalized advertising and newsletters. Therefore you
the constructs of this study have to be converted find already in the definition of CRM the importance
from the B2C sector to special aspects of the B2B of IT (Hippner 2004).
sector. Therefore the consciously determination of a To get an IT that covers all aspects of CRM there
customer relationship is not part of this model (Rapp is a combination of the implementation of mobile IT
2000). So we get two antecedents for CRM and of a CRM system necessary. Therefore the
performance. variable implementation of information technology
First there are the actions to acquire new has two antecedents.
customers and second there are the actions to force Only if the mobile technology and all other
the relationship to profitable customers and also to channels to the customer are integrated into the
recommit inactive profitable customers (Reinartz CRM system to ensure a bidirectional data flow on a
2004). consolidated customer data base all CRM processes
The actions to acquire new profitable customers and activities can be properly supported. This aspect
and to build a relationship to the already profitable is shown by hypotheses 3 and 4.
customers haven an influence on the CRM per-
formance of an enterprise (hypotheses 1 and 2).

416
 A RESEARCH MODEL OF CUSTOMER RELATIONSHIP MANAGEMENT SYSTEMS FOR MOBILE DEVICES -
Description of a Research Model about Customer Relationship Management Projects

3.3 CRM Performance and • process for mobile devices (e. g. in-
Implementation of IT as crease of process automation for sales
Antecedents of Satisfaction and service people)
The antecedents to these seven constructs are the
To measure enterprises satisfaction with the results two variables “CRM performance” and the “im-
of its determined CRM project in combination with plementation of IT”. Hence the relationship between
the usage of mobile devices the variable these two variables and the variables in the certain
“satisfaction with the CRM project” will be used. dimensions are described in the hypotheses 7 until
This variable shows the over all impression and 20 (see Figure 2). All hypotheses in the research
covers the success of the CRM project over all. model show the following description – for example
Therefore also the question is formulated, if the H7:
enterprise would repeat the CRM project. “The CRM performance has an influence on the
The satisfaction with the CRM project is in- achieved objectives of the enterprise in the
fluenced by the two variables CRM performance and dimension finance.”
implementation of IT (see also hypotheses 5 and 6 in
Figure 2).
4 RESEARCH METHOD AND
3.4 CRM Performance and
Implementation of IT as FURTHER STEPS
Antecedents of Achieved Objectives
4.1 Research Method
Unlike the variable satisfaction with the CRM
project the variables to measure the achieved The proposed research model and its hypotheses
objective cover certain aspects of CRM projects. represent a structural equation path model. This
Therefore a pre-study was undertaken to collect the model can be tested by the Partial Least Squares
objectives of CRM projects for mobile devices (PLS) analysis. The PLS procedure was invented by
(Hartel 2006). The results of this pre.-study have Hermann Wold. It is a second-generation multi-
been grouped by topics. According to this pre-study variate technique which has the ability to model
we can differ between objectives related to the latent constructs under conditions of non-normality
enterprise and to the field staff. Furthermore they (Chin 1999).
can be assigned to sub-groups – so called dimen-
sions. 4.2 Further Steps
The achieved objectives concerning the whole
enterprise are grouped by four dimensions con- To prove the proposed research model further
cerning the following aspects: empirical research is needed. Therefore the variables
have to be operationalized by indicator variables.
• finance (e. g. cost reduction or increase
After this the concept of the study has to be planned.
of up- and cross-selling)
This covers the aspects like: data base, sample size,
• customer (e. g. improvement of cus- kind of questioning etc. Each indicator variable
tommmer communication or reducetion refers to a question in the questionnaire.
of complaints) Because of the different market of CRM systems
• employees (e. g. improvement of the only enterprises of one CRM market should be
communication between employees) chosen for interviews – e. g. American CRM market
• processes (e. g. improvement of effect- or German speaking CRM market. Thus an
tiveness in sales) influence by the market to the study results can be
The achieved objectives concerning the filed excluded.
staff using mobile devices are grouped by the By choosing the enterprises for interviews there
following three dimensions: should be avoided that there is any influence by a
• time and costs (e. g. reduction of cycle special CRM software product or a consulting
time or time saving by administrative company during the CRM project. The chosen
work for sales people) enterprises should be free from such influences.
• quality (e. g. improvement of data and It is also important to select companies which
information quality for sales people) have conducted a CRM project and completed it.
Therefore it can be ensured that first influences as

417
ICE-B 2008 - International Conference on e-Business

well as long term influences of the CRM project for Hippner, H., 2004. Zur Konzeption von Kunden-
example in the dimension finance can be observed. beziehungsstrategien (in German). In Hippner, H. &
Wilde, K. (Ed.), Management von CRM-Projekten,
Gabler, Wiesbaden, pages 33-66.
Jayachandran, S.; Hewett, K.; Kaufman, P., 2004.
5 SUMMARY Customer Response Capability in a Sense-and-
Response Era: The Role of Customer Knowledge
The described research model shows a concept to Process. In: Journal of Academy of Marketing Science
analyze CRM projects in enterprises of the business 32, 3, pages 219-233.
Rapp, R., 2000 (in German). Customer Relationship
to business sector. The model covers aspects of
Management - Das neue Konzept zur Revolutio-
CRM strategy and CRM concepts as well as the nierung der Kundenbeziehungen. Campus, Frankfurt a.
support of the implemented information technology M.
respective CRM systems. The model points out that Reinartz, W.; Krafft, M.; Hoyer, W. D., 2004: The
the CRM performance and the IT have an influence Customer Relationship Management Process: Its
on the achieved objectives of an CRM project. So Measurement and Impact on Performance. In: Journal
the model shows that IT is not the only but an of Marketing Research XLI, pages 293-305.
important factor to the success of such a CRM Ryals, L.; Knox, S., Maklan, S., 2000. Customer Relation-
project. ship Management (CRM): Building the Business Case.
London: Financial Times Prentice-Hall.
The most important points of the model are that
Salomann, H., Dous, M., Kolbe, L. & Brenner, W., 2005.
the variable “implementation of mobile information Customer Relationship Management Survey – Status
technology” covers all aspects for the use of mobile Quo and Future Challenges. Institute of Information
devices for field staff in sales and services. It is also Management at the University of St. Gallen.
important that the objectives of the field staff are Selchert, M., 2005. CFROI of Customer Relationship
measured and analyzed by three dimensions con- Management. In: Managementschriften. Kremin-Buch,
cerning aspects like time and costs, quality and B.; Unger, F.; Walz, H. (Hrsg.), Wissenschaft und
processes for mobile devices. Praxis, Sternenfels.
Sundararajan, P., 2002. Emerging Mobile Customer
Relationship Management. In Applications in
Financial Services. EAI Journal, Mai 2002, S. 44-47.
REFERENCES Thompson, E., Davies, J. & Frey, N., 2002. Gartner's 2002
European CRM Survey: As Projects Progress,
Alt, R. & Puschmann, T., 2004. Successful Practices in Challenges Abound.
Customer Relationship Management. In Proceedings Wilson, H.; Daniel, E.; McDonald, M., 2002. Factors for
of the 37th Hawaii International Conference on Success in Customer Relationship Management
System Sciences, Hawaii. IEEE Computer Society. (CRM) Systems. In: Journal of Marketing Mana-
Chin, W. W.; Newsted, P. R., 1999: Structural equation gement 18, 1/2, pages 193-219.
modeling analysis with small samples using least Zahay, D.; Griffin, A., 2002. Are customer information
squares. In: Statistical strategies for small sample systems worth it? Results from B2B services. In:
research. Hoyle, R. H. (Hrsg.), Thousand Oaks et al. Marketing Science Institute, Report No. 02-113,
pages 307-342. Cambridge.
Day, S. G.; Van den Bulte, C., 2002. Superiority in
Customer Relationship Ma¬nagement: Consequences
for Competitive Advantages and Perfor¬mance. In:
Marketing Science Institute, Report No. 02-123,
Cambridge.
Dyché, J., 2002. The CRM-Handbook. Addison-Wesley,
Boston.
Greve, G., 2006 (in German). Erfolgsfaktoren von
Customer-Relationship-Management-Imple¬men¬
tierungen. Deutscher Universitäts-Verlag, Wiesbaden.
Hampe, F. & Schwabe, G., 2002 (in German). Mobiles
Customer Relationship Management. In Reichwald, R.
(Ed.), Mobile Kommunikation, Gabler, Wiesbaden.
Hartel, M.; Bulander, R., 2006: Success Factors of mobile
CRM Projekts - An Overview. In: Proceedings of the
International Conference on E-Business (ICE-B 2006).
INSTICC Press, Setúbal, Portugal, pages 306-311.

418
POSTERS
 A STUDY OF INNOVATION DIFFUSION OF ELECTRONIC
PATIENT RECORDS FOR SUPPORTING MEDICAL PRACTICE

Vincent Cho and Geoffrey Lieu

Department of Management and Marketing, The Hong Kong Polytechnic University, Hung Hom, Hong Kong
[email protected], [email protected]

Keywords: Innovation diffusion, electronic patient record.

Abstract: This paper proposes a study on the underlying factors affecting the adoption, routinization and infusion of
electronic patient record in the clinics of Hong Kong. We suggest using a focus group to identify the
potential antecedents for the three stages of innovation diffusion (adoption, routinization and infusion).
Then a theoretical framework based on the antecedents and their impact on innovation diffusion will be
layout. It will be verified upon a survey sending to the medical practitioners in Hong Kong.

1 INTRODUCTION the Medical Group Management Association in US,

one in five said they were using an EPR and 40
We are well into the digital information age. Digital percent of those without one told they plan to
communications and information resources affect acquire the technology within the next two year.
almost every aspect of our lives – business, finance, Clearly, the EPR is of growing importance for many
education, government and entertainment. Clinical physician practices.
practice is highly information intensive, but it is one In face of adoption barriers, there has been much
of the few areas of our society where computer research outlining the healthcare system’s move
access to information has had only limited success. towards EPRs for example (Ross and Lin, 2003;
Most IT practices in health care by physicians Tachinardi et al. 2001; Van’t Riet et al., 2001).
have been applied to office management in areas However, most of these studies are US based. This
related to accounting of the business, and the study attempts to investigate the facilitating and
scheduling on patients’ booking. The adoption of inhibiting factors that affect health care practitioners
Electronic Patient Records (EPRs) - medical to adopt EPRs in Hong Kong; to understand the
computerized systems that organize the information health care practitioners on their attitude and
on a patient’s treatment, diagnosis and results from knowledge towards EPRs in health care practice and
laboratory and other testing – appears limited. The to explore the existing utilization and future
possibility of instant, universal access to up-to-the- intention on EPRs in private sector of the health care
minute, accurate patient information is a goal that is industry.
actively sought throughout health services
organizations. It is increasingly recognized that
EPRs bring along the quality benefits of electronic 2 THEORETICAL FRAMEWORK
documentation and viewing, prescription and test
ordering, care management reminders, and To predict acceptance of technology, a number of
messaging, among other medical systems. Thus, intention-based theories have evolved, i.e. the theory
EPRs are important tools for improving patient of planned behavior (Ajzen, 1991), theory of
safety and quality of care, especially by promoting reasoned action (Fishbein and Ajzen, 1975) and the
the practice of evidence-based medicine. technology acceptance model (Davis, 1989; Davis et
Despite this potential for quality improvement, al. 1989). According to these theories, user beliefs
however, few physician practices use EPRs. and attitudes about IT influence adoption and usage
Nevertheless, interest in EPRs is substantial. A behaviours. With few exceptions, however, most
recent survey in 2005 indicated that among 1061 studies using these theories have ignored the
respondents to a random sampling of members of temporal dimension and the antecedent variables

421
ICE-B 2008 - International Conference on e-Business

that may affect beliefs and attitudes at different • Initiation – Analyzing organizational needs
stages of the adoption process. Those that have, and potential IT solutions
stress its importance and the need for further study. • Adoption – Negotiating to get
For example, Venkatesh and Davis (2000) report organizational backing for IT
that the same variables had different effects at implementation
different stages of the adoption process, and • Adaptation – Developing, installing and
Fichman and Kemerer (1999) emphasize the need to maintaining the IT application,
capture the time of deployment instead of, or in revising/developing organizational
addition to, time of acquisition as the bases for procedures, training of end-users
diffusion modeling, driven the observed pattern of • Acceptance – Inducing the organizational
cumulative adoptions varies depending on which members to use the technology
event in the assimilation process (i.e. acquisition or • Routinization – Encouraging the use of the
deployment) is treated as the adoption event. IT application as a normal activity
Further, Agarwal and Prasad (1997) support this • Infusion – Effective use of the technology
view that intention-based models may not explain results in the intended benefits (increased
user adoption behavior at the different stages of the organizational effectiveness) of the IT
adoption process. being obtained.
Based on this evidence, the current study
considers the Rogers’ (1995) stage-based diffusion Initiation, adoption and adaptation require both
of innovation model to be the most appropriate to managerial and end-user input and buy-in, and the
guide its investigation of the formation and change remaining three stages require necessary dialogues
over time of user attitudes and subsequent between organizational members for progression
acquisition and deployment decisions. through each stage to occur. Additionally, it is
Everett Rogers defines diffusion as “the process widely recognized that successful implementation
by which an innovation is communicated through depends upon gaining organizational members’,
certain channels over time among the members of a targeted as end-users of the innovation, appropriate
social system” (Rogers, 1983, p.5) where innovation and committed use of an innovation (Leonard-
has been described as an idea, material, or artifact Barton and Deschamps 1988; Klein and Sorra 1996).
perceived to be new by the relevant unit of adoption It is through the development of a critical mass of
(Zaltman, Duncan, and Holbek, 1973). There are individual routinization and infusion that eventual
two types of communication channels have been organizational infusion of an innovation is achieved
influential in diffusing technology – mass media (Tornatzky and Fleischer 1990; Klein and Sorra
channels and interpersonal channels. Mass media are 1996), and organizational benefits might then be
radio, television, newspapers, and so on, which obtained.
enable a source of one or a few individuals to reach Based on the situation in Hong Kong’s clinical
an audience of many. And interpersonal channels are practices, most private clinics are either solo
face to face, telephone, and personal networks. In his practices or partnerships of a few medical doctors
review of innovation diffusion, Rogers (1995) that are small in size. Thus the respective process on
reported mass media channels were most influential EPR initiation, adaptation and acceptance are rather
in introducing potential adopters to an innovation, straight forward. In this regard, we would like to
whereas interpersonal channels were more focus our study on the other three different stages:
influential in subsequent stages. adoption, routinization and infusion on EPRs in
Innovation diffusion research postulates that supporting medical practice.
many different outcomes are of interest in The measure on adoption is based on whether
technology adoption, including the initial adoption, the organization has implemented any EPR.
the subsequent routinization and infusion of the Routinization is measured by the usage of the EPR
innovation. This view is consistent with the stage according to the daily tasks of a clinic. Infusion is
model as proposed and empirically validated by measured by the extent of the EPR being integrated
Cooper and Zmud (1990). These stages of with other internal systems within the clinic or
implementation (as shown below) are not external systems outside the clinic. Moreover, the
necessarily sequential, and should be considered antecedents on these three essential stages will be
activities that may occur in parallel (Cooper & identified.
Zmud 1990):

422
 A STUDY OF INNOVATION DIFFUSION OF ELECTRONIC PATIENT RECORDS FOR SUPPORTING MEDICAL
PRACTICE

3 DATA COLLECTION 4 METHODOLOGY

A trivial approach to study different stages of a We shall obtain the means, standard deviations, and
clinic would be longitudinal tracing, but this takes bivariate correlations for all data used to analyze
years on the data collection. Nevertheless, we predictions of all variables. We intend to perform a
assume different clinics have different extents on the factor analysis on the reasons for adopting new
diffusion stages. Some earlier adopters would be technologies as well as those for not adopting these
more focused on infusion at current moment while technologies. A multiple ordinary least squares
some late adopters are still struggling with the (OLS) regression analysis will be the primary
routinization of the EPR practice. Thus a cross statistical technique to be employed in our study. We
sectional approach would be adequate for shall control for complementarities in the variables
understanding on the diffusion situation of EPRs in and also check whether control variables will have
Hong Kong. Moreover, we suppose most clinics are any significant influence on the data.
not purely on a single stage of the EPR diffusion,
they would be 70% adopting on the EPR, with 50%
routnizing the EPR practice and 5% infusing the 5 CONCLUSIONS
EPR with their daily tasks. In this regard, cross-
sectional approach would make more sense to have With our results on the survey, we will understand
the overall picture. the facilitating and inhibiting factors as well as the
Data for this study will be firstly gathered in a existing practice of EPRs in Hong Kong private
focus group interview from which the possible clinic. This study will establish a theoretical
antecedents and measurements on adoption, examination on the diffusion model and inject the
routinization, and infusion will be determined. A managerial insight on how to utilize the EPRs to a
structured questionnaire will be constructed based greater extent.
on the literature with amendments from the focus
group to fit the Hong Kong medical practice. The
questionnaire will be pilot tested, revised if
necessary, and then sent extensively by mail to ACKNOWLEDGEMENTS
private sector doctors. From the survey, the
respondent (the principle doctor who hosts the This research was supported in part by The Hong
clinic) will be asked about the current usage on EPR Kong Polytechnic University under grant number A-
which consists of the three dimensions on the extent PA7X.
of adoption, routinization, and infusion. The
respondent will then evaluate the importance of
some pre-defined antecedents on the three respective REFERENCES
dimensions of usage. Some open-ended questions
will be supplemented any other antecedents not to be Agarwal, R., Prasad, J., 1997. The role of innovation
included. characteristics and perceived voluntariness in the
The Hong Kong Medical Association (HKMA) acceptance of information technologies, Decision
has available to the public on the Internet a directory Sciences, 28, 3, 557 – 582.
of its members by clinical specialty and geographic Ajzen, I., 1991. The Theory of Planned Behavior,
location. This list includes essentially all registered Organizational Behavior and Human Decision
private practice doctors in Hong Kong, estimated at Processes, 50, 2, 179-211.
Cooper, M., 1991. Failure time analysis of office system
about 4,000 in total. use, Journal of the American Society for Information
As the data collection instrument will be a Science, 42, 9, 644 – 656.
mailed questionnaire, we will randomly sample Cooper, R.B., Zmud, R.W., 1990. Information Technology
2,000 doctors listed in the HKMA directory. For the Implementation Research: A Technological Diffusion
replied respondent, we will mail a $50 Park’N shop Approach, Management Science, 36, 2, 123-139.
couple (for the first 400 respondents) and a summary Davis, F.D., 1989. Perceived Usefulness, Perceived Ease
of our finding as a reward. Of Use, And User Acceptance, MIS Quarterly, 13, 3,
319-339.
Davis, F.D., Bagozzi, R.P., Warshaw, P.R., 1989. User
Acceptance Of Computer Technology: A Comparison
Of Two Theoretical Models, Management Science, 35,
8, 982-1002.

423
ICE-B 2008 - International Conference on e-Business

Fishbein, M., Ajzen, I., 1975. Belief, Attitude, Intention

and Behavior: An Introduction to Theory and
Research. Reading, MA: Addison-Wesley.
Klein, K.J., Sorra, J.S., 1996. The Challenge of Innovation
Implementation, Academy of Management Review, 21,
4, 1055-1080.
Leonard-Barton, D., Deschamps, I., 1988. Managerial
Influence in the Implementation of New Technology,
Management Science, 34, 10, 1252-1265.
Rogers, E. M., 1995. Diffusion of Innovations, 4 ed. The
Free Press: New York.
Ross, S.E., Lin, C.T., 2003. The effects of promoting
patient access to medical records: A review, Journal
of American Medical Informatics, 10, 2, 129– 138.
Tachinardi, U., de SaRebelo, M., de Magalhaes Oliveira,
P.P., Jr., Pilon, P.E., 2001. Real time information from
bedside monitors as part of a webbased patient record,
in: Proc. AMIA Symposium, 691–695.
Van’t Riet, A., Berg, M., Hiddema, F., Sol, K., 2001.
Meeting patients’ needs with patient information
systems: Potential benefits of qualitative research
methods, International Journal of Medical Informatics
64(1) (2001) 1–14.

424
 MANAGEMENT INFORMATION SYSTEMS IN ROMANIAN
UNIVERSITIES

Ana-Ramona Bologa, Razvan Bologa

Computer Science Department, Academy of Economic Studies, Piata Romana 6, Bucharest, Romania
[email protected], [email protected]

Gheorghe Sabau, Mihaela Muntean

Computer Science Department, Academy of Economic Studies, Piata Romana 6, Bucharest, Romania
[email protected], [email protected]

Keywords: The paper Romanian universities, management information systems, ERP systems, integrated information
solutions.

Abstract: This paper shortly presents the situation of the Romanian universities regarding information systems
implementation and deployment. The information presented is the result of a study regarding the current
state of the Romanian universities in the process of data and information system integration, performed at
the end of 2007 in 35 accredited universities. This study was used as a base for identifying and analyzing
the main factors of influence for developing an integrated university environment and for identifying
concrete action directions for accomplishing that integration.

1 INTRODUCTION 2 PROBLEM FORMULATION

AND GENERAL CONCEPTS
The implementation of an information system
dedicated to the university management is nowadays Identifying the most important specific aspects
a fundamental option for the greatest majority of the related to the implementation of university
universities that understood the new trends at governance systems in Romanian, and generally
international level. A higher-education information speaking Eastern European, universities is the focus
system can be used as a vehicle for professionalizing of our current research. Local universities have
and transforming the traditional universities and for different process from their Western counterparts
developing an integrated and standardized Romanian and that is why implementations of solutions
higher education environment, in the European developed in Western countries had limited success.
context. The incompatibility with international solutions
The present paper will present the results of a has led to a situation where local universities are
first phase of a national research project, “Integrated using a high number of various small and poor
Information Solutions for Competitive Management information systems developed in-house or by small
in Romanian Universities”. The objective of this local companies. Almost every faculty or department
first phase was to perform a study of Romanian has its own software applications, developed in-
universities in order to obtain a complete view over house, applications that use various operation
the present situation of existing management systems, tools, databases and protocols. Those
information systems, their integration level and the applications are managed by that specific
problems faced in using those systems. faculty/department and there is no integrated view
The following phases will analyse the solutions over the activities developed in the university.
that foreign universities apply, and finally make a Our team has achieved a study about the current
pilot implementation in the Academy of Economic state of the Romanian universities in the process of
Studies of Bucharest. data and information system integration.
This study is going to be used as a base for
identifying and analyzing the main factors of

425
ICE-B 2008 - International Conference on e-Business

influence for developing an integrated university Education in a first attempt of covering the gaps the
environment and for establishing concrete action Romanian universities presented in the field of
directions for accomplishing that integration. applications for institution management. It was
We used as a starting point the list of all developed by a Romanian software company and
Romanian universities, offered by the Ministry of proposed three modules: Financial, Research and
Education (Romanian universities ), which consisted Payments. This solution was implemented in about
of 56 accredited state universities, 32 accredited 40 public universities and some of them are still
private universities and 24 temporary accredited using it.
private universities. University Management System (UMS) should be
an integrated information system dedicated to higher
education institutions (public or private). Developed
3 APPLICATION INTEGRATION and provided by a Romanian company, Redpoint SA
LEVELS since 2004, UMS ensures the management of
academic processes, students, academic staff, tuition
fees, admission process, graduation process,
Through our research we will use the conventional
three-tier approach. When trying to connect to a scholarships, lodging and accommodation, diplomas,
certain application, its architecture must be etc.
considered. Most applications have a three layer Naum Consult System was developed by Naum
structured architecture (Microsoft - Integration Consult SRL and has been provided since 2003. The
Patterns): Presentation level – is the level that system has modules for: accounting, provision
displays the information for the final user and allows management, assets management, student
him to input data; Business logic level – contains the management, HR, financial.
business functions that action on business data; Data GESCO 2001 should be an information system for
level – accomplishes the persistent data storing in
the management of higher education institutions
data stores. This level is also called resource level.
schooling process developed by a Romanian
Similarly, there are three connection ways
between applications and integration level company, Genisoft Group SA. The system offers
(Microsoft - Integration Patterns ): Presentation many facilities such as: student performances
level integration – the integration level can extract management, budget assessment at department,
information from the user presentation level using a course and allows a correlation with data from
technique called “screen scraping”; Function level SICOB accounting module.
integration – the interaction between the integration
level and the business logic level is accomplished by
application or service interfaces; Data level 5 STUDY RESULTS ANALYSIS
integration – the integration level can move data to
and from data level.
5.1 Implemented Solutions Analysis
The study was conducted on a number of 35 of the
4 SOFTWARE SOLUTIONS Romanian universities, including both public and
FOR HIGHER EDUCATION private institutions and the full range of Carnegie
MANAGEMENT ON segments. This can be considered a representative
sample of Romanian universities, encountering a
ROMANIAN MARKET percent of 31.25% of total accredited universities.
This study addressed a number of questions such as:
The first phase of our research was to gather data
about the current commercial systems used by - What kinds of information systems were developed
universities. As expected, these were marginal within Romanian universities and for what
applications developed by local companies with a functional areas?
limited presence. Our field studies revealed that the - Do those information systems succeed in fulfilling
number of players on the market was reduced, the the information requests of Romanian universities?
following being the most representative: - Which were the major suppliers of software
solutions for Higher Education Management on
SICOB (Information System for Public
Romanian market?
Organization Management) should be a result of a
Almost all universities have a financial solution,
project financed in 2001 by the Ministry of
and Sicob was the major vendor for the financial

426
 MANAGEMENT INFORMATION SYSTEMS IN ROMANIAN UNIVERSITIES

module, encountering 36% of total number of 5.2 Integration Level

implementations, while Naum Consult follows with
a percent of 12%. Anyway, most universities use A real process level integration is presented only by
self-developed solutions for financial and accounting University Management System solution. As its
management activities. developer, the company Redpoint, initiated
Most universities felt the need to implement a collaboration with SAP corporation in order to
Students management module, so that 77% of develop the following modules: financial
universities currently use such a module. 41% of accounting, management accounting, payroll, virtual
those preferred the UMS solution, 11% chose Gesco, library, this solution has a big chance to become a
11% chose Naum Consalt and there is an important complete and competitive software package for
percent of 37% that still use self developed university management. In Naum Consult we have a
solutions. Although the Student module it wasn’t partly data level integration by using a common
very popular two or three years ago, the universities database in several different modules. SICOB, on
have caught up and now most of them use a Student the other side, is a simple accounting solution, so the
solution, even if not a professional one. integration can be made only external, by another
Self-developed solutions are very popular for application or support platform. As for the university
Financial and HR/Payrolls modules, as they were in-house solutions, there the integration is almost
preferred in order to make cost savings but also to inexistent. The modules doesn’t communicate
cover the incomplete functionality of the existing between them, at most they are exported from one
software. Another important remark is that big module, processes and transformed and then
universities use all the HE ERP modules, imported into other modules.
commercial or self developed. Small universities, on
the other side don’t use commercial solutions, 5.3 The Example of the Academy of
because of their cost. Public universities always use Economic Studies, Bucharest
Financial and HR/Payrolls modules, regardless of
developer, as they have to make mandatory specific A preliminary analysis of the information system in
reports. Our study indicates that the number of our university, the Academy of Economic Studies,
players on the HE ERPs market is small. Most revealed the existence of 7 major independent
universities use in-house or low-end solutions. The information systems, which are independent and
chart 5.1 presents a comparison between the main communicate with each other only by export files or
four software solutions used in Romanian specially developed transfer programs. Here is a
universities. short description of those systems and their
relationships with each other.
Table 5.1: University application analysis by architecture
1. Student Management Application is a client-
and implementation technologies.
server application that uses a Microsoft SQL Server
Gesco UMS Sicob 2005 database. Its interface was developed in Visual
Operating Windows Sun Server Fox Pro and Visual Basic. It includes three modules:
Windows
system 2003, Unix, a. Student Master Data: includes student
2000 Server
Linux information, faculties, degrees and programs
Database Oracle Progress information, their curricula, courses etc. It uses a
SQL Server SQL Server Workgroup specially developed transfer program in order to
MySQL DB extract information from Admission Management
Development Java Progress Application.
Microsoft
tools J2EE, Actuate b. Student Academic History: includes a list of all
.NET
SWING, Developer grades up to and including current term, earned
(ASP, Web,
JSP, Struts, credit points, reprogrammed exams. The secretaries
XML, C)
EJB are responsible for updates.
Reporting Embedded, Embedded, c. Student Bills: includes all kind of student fees,
and analysis based on integrated
charges, payments. Data are updated by the pay
XML, self with MS
offices (approximately 10 workstations).
transfer in Office
.pdf or .xls 2. Student Admission Management Application
uses as inputs data regarding student personal data,
their admission options, faculties and
specializations, student photos etc. It provides basic

427
ICE-B 2008 - International Conference on e-Business

data for the Student Management Application and it 6 CONCLUSIONS

uses outputs from Student Bills module, in order to
check out tuition fees payment. This application was The Romanian learning system was recently
developed entirely in Visual Fox Pro. integrated, at least at a formal level, in the European
3. The Academy Web Site: centralizes all the learning system, automatically bringing the need of
relevant information for various visitors category, adapting to European practices (e.g. ECTS).
students, inside and outside the university. Data are The capacity of Romanian universities
not automatically displayed by interrogating a information systems to respond to the demands and
database, but manually filled in by secretaries and the challenges of European education system was
then converted to HTML format. As a result, there is very modest until now. Universities should realize
a high incidence of human errors in data and many that they are not so drastically different and given
out of date or incomplete on-line information. that, they should collaborate in order to influence
vendors to become more sensitive to higher
4. Social Information System was developed in
education needs. This way, ERP vendors should
Visual Fox Pro and includes four modules: Lodging
provide some best practice models to reduce the
and accommodations-student lodging on university
costs of ERP implementation.
campus; Lodging Fees; Scholarships and financial
European integration of Romanian higher
aids; Transportation reimbursement-public
education system might benefit from the support of
transportation expenses reimbursement. It imports
European funds. But, in order to access the
data from the Student Management Application.
European funds, the universities should manifest
interest, make congruent efforts and put some
pressure on the decision factors, on one side, and, on
the other side, the political support is also very
important. For the moment none of them seems to be
strong enough.

REFERENCES
Ana-Ramona Bologa (Lupu), ERP for Romanian Higher
Education, the 8th Economic Informatics Conference,
“Informatics in Knowledge Society”, ASE, Bucharest,
p. 205-210, May 2007, ISBN 978-973-594-921-1;
Marti Harris, Michael Zastrocky, Jan-Martin Lowendahl ,
Magic Quadrant for Higher Education Administrative
Suites 2006, Gartner Industry Research 16 September
2006;
Ion Lungu, Ana-Ramona Bologa, Adela Bâra, Vlad
Figure 5.1 Communication flows in the applications of the Diaconiţa – Integrarea sistemelor informatice, Editura
Academy of Economic Studies. ASE, Bucharest, 2008;
Microsoft - Integration Patterns http://download.
5. SICOB is used for financial and accountability microsoft.com/download/a/c/f/acf079ca-670e-4942-8a
53-e587a0959d75/IntPatt.pdf;
specific operation.
The list of accredited Romanian universities, http://
6. Human Resources and Payrolls Application www.edu.ro;
uses Microsoft SQL Server database and includes 2 Ana-Ramona Lupu, Razvan Bologa, Gheorghe Sabau,
specific modules: Payrolls and. Human resources. Mihaela Muntean, The Romanian Universities in the
Process of Data and Information, in The Proceedings
7. Library for Universal (L4U): a classic system of The 7th WSEAS International Conference on
for library management that uses data exported from Artificial Intelligence, Knowledge Engineering and
Student Management Application and personnel Data Bases (AIKED'08), Feb. 2008, Cambridge, U.K.,
database, after a previous filtering and p 527-532.
transformation by a specially designed program.
There is also Web-based Virtual Campus
management application for distance learning
programs.

428
 A FEATURE EXTRACTING METHOD FOR TAMPER
DETECTION IN PRINTED DOCUMENTS

Yoshiyasu Takahashi, Takaaki Yamada and Seiichi Susaki

Systems Development Laboratory, Hitachi ltd., Totsuka-ku Yoshida-cho 292, Yokohama, Kanagawa, Japan
{yoshiyasu.takahashi.gq, takaaki.yamada.tr, seiichi.susaki.gw}@hitachi.com

Keywords: Feature Extracting Method, Dot-Pattern Code, Digital Watermarking, Electronic Delivery, e-Government.

Abstract: In this paper, we propose our feature extracting method for tamper detection in printed documents. To
detect the tamper of the printed document, a feature extracting method is needed. In this paper, we describe
our feature extracting method. Our feature extracting method is based on the location of the mean point of
each dot. We have estimated our feature extracting method’s probability of the collision, its uniformity of
the distribution, its invariability during D/A and A/D transform and its invariability during the ordinal
change of paper. We have found that our feature extracting method can extract desirable feature value.

1 INTRODUCTION In this paper, we describe the requirements for

the feature extracting method and introduce our
Since 2000, the Japanese Government has promoted innovative method to extract the paper’s feature. We
the IT strategy along with the implementation of also make estimation of our method and compare
various structural reforms. (IT Strategic with another extracting method.
Headquarters, 2007).
However, several issues still remain to solve.
One of the issues is the lower utilization rate of e- 2 REQUIREMENTS FOR THE
government. FEATURE EXTRACTING
We consider the reasons of the lower utilization
rate are three; complicated preparation for online METHOD
application, security worry and the lack of the means
of the online delivery. For the feature extracting, the requirements are
To resolve the third reason, we can envisage a following five.
system that the government delivers certificates 1) The feature values calculated from two
online and the user also receives them online, and different documents should not be equal
the user prints them with their home printers and use 2) The distribution of the feature values should
or submit them to the destination in paper format. be uniform.
To bring the scenario to fruition, the two means 3) The document should not be calculated from
are needed; a means to prove that the paper is its feature value.
original print and not copy, and a means to prove 4) The feature values should not vary during the
that the content of the paper is correct and not D/A and A/D transform.
tampered. 5) The feature values should not vary if the paper
To detect the copy, the “copy indicator” document is folded or got wrinkles or tainted.
technique is widely known. For the feature extracting, the problem is more
To detect the tamper of the paper, the extracting difficult than the hash values because of the
method of the paper’s feature is needed. With that requirement 4 and 5. Because the D/A and A/D
method, we can extract the feature of the printing transform is inevitable, the feature value should not
document and print it within the paper. After the vary during them. Moreover, because paper is used,
paper is distributed, the verifier can extract both the folding, wrinkles or taint easily occur, the feature
feature of the paper and the feature embedded, and value should keep the same value during the ordinal
compare them. However, the extracting method of alteration of the document. At the same time, if the
the paper’s feature is not well known. significant change occurs in the document, even if

429
ICE-B 2008 - International Conference on e-Business

embed and print

Tax Tax
123,456 yen (1.2, 3.4)
123,456 yen
(5.6, 7.8)
calculate ・・ ・
median points data dot-pattern code

image correction median points data of the paper

tampered? (2.5, 6.3)
and calculate (4.8, 7.1) automatic verification
・・ ・
Tax
654,321 yen scan (1.2, 3.4)
tampered!
(5.6, 7.8)
read
・・ ・
the embedded code embedded median points data

Figure 1: Overview of the verification.

the change is minute, the feature value should watermarking technique for printed documents by
change the other values to represent the change of superposing dot pattern blocks on backgrounds of
the document. the document image (Suzaki, 2003).

3 RELATED WORKS 4 PROPOSING THE FEATURE

EXTRACTING METHOD
The ideal feature value has not been found yet.
However, several techniques that meet a part of the 4.1 Brief Overview
requirements are studied to extract the paper’s
feature to detect the tampering. Figure 1 shows the overview of our document
There are three types of the paper’s feature verification system.
below. To print the certificate, the following procedures
1) using the document itself are performed.
2) using the semi-fragile watermarking 1) The issuer makes the certificate.
3) using the feature of the paper 2) The system calculates the feature value from
First, there are methods to use (a part of) the the certificate, and embeds it into the certificate
original document, such as the text data or the image itself. The concrete means of extracting and
data. To embed them, 2D-code is usually used. To embedding of the feature value are described later.
check the document’s tamper, usually a man power 3) The system prints the feature value embedded
is needed; a verifier scans the code and extract the certificate with printers. In the electronic delivery,
document’s text data or image data, and compare this process can be performed at the user’s home.
them, watching side-by-side. Its cons are its easiness After the issue, because the certificate circulates
to implement and its robustness for the paper’s among many people, there is a risk of tampering the
conditions, and its pros are that it needs a verifier’s certificate. Therefore, the submitted certificate needs
help and time to verify. to be verified. With the following procedures, the
Second, the use of semi-fragile watermark verifier can verify the certificate.
method has been proposed. The semi-fragile 1) The verifier scans the certificate and detect the
watermark is a watermark which is robust to some embedded feature value.
degradation such as compression, while it is at the 2) The verifier also calculates the feature value
same time destroyed if the embedded image is from the submitted certificate.
tampered with. However, the semi-fragile watermark 3) The verifier compares the two feature values.
which is robust to the paper’s folding or wrinkles or When the significant difference is detected, the
tainting is not well-known. Therefore, workable verifier judges the certificate as tampered. The
proposal of this type of method has not been yet. concrete means of extracting and embedding the
Third, using the feature of the paper has been feature value are described below.
proposed. Suzaki and et al. has proposed the

430
 A FEATURE EXTRACTING METHOD FOR TAMPER DETECTION IN PRINTED DOCUMENTS

4.2 Extracting the Feature Value number of Times New Roman font. We drew a
character one by one on the white image and
In our method, we use the median point as the calculated its median point.
feature value of the document, as shown in Figure 2. The result is shown in Figure 3. In Figure 3, the
In the Figure 2, the plus sign illustrates the median median points of each character are plotted with dots.
point of the character “4”. From the Figure 3, though there is some bias, we can
find that the median points are distributed almost
uniformly. Moreover, there are no dots which collide
against another dot.
To compare, we also calculated the area of each
alphabet and number. In Figure 4, the histogram of
the result is shown. The x-axis is the area and the y-
axis is occurrence. From Figure 4, we can find the
area is distributed not uniformly. Indeed, nine
Figure 2: Sample of the feature value. characters collided; i.e. they have the same area.
Therefore we can conclude that our feature value is
The benefits of the median point as the feature better than the area.
value are the robustness against the blur because of
the scattering of the tonner or ink of printers and of 15

the focus error of scanners. Figure 2 also illustrates 10

this merit. In Figure 2 left, the character “4” is not
blurred. This illustrates the character “4” to calculate 5
the feature value in issuer’s computer. After the
printing, circulating, and scanning, the character 0
y

-15 -10 -5 0 5 10 15
may blur. The Figure 2 right illustrates this situation.
-5
Although the blur, the median point of the
character “4” of the Figure 2 right does not vary -10
very much. That is because the width of the line of
the character is almost the same. -15
To embed the feature value into the Document, x

we use the dot-pattern code (Takahashi 2007, 2008),

which the authors have developed before. The dot- Figure 3: Median points of a character.
pattern code is the information embedding method
onto the paper with minute dots.
14
12
10
5 ESTIMATION
occurrence

8
6
5.1 Assessment Viewpoints of the 4
Feature Value 2
0
The viewpoints of the assessment are following five.
(A)probability of the collision
6

5
8

2
18

79
7.

9.

1.

3.
30

42

55

67

(B) uniformity of distribution

(C) probability of inverse calculation Figure 4: Area of a character.
(D)invariability during D/A and A/D transform
(E) invariability during the ordinal change of 5.3 Invariability during D/A and A/D
paper
Transform
5.2 Probability of the Collision and To estimate the invariability during D/A and A/D
Uniformity of Distribution transform, we viewed the difference of the median
point’s change between before and after printing the
To estimate the probability of the collision and the document.
uniformity of distribution, we calculated the median
points of characters. The characters are alphabet and

431
ICE-B 2008 - International Conference on e-Business

We calculate the median point and area of a mean point of each dot, and is expected to be
character “a” in digital format. Then, we print it and applicable to home printers such as inkjets.
scanned and calculated the median point and area We have estimated the probability of the
again from the scanned image. Before calculating collision and the uniformity of the distribution of our
the median point and area, we converted the scanned feature extracting method. We have found that the
image to binary image with various thresholds. feature value extracted from one character is
distributed uniformly and do not collide each other.
Table 1: Invariability during D/A and A/D transform. We have also checked invariability during D/A
and A/D transform and we have found that almost
thrs median point diff area diff every character can be distinguished with our feature
192 (149.12, 146.86) 2.33 260 128 extracting method even after D/A and A/D transform.
128 (150.28, 148.44) 0.58 186 54 We have also found that our feature extracting
64 (150.02, 148.22) 0.76 152 20 method is better than the area.
32 (150.49, 148.31) 0.78 116 16 We have also checked the invariability during
(150.41, 148.89) the ordinal change of paper and found that our
16 0.34 108 24
method has enough and better invariability than the
8 (150.46, 148.75) 0.44 104 28
area.
orig (150.08, 148.98) --- 132 --- Therefore, we can conclude our proposed
The result is shown in Table 1. From Table 1, we method can extract desirable feature value.
can find that though the threshold varies (thrs), the
median point does not vary very much. On the other
hand, the area varies very much and the difference REFERENCES
of the area of the original image. There are twelve
characters whose area’s difference is less than 54. IT Strategic Headquarters, 2007. Priority Policy Program
Therefore we cannot distinguish these twelve 2007. The Japan Government. Japan.
characters with area. Ching-Yung Lin, Shih-Fu Chang, 2000. Semi-Fragile
Watermarking for Authenticating JPEG Visual
Content. In SPIE Security and Watermarking of
5.4 Invariability during the Ordinal Multimedia Content II, pp.140-151.
Change of Paper Ke DING, Chen HE, Ling-ge JIANG, Hong-xia WANG,
2005. Wavelet-Based Semi-Fragile Watermarking
To estimate the invariability during the ordinal with Tamper Detection. In IEICE Trans. on
change of paper, we viewed the difference of the Fundamentals of Electronics, Communications and
median points’ change between before and after Computer Sciences, Vol.E88-A, No.3, pp.787-790.
folding of the paper. The experiment is done as 5.3. Suzaki M., Mitsui Y., Suto M., 2003. New Alteration
Detecting Technique for Printed Documents using Dot
Pattern Watermarking. In SPIE, Volume 5020, pp.
Table 2: Invariability during the ordinal change of paper.
665-676.
thrs median point diff area diff Takahashi Y., Yamada T., Ebisawa R., Fujii Y., Tezuka S.,
2007. Research and Development of Dot Pattern
192 (146.03, 144.40) 6.11 296 164 Technology for Commodity Printers. In Proc. of FIT
128 (150.59, 148.79) 0.53 164 32 2007, the Forum on Information Technology, pp. 325-
64 (150.51, 148.59) 0.57 59 73 326.
Takahashi Y., Yamada T., Ebisawa R., Fujii Y., Tezuka S.,
orig (150.08, 148.98) --- 132 ---
2008. Information Embedding Method for Home
The result is shown in Table 2. In this Printing of Certifications. In Proc. of ICACT2008, the
experiment, our median point shows near to original 10th International Conference on Advanced
Communication Technologies, pp.2116-2120.
median point. Therefore we can conclude that our Echizen I., 2007. Digital Watermarking Technique and its
proposed method has good invariability during the Application. In IPSJ Magazine Vol.47 No.11, pp.1243-
folding of the paper. 1249.

6 CONCLUSIONS
In this paper, we have introduced our feature
extracting method of paper document. Our feature
extracting method is based on the location of the

432
THE AFFECTION OF DEMOGRAPHIC CHARACTERISTICS ON
MEDIA CHOICE
A Case about Chinese Miniature Automobile Consumers

Dao-ping Chen
School of Economy and Business Administration, Chongqing University, Chongqing, P.R. China
School of Economics and Management, Chongqing Normal University, Chongqing, P.R. China
[email protected]

Wei Liu
School of Economy and Business Administration, Chongqing University, Chongqing, P.R. China
[email protected]

Keywords: Miniature automobile consumers, demographic characteristics, media choice, optimal scaling regression.

Abstract: This paper explores the media choice of Chinese miniature automobile consumers in obtaining purchasing
information about miniature automobile. A worldwide survey involved a majority of Chinese areas was
conducted. The survey focused on consumers' demographic characteristics and media choice. The result of
the survey shows that consumers chiefly choose newspaper, TV or automobile exhibitions as the media to
obtain purchasing information about miniature automobile. Then, based on optimal scaling regression
analysis of statistics, the influences of demographic characteristics on media choice were studied. The study
finds that consumers' demographic characteristics significantly affect their media choice. In detail,
consumers' living city, occupation and education are three most important characteristics that affect the
media choice of consumers.

1 INTRODUCTION demographic characteristics on media choice of

Chinese miniature automobile consumers.
Consumers can obtain purchasing information from
various media such as newspaper, TV, magazine, or
Internet. Many scholars studied the consumers' 2 QUESTIONNAIRE AND
media choice behaviour (Wei and Pan, 1999, Hung, RESPONSE
Gu, and Tse, 2005, Geist, 2004, Ducoffe, 1996,
Yoon and Kim, 2001, Milsom, 2003). These studies Newspaper, magazine, TV, Internet, and automobile
found that consumers which are from different exhibition are selected as the resources from which
countries or purchase different products have Chinese consumers get purchasing information about
different choice to media in obtaining purchasing miniature automobile. Gender, living city,
information. occupation, education, age, and monthly household
There are many factors influencing consumers' income are used to describe demographic
media choice. Demographic characteristics, such as characteristics. Gender has two levels (1=male,
age, education, occupation, and income, are 2=female), living city four levels (1=big city, 2=city
generally regarded as important factors affected the of middling size, 3=county seat, 4=villages and
media choice of consumers (Westbrook and Fornell, towns), occupation seven levels (1=government
1979, Andreasen and Ralchford, 1976, Schaninger servant, 2=employee of national enterprise,
and Sciglimpagiia, 1981, Newman and Staelin, 1973, 3=employee of private enterprise, 4=employer of
Chiteji and Stafford, 1999, Claxton, Fry, and Portis, individual enterprise or partnership enterprise,
1974). This study has the main objective of 5=farmer, 6=professional (lawyer, accountant,
empirically determining the affection of teacher, doctor, athlete, reporter etc.), 7=other),

433
ICE-B 2008 - International Conference on e-Business

education six levels (1=junior high school or below, Table 1: Media choice by consumers (N=2 343).
2=senior high school, 3=technical secondary school, Information Frequency Percent
4=junior college, 5=college or university, sources (%)
6=graduate student), age seven levels (1=18-21years, Newspaper 986 42.1
2=22-25, 3=26-29, 4=30-34, 5=35-39, 6=40-59, Magazine 281 12.0
7=60 or above 60 years), and monthly household TV 863 36.8
income four levels (1=less than ￥2 000, 2=￥2 000 Internet 474 20.2
to less than ￥5 000, 3=￥5 000 to less than ￥8 000, Automobile 559 23.9
4=more than ￥8 000). exhibition
The data used in this study are obtained from a
large survey sponsored by an automobile group 3.2 The Influence of Demographic
company in China. 2 630 questionnaires are sent 263 Characteristics on the Media
dealers of the automobile group through mails, Choice
commencing in December 2003, at three month
intervals. A dealer selected is responsible for 10 We conducted optimal scaling regression analysis.
questionnaires. The objects of the survey are Newspaper, TV, Internet, automobile exhibition, or
consumers who visit the shop of dealers and have magazine is dependent variables and gender, living
intention to purchase miniature automobile. The city, occupation, education, age and monthly
survey areas involve 31 provinces, municipalities household income are independent variables. The
and autonomous regions of Chinese Mainland. A regression result is shown in table 2.
total of 2 623 usable responses are received, yielding
a response rate of 99%. 280 questionnaires in 2 623 3.2.1 Newspaper
are discarded because they are not perfectly filled.
The remaining 2 343 questionnaires are used for the From table 2, we know that the regression model is
final analysis. All data are processed by SPSS 15.0. statistically significant at the 0.01 level (F=3.210,
Sig.<0.01). Of six demographic characteristics, four
characteristics, living city (F=29.476, Sig.<0.01),
3 RESULTS occupation (F=5.929, Sig.<0.01), monthly household
income (F=7.693, Sig.=0.006<0.01), and age
(F=2.901, Sig.=0.021<0.05) have a significant
3.1 The Media Choice of Chinese influence on dependent variable newspaper. Living
Consumers city (62.2%) is the most important characteristics,
followed by monthly household income (20.0%),
The result of the media choice by Chinese miniature occupation (8.2%), and age (5.3%). Gender and
automobile consumers is shown in able 1. The education have not significant affection on
percentage of the media choice breakdown in dependent variable newspaper.
descending order is as follows: newspaper (42.1%),
TV (36.8%), automobile exhibition (23.9%), Internet 3.2.2 Magazine
(20.2%), and magazine (12.0%).The percentage of
newspaper is higher than that of TV, which shows The result of ANOVA shows that the regression
that it is more possible for consumers to obtain model on magazine has statistical significance at the
advertising information from newspaper than from 0.01 level (F=2.600, Sig.<0.01). From table 2, we
TV although Chinese families almost have TV. can see that occupation (F=11.032, Sig.<0.01),
Though Internet has had a rapid development in education (F=19.720, Sig.<0.01) and age (F=4.858,
China in recent years, it is early for Internet to be Sig.<0.01) produce significant influence on
regard as a main medium of consumers for obtaining dependent variable magazine. The three
automotive advertising information. It appears demographic characteristics breakdown in
surprising that the percentage of magazine, as an descending order of importance are as follows:
important traditional medium, is only 12.0%, which education (55.6%), occupation (32.7%), age
may be because Chinese miniature automobile (11.6%).
consumers are not known as chief target of
From table 2, we can know that gender, living city
magazine.
and monthly household income have not significant
affection on choice of consumers on magazine.

434
 THE AFFECTION OF DEMOGRAPHIC CHARACTERISTICS ON MEDIA CHOICE - A Case About Chinese
Miniature Automobile Consumers

Table 2: Results of Regressions of Consumer Characteristics on Media Choice (N=2 343).

Media Characteristics ANOVA Beta df F Sig. Importance
Newspaper Gender .022 1 1.063 .303 .015
Living city .114 2 29.476 .000 .622
F(16, 2 326)
Occupation -.054 6 5.929 .000 .082
=3.210
Education -.018 2 .625 .536 .028
Sig.=0.000
Age .036 4 2.901 .021 .053
Income -.059 1 7.693 .006 .200
Magazine Gender .033 1 2.405 .121 .013
Living city -.020 2 .825 .438 -.024
F(19, 2 323)
Occupation .071 6 11.032 .000 .327
=2.600
Education -.101 5 19.720 .000 .556
Sig.=0.000
Age .046 3 4.858 .002 .116
Income -.008 2 .144 .866 .012
TV Gender -.028 1 1.799 .180 .078
Living city -.078 1 13.747 .000 .416
Occupation F(15, 2 327)
.050 6 5.416 .000 .153
=2.108
Education -.066 4 9.063 .000 .200
Sig.=0.008
Age .027 2 1.721 .179 .071
Income .032 1 2.345 .126 .081
Internet Gender .011 1 .311 .577 -.004
Living city .053 2 6.917 .001 .060
F(23, 2 319)
Occupation .161 6 52.029 .000 .331
=16.133
Education -.221 5 91.927 .000 .524
Sig.=0.000
Age .084 6 18.557 .000 .062
Income -.047 3 5.598 .001 .027
Automobile Gender -.049 1 5.413 .020 .187
exhibition Living city .052 2 5.700 .003 .228
F(16, 2 326)
Occupation .060 6 8.227 .000 .264
=2.479
Education -.054 4 6.099 .000 .228
Sig.=0.001
Age .014 1 .473 .492 .026
Income .047 2 4.972 .007 .068

3.2.3 TV From table 2, we can see that living city (F=6.917,

Sig.=0.001), occupation (F=52.029, Sig.<0.001),
The result of ANOVA of regression model on TV education (F=91.927, Sig.<0.001), age (F=18.557,
shows that the model is statistically significant at the Sig.<0.001) and monthly household income
0.01 level (F=2.108, Sig.=0.008). Living city (F=5.598, Sig.=0.001) produce significant influence
(F=13.747, Sig.<0.001), occupation (F=5.416, on dependent variable Internet Owing to table 3, we
Sig.<0.001) and education (F=9.063, Sig.<0.001) can know that education (52.4%) is the most
produce significant influence on dependent variable important of the five characteristics, followed by
TV. The three characteristics are put in descending occupation (33.1%), age (6.2%), living city (6.0%)
order of importance as follows: living city (41.6%), and monthly household income (2.7%). Gender is
education (20.0%), and occupation (15.3%). only characteristic which is not statistically
Moreover, from table 2 we can know that gender, significant in this case of Internet.
age and monthly household income have not
significant affection on dependent variable TV. 3.2.5 Automobile Exhibition
The result of ANOVA of regression model shows
3.2.4 Internet that the model on automobile exhibition is
statistically significant at the 0.01 level (F=2.479,
The result of ANOVA of regression model on Sig.<0.01). From table 2, we can see that consumers'
Internet shows that the model is statistically gender (F=5.413, Sig.=0.020<0.05), living city
significant at the 0.01 level (F=16.133, Sig.<0.001). (F=5.700, Sig.=0.003<0.01), occupation (F=8.227,

435
ICE-B 2008 - International Conference on e-Business

Sig.<0.001), education (F=6.009, Sig.<0.001) and This study is somewhat limited in the sense that
monthly household income (F=4.972, Sig.=0.007 the sample is confined to consumers entering the
<0.01) produce significant influence on their choice shop of dealer and having intention to purchase
on automobile exhibition. The five characteristics miniature automobile, therefore, our results are only
are ranked in descending order of importance as a first step toward understanding Chinese miniature
follows: occupation (26.4%), education (22.8%), automobile consumers. A more comprehensive
living city (22.8%), gender (18.7%), and monthly survey that focuses on all consumers of automobile
household income (6.8%). Age has not significant in China should be conducted in the future. If so, a
impact on dependent variable automobile exhibition. more representative sample could be obtained. By
the new sample, a contrast could be done on the
media choice between consumers of miniature
4 SUMMARY AND automobile and consumers of other automobile
categories.
CONCLUSIONS
The study focuses on the media choice of Chinese
consumers in obtaining purchasing information REFERENCES
about miniature automobile and on the influence of
Andreasen, AR., Ralchford, BT., 1976. Factors affecting
their demographic characteristics on the media
consumers' use of information sources. Journal of
choice. The results show that Chinese consumers Business Research, 4(3): 197-212.
mainly choose newspaper, TV, and automobile Chiteji, NS., Stafford, FP., 1999. Portfolio choices of
exhibitions to obtain purchasing information about parents and their children as young adults: asset
miniature automobile. Chinese consumers appear not accumulation by African-American families. American
to prefer magazine to gain advertising information Economic Review, 89(2): 377-380.
about miniature automobile. Claxton, JD., Fry, JN., Portis, B., 1974. A taxonomy of
The study finds that the demographic prepurchase information gathering patterns. Journal of
characteristics of Chinese consumers produce a Consumer Research, 1(3): 35-42.
Ducoffe, RH., 1996. Advertising Value and Advertising on
significant impact at the 0.01 level on whether
the Web. Journal of Advertising Research, 36
consumers select newspaper, magazine, TV, (September/ October), 21-35.
Internet, or automobile exhibition as a medium of Geist, LC., 2004. Automakers Stretch for High-impact
obtaining purchasing information. Of all six Magazine Ads. Automotive News, 10/11/2004, Vol. 79
characteristics variable, living city, occupation and Issue 6116, p26B-26E, 3p.
education are the three most important Hung, K., Gu, FF., and Tse, DK., 2005. Improving Media
characteristics which affect the media choice. Decisions in China. Journal of Advertising 34
Gender has not significant influence on almost all (Spring), p49, 15p.
media choice of consumers. Milsom, P., 2003. More Car Buyers Turn to Web for
Research. New Media Age, 8/21/2003, p13, 1/2p, 1c.
The results provide substantial implications for
Newman, JW., Staelin, R., 1973. Information sources of
miniature automobile manufacturers which face durable goods. Journal of Advertising Research, 13(2):
advertisement decision. If an automaker intends to 19-29.
select a medium from various media for advertising, Schaninger, CM., Sciglimpagiia, D., 1981. The influence
newspaper or TV should first be considered. Another of cognitive personality traits and demographics on
implication derivable from this study is that consumer information acquisition. Journal of
manufacturers should conduct a classification on Consumer Research, 8(2): 208-216.
consumers according to their living city, occupation Westbrook, RA., Fornell, C., 1979. Patterns of information
or education characteristics. Manufacturers should source usage among durable goods buyers. Journal of
Marketing Research, 16 (3): 303-312.
integrate several media to publish advertising
Wei, R., Pan, ZD., 1999. Mass Media and Consumerist
information suitable for the consumer categories. For Values in the People's Republic of China.
example, if they consider to use newspaper, TV or International Journal of Public Opinion Research, 11
Internet for advertising, automakers should chiefly (1), 75—97.
focus on the appeals of consumers living in the city Yoon, SJ., Kim, JH., 2001. Is the Internet More Effective
of middling size, or individual-enterprise or Than Traditional Media? Factors Affecting the Choice
partnership-enterprise employer doing pioneering of Media. Journal of Advertising Research, 41
work or consumers with a low education level in (November/December), 53-60.
advertising design of Chinese miniature automobile.

436
 WEB BASED COLLABORATIVE DOCUMENT CREATION AND
REVIEW SYSTEM

Marius Ioan Podean, Raluca Arba

Faculty of Economics and Business Administration
Babes Bolyai University of Cluj Napoca, Teodor Mihali 58-60, Cluj Napoca, Romania
[email protected], [email protected]

Loredana Muresan
Faculty of Economics and Business Administration
Babes Bolyai University of Cluj Napoca, Teodor Mihali 58-60, Cluj Napoca, Romania
[email protected]

Keywords: Collaborative document editing, document management, workflow, virtual teams, XML.

Abstract: An important aspect in distributed teams and organization is their ability to manage documents. A
collaborative document editing system that integrates functionalities from document management systems,
workflow, collaborative editing with support for virtual teams can increase team efficiency and allow users
to concentrate their efforts on content development. This paper reports a case study implementing this
approach in collaboratively creating scientific papers. The use of XML when treating documents proves to
be the appropriate solution to develop user and document centered systems.

1 INTRODUCTION geographically distributed teams (Nallaparaju et al.

2005). Using WMS, the organization can integrate
Creating scientific papers is most of the time a very different software technologies, leading to the
complex and elaborate task. This type of documents improvement of the collaborative activities
usually is written by more then just one person, and (Aversano et al. 2001). This technology leaves
often the authors are located in different parts of the authors with the consumption of a great deal of time.
world. This collaborative process is sometimes Collaborative editing allows multiple persons to
slowed down by the fact that dealing with the edit simultaneously the same document, see who is
technology used to support it is very much time working on the document and watch in real time the
consuming. Taking in consideration the authors need changes that they have made (Raikundalia and
for location and time independence, the usage of Zhang 2004). In order to have sufficient knowledge
different operating systems and applications, at this about the changes that others perform upon the
moment technology offers solutions that satisfy just document, group awareness mechanisms have been
partially the aforementioned requirements. created, such as: telepointers (multiple cursors of
Document management systems (DMS) focus on users appear within the document), radar views,
tracking and storing documents created and multi-user scrollbars and, as shown by Raikundalia
exchanged by there authors (Aversano et al. 2001). and Zhang (2004), structure-based multi-page view,
They provide components for defining metadata for point jumping mechanism and user info list. The
the documents (i.e. date of creation, authors, version main downside of this solution is that the
etc.), indexing (usually based on metadata), storage implementations available are platform-specific and
and retrieval (based on the unique document can generate conflicts between members when
identifier). someone changes often content created by others.
Workflow management systems (WMS) allow The collaboratively edited file is stored on the
the automation of processes within an organization, document owner’s computer, leading towards
enabling greater coordination and control among versioning problems when participating members
make local copies of the document.

437
ICE-B 2008 - International Conference on e-Business

Wikis represent “a piece of software that allows chapters of the document, chapters on which the
users to add, modify, and/or delete information from others could only place comments, content changing
a knowledge base via the web” according to Spek not being allowed, avoiding conflicts and stepping
(2008). As the author of this definition emphasis as a on each others and allowing individuals to reconcile
main characteristic, wikis are anarchistic systems, with the teams goals. The application facilitates
many implementations allowing anonymous users to communication through synchronous and
modify the content. On wiki systems “conflicts can asynchronous channels. Documents being stored in
quickly result in ‘edit wars’ when multiple users XML files and using a web based user interface
keep on reverting each others changes because they makes the system work on different platform and
don’t agree” as shown by Spek (2008). allows users to export documents in different open
Creating research papers requires a system that is formats, representing a combination of best practices
simple to use and allows users to focus their efforts specific for the previous discussed systems. At this
on the content rather that on the technology used to moment the application offers no version control
create it. The system has to support the space capabilities, and therefore authors could not revert
independence of the authors and integrate their documents to older versions.
collaborative effort in a common workplace in order As mentioned earlier, Dante is both a user and
to obtain greater team efficiency (Guerrero et al. document centric system, supporting collaboration
2004). Collaborative work involves information in virtual teams and efficient document
exchange in order to support negotiation and management. As described by Millward and
communication between group members and Kyriakidou (2004), it is important for virtual teams
different mechanisms through which the team can to be a “singular concrete entity” with the following
regulate and manage itself in order to be goal characteristics: stability, regular interaction,
directed (Millward and Kyriakidou 2004). A more symbiosis and member proximity. Following this
supportive system would have document requirements, in Dante teams are organized around
management facilities and support for task the document that they are creating. Member
automation. To achieve greater efficiency the system proximity results from the fact that each member can
will have to be user centered and non–restrictive view the most recent version of the chapters that the
regarding de operating system. others have created and that everybody knows who
In this paper, we start by presenting current is responsible for a particular chapter. Each author
technologies used for document management and can review others work and make suggestions
editing highlighting their main characteristics and related to each piece of text using the commenting
downsides. We will continue in Section 2 with the tools. The symbiosis of the team is supported by the
theoretical approach of our system and then discuses fact that each person’s responsibilities are clearly
in Section 3 the implementation and the technical drawn and all members have the same rights, all
details concerning it. Based on the model we’ve depending upon others in improving their work.
proposed, in Section 4 we shall present some Interaction between members is supported by both
conclusions and further work. synchronous and asynchronous mechanisms.
In collaborative real time editing systems several
users can edit a file using different computers. An
2 THEORETICAL APPROACHES important aspect for this type of systems is group
awareness (GA). GA provides users information
In this paper we shall present our implementation of about the status of a document and changes made by
such a system that aims to cover the aforementioned others. As shown in Raikundalia and Zhang (2004),
requirements. Dante is a web based system designed several GA techniques have been identified:
to be a good support for virtual teams in elaborating telepointers (multiple cursors are shown within the
scientific papers. It offers document management document), radar views, multi-user scrollbars and
facilities and process automation for repetitive tasks. structured-based multi-page view, point jumping
Since all the data the system uses is stored in XML mechanism and user info list. In Dante, the GA
files, Dante can be document and user centered, problem is solved by using a structured-based multi-
allowing authors to easily edit, review and export in page view panel for displaying a project. One of the
different formats their work. Teams are building main downsides of all collaborative editing systems
around the document allowing them to be goal (CES) is that the document is saved on the document
directed; all members of the team having the same owner’s computer, all others participants being
rights. Each author is responsible of editing different allowed to save a copy of the document leading in

438
 WEB BASED COLLABORATIVE DOCUMENT CREATION AND REVIEW SYSTEM

time to lost edits and versioning problems. Dante automation for different parts of the process (i.e.
being a web based system it has the advantage of formatting the document according to different
delivering file storage facilities. All changes made templates)
by the authors are saved on the server which will
As shown in Figure 1, Dante is a web based
offer only the most recent version of the document.
system design upon de model-view-controller design
Authors edit different parts of the document and
pattern which stores all data in XML files. This
send changes to the server via AJAX.
approach allows a great flexibility in handling
As shown in Leone, Hodel and Gall (2005),
document and user interaction. The presentation
combining CES and DMS can result in greater
layer sets the appropriate tools according to user
performance improvements. Dante takes a similar
level and rights and manages the client-side
approach providing solutions and process
communication with the controller layer using
automation for creating, storing retrieving, editing
AJAX.
and exporting documents collaborative, anytime on
The presentation layer defines de following main
the most used platforms. All documents have
tools: Editor, Partners and Chat. The Editor allows
metadata to easily manage them and provide extra
users to manage old projects, start new ones,
search capabilities. Another important aspect is the
collaboratively edit current projects and export
presence on the internet and the collaboration with
documents in different open formats and format the
persons from outside the team using e-prints.
content using predefined templates. When starting a
Lawrence (2001) shows that articles freely available
new project, the project owner must define a set of
online are more highly cited and recommends, in
properties, like name (works like an identifier for the
order to achieve greater impact and faster scientific
project), type (according to the type that has been
progress, that authors should aim to make research
selected, a certain template will be used to format
easy to access. Content can easily be transformed
the content when users export the document) and a
into HTML and allow others to post comments on
list of members that will participate to the project
the article if the team wishes so.
(after defining the project, all participants will have
the same rights) (Leone, Hodel and Gall 2005).
During development extra sets of metadata will be
added to the project allowing users to consult the
state of the project, last modified date etc. Each
author can edit one or more chapters from the
document and place comments on those chapters
edited by the others. Chapters are presented as
elements in a tree menu and can bee accessed in
different windows. When accessing a particular
chapter, the system checks the metadata associated
to it and determines if the current user is the author
of the chapter and displays a new menu that allows
him to edit the content if so, otherwise allowing him
only to add new comments to the component
Figure 1: Dante architecture. elements or edit comments define earlier. The editor
is not made up from elements that will allow authors
To summarize, the main characteristics that we to format the content (i.e. defining font types,
consider essential for an efficient collaborative paragraph alignment etc.), but from elements that
document creation and review system are: represent structural components of document (i.e.
paragraph, note, quote, table etc.). The formatting of
support for virtual team efficiency
document will be done automatically by the
user and document centered
application according to the template that the
support users location and time independency
document owner had defined. Each structural
platform independent
component of the document can be commented by
efficient group interaction
the rest of the team, the editor defining special zones
allow users to concentrate on the content and
at the end of each element where these comments
waste as little as possible time with the technology
can be consulted. When accessing chapters for
that supports the process
whom the current users is not the owner, a different
document management facilities
editor is loaded allowing the user to define and edit

439
ICE-B 2008 - International Conference on e-Business

comments and reply to those that others have The Model layer consists mainly from two
defined. Before finalizing the project, the chapters subcomponents: the Document manager and the
can be exported in different open formats (i.e. PDF) Repository. The Repository is a collection of DTD,
or to a link in order to be accessed by those that are XML and XSLT files used by the Document
not participating to the project. manager module.
The Chat sections allows user to communicate The Model layer consists mainly from two
using synchronous channels. The section defines two subcomponents: the Document manager and the
main channels: Groups and Personal. The Groups Repository. The Repository is a collection of DTD,
channel lists all active projects for the current user XML and XSLT files used by the Document
and allows him to communicate with the members manager module.
of each project in different panels. When a particular
group is selected, the user can communicate with the
members of that project that are online, the messages
being available for all co-workers. All messages are
stored in XML files and associated to the project,
allowing user to consult the discussion archive or the
project anytime. On the other hand, the Personal
channel allows user to communicate with each
friend defined in the Agenda using private channels.
These messages are not stored in the message
archive, only if the users decide so.
The Partners section has two main
Figure 2: Document workflow.
subcomponents: Agenda and Invitations. The
Agenda allows users to manage their partners’
The document type definition files describe the
contacts and export them to micro-formats like
structure of the documents and all metadata that can
HCard. The user’s personal data can be exported in
be added to them. In this case, to describe the
VCard format. The Invitations subcomponent
structure of the scientific papers we’ve created a
manages the user’s invitations to participate at
document type called xmlDocument. The structure
different projects (only after accepting an invitation
of this type is in a way similar to the more mature
a user can actively participate to a project). All
DocBook xml vocabulary (Walsh 2005), being more
personal data is stored in XML files, allowing the
simplified and particularized for scientific papers.
system to easily integrate them into the projects
The xmlDocument DTD defines the structure of the
content when exporting a project in a final state
whole document viewed as a project. The two main
according to different templates.
elements of an xmlDocument are docInfo, which
In the next section we shall discuss the details
describes the metadata associate with the document
regarding the implementation of our web based
(i.e. document name, authors, document type, last
collaborative document creation and review system.
update, version etc.) and docBody. The docBody
defines the content of the document (i.e. abstract,
references, appendixes etc.) and for to each chapter
3 IMPLEMENTATION the metadata associated (id, title and author) and a
link to the XML file that defines its content. This
As previously mentioned, Dante is a web based separation of each chapter in different XML files
system design upon de model-view-controller allows a more flexible management of content and
design-pattern which stores all data in XML files. metadata. The vocabulary that describes the
The View layer personalizes the user interface chapter’s elements is defined by the xmlChapter
according to user’s rights and permissions and document type definition, implementing means of
displays the appropriate editor after reading the storing and identifying elements. As mentioned
documents metadata. It also implements a earlier, each author can define comments for the
communication module which transfers data to the chapters that he is not editing. These comments are
Controller layer using AJAX. The Controller layer stored in separate XML files, and the vocabulary
handles the events triggered from the UI and calls describing them defines elements to uniquely
the appropriate handler from the Model layer. We identify de comment, describe its characteristics (i.e.
will concentrate our attention on the Model layer author, date etc.) and link it to the element to which
witch manages all XML documents. it referees.

440
 WEB BASED COLLABORATIVE DOCUMENT CREATION AND REVIEW SYSTEM

In order for the Document manager to easy set client-side are in some manner a transposing of
the corresponding roles and rights, a document type those residing on the server being also a wrapper
definition is implemented to store the projects. This upon the document. The Session Manager module
DTD allows defining for each user the projects for from the Controller layer determines which script to
which he is a collaborator or the owner. The be loaded according to conjuncture determining the
Document manager module consist in several operations that a user can execute on a particular
classes based on the DOM implementation which document. When editing a particular chapter, all new
automates the processes concerning the creation, elements and all changes are stored in different
storage, retrieval and the export of the documents. queues and only when the users decides to save the
All server side scripts have been implemented in document the content of these queues is send using
PHP which includes a good support for XML, AJAX to the server. These classes are also based on
complying with the commonly used standards (SAX, the DOM model, the content being sent to the server
DOM, SimpleXML, XMLReader, XMLWriter, and representing a DOM node to be inserted in the
the XSLT). Some of the classes defined in the document residing on the server. The queues are
document manager come as a wrapper for the DTD gradually discharged when a component element is
residing in the repository: the xmlChapter class, for successfully saved on the server. This approach
example, handles the creation and the update of allows defining specific UI elements and behavior to
different document objects, prepares the content for each document type being treated. The user screen is
printing accessing the appropriate class when blocked while saving the content of the queues,
exporting (i.e. the pdfPrinter class when exporting to ousting the chances to make unsafe changes to the
PDF). This module generates the appropriate user content. When appropriate, an XSLT processor is
interface according to the user’s role and rights, used on the client-side to reduce the charge on the
loading the corresponding XSL and parsing the server.
requested XML file. For the same DTD the We have chosen this modular approach based on
Repository defines different XSL transformation document types and user interfaces particularized on
files for each role and user-level, restricting the roles/rights and document types in order to facilitate
access to different actions that the user can perform further development of the system in such a manner
upon the document. Most of the UI used by Dante is that it could handle a lot more document types.
generated using XSL transformations.
The Chat module works in a similar way,
defining a DTD for storage purposes and different 4 CONCLUSIONS
XSL transformation schemes for display. As
mentioned previously, the user can communicate The work reported in this paper has addressed the
using private channels or rooms dedicated to problem of integrating document management
different projects. The messages exchanged between functionalities and workflow capabilities into single
users in these project related rooms are bound to the system with support for virtual teams in order to
project using metadata (although physically residing achieve a very efficient solution for collaborative
in different locations) allowing project based document editing. Allowing users to edit
message archiving. This approach allows members collaboratively documents and supporting their
to catch-up with the team when not being able to needs for time and location independence can result
join the group. The Agenda module stores member’s in more team efficiency. Users have to be able to
personal data (this being done also using XML files) concentrate their efforts on content and reduce as
and allows users to export this information in HCard much as possible the time used to handle and
and VCard microformats. All templates require a integrate technologies. An approach that integrates
minimum of identification data for the authors, the functionalities from document management systems,
project importing all required data when formatting workflow management and collaborative editing
from this module. We have tried to structure the proves to be a much more user centered and
implementation as much as possible according to supportive collaborative solution. As such an
functionality of the whole system, defining a implementation, Dante tries to offer sufficient
modular structure. support in order for the teams to be goal directed and
The user interaction is managed using efficient. Using a web based solution does not
JavaScript. Because the communication from the impose restrictions on the users regarding operating
server to the client is mostly done using XML systems or software. Taking in consideration the
chunks (or an entire file), the classes defined on the multitude of solutions available, we have to consider

441
ICE-B 2008 - International Conference on e-Business

using as much as possible open formats in order for Workflow and Collaborative Document Authoring in
the content to be easily integrated. As shown in this Medical Research, in Proceedings of Tenth
paper, using XML as building blocks allows the Australasian Document Computing Symposium,
system to be document and user centered. We have Sydney, Australia,
http://eprints.vu.edu.au/archive/00000648/
presented a case study of a particular field where Raikundalia, G & Zhang, H 2004, ‘Novel Group
such an implementation would be a great support. Awareness Mechanisms for Real-time Collaborative
As further work, we intend to develop our system Document Authoring’, The Ninth Australasian
toward a framework that can handle a wide range of Document Computing Symposium (ADCS 2004),
documents, taking in consideration the increased University of Melbourne, Australia,
need for collaborative document editing in a http://www.cs.mu.oz.au/~alistair/adcs2004/papers/pap
multitude of working fields. er05.pdf
Spek, S 2008, ‘Wikis are good for knowledge
management’, submitted 6 February, 2008,
arXiv:0802.0745v1
ACKNOWLEDGEMENTS Walsh, N 2005, DocBook 5.0: The Definitive Guide,
O'Reilly & Associates
The work presented has been funded by the research
grant “Intelligent System for Business Decisions
Support”, Director Nitchi Stefan, PhD., Professor,
PNII Program, 91-049/18.09.2007 supported by
Higher Education Ministry.

REFERENCES
Aversano, L, Canfora, G, Lucia, A & Gallucci, P 2001,
‘Integrating Document and Workflow Management
Systems’, Proceedings of the IEEE Symposia on
Human-Centric Computing Languages and
Environments,
http://csdl2.computer.org/comp/proceedings/hcc/2001/
0474/00/04740328.pdf
Guerrero, L & Collazos, C & Pino, J & Ochoa, S &
Aguilera, F 2004’Designing Virtual Environments to
Support Collaborative Work in Real Spaces’, in
Journal of Web Engineering, vol.2, no.4, pp. 282-294
Hampel, T, Keil-Slawik, R 2000, ‘A Collaborative
Document Management Environment For Teaching
And Learning’, in Proceedings of the Third
International Conference on Collaborative Virtual
Environments. San Francisco (Ca.), USA, pp.197-198.
Lawrence, S 2001, ‘Online or invisible?’, Nature, vol. 411,
no. 6837, p.521,
http://www.idemployee.id.tue.nl/g.w.m.rauterberg/pub
lications/CITESEER2001online-nature.pdf
Leone, S & Hodel, T & Gall, H 2005, ‘Concept and
Architecture of a Pervasive Document Editing and
Managing System’, in Proceedings of the 23rd annual
International Conference on Design of
Communication (SIGDOC), Coventry, United
Kingdom, ACM Press, pp. 41-47
Millward, L & Kyriakidou, O 2004, ‘Effective virtual
teamwork: a social-cognitive and motivational model’,
in Virtual and Collaborative Teams: Process,
Technologies and Practices, ed. Idea Group Inc.,
United Kingdom
Nallaparaju, V & Raikundalia, G & Brand, C & Bain, C &
Hutchinson, A 2005, ‘An Experimental Study of

442
 ANTECEDENCES AND CONSEQUENCES OF E-SERVICE
QUALITY ACROSS INDUSTRY SECTORS

Dauw-Song Zhu
Business Administration & Accounting Department, National Dong Hwa University
No. 1, Sec. 2, Da Hsueh Rd. Shou-Feng Hualien, Taiwan 974
[email protected]

Chih-Te Lin
Business Administration Department, National Dong Hwa University & Lecturer of Food and Beverage Management
Department, Taiwan Hospitality & Tourism College
No. 2, Lane 46, Jhongmei 3rd St, Hualien, Taiwan 970
[email protected]

Yu-Ling Su
Business Administration Department, National Dong Hwa University & Lecturer of Food and Beverage Management
Department, Taiwan Hospitality & Tourism College
No. 2, Lane 46, Jhongmei 3rd St, Hualien, Taiwan 970
[email protected]

Keywords: e-Service quality, e-commerce quality, online quality, Internet quality.

Abstract: A few studies are available on the different sectors basis on the determinants of success in online retailing.
This study tries to identify which e-Service dimensions serve as determinants of e-Service quality (Zeithaml
et al., 2002). Therefore we focused on analyzing impacts on overall e-Service performance of various
e-Service quality dimensions in internet retailing. We analyzed the impact of e-Service quality dimensions
of online retailers along several sectors. We found that while many dimensions of online service
performance were similar in their impact across all sectors, several dimensions in particular exhibited
sector-by-sector performance differences.

1 INTRODUTION e-Service quality on the web has thus become an

issue for ensuring customer satisfaction, loyalty and
What companies on the Internet require is to deliver firms’ profitability and service, both for industry and
high value to the customer, build customer loyalty, the public sector (Buckley 2003). Thus, we examine
encourage repeat purchases and maintain long term each of several industry sectors separately to
relationships with the customers (Gurau 2003; distinguish the web site dimensions that are
Parasuraman and Zinkham 2002), since recent associated with creating overall satisfaction in the
research indicates that ‘positively predisposed online consumer. Here we concisely review
on-line customers’ demand much more (Van Riel et literatures that have empirically examined
al. 2003). To do so, as Zeithaml et al., (2002, p362) antecedents and consequences dimensions of
argue, companies need to shift the focus of e-Service quality in electronic services. Table 1 is a
e-business from e-commerce to e-Service. Such shift list of the variables that have been examined in these
implies understanding what customers’ value in an literatures, both in multi-industry sector e-retailing
on-line encounter, how they perceive the experience literatures, and in single industry studies related to
and how they evaluate its quality (Lauren et al, general e-Service, online banking/ financial services,
2006). travel services, book retailing, and health care
Understanding, measuring and managing services.

443
ICE-B 2008 - International Conference on e-Business

2 ANTECEDENCES 2002; Lee, G. G. & Lin H. F., 2005), functional

quality is the antecedent across e-retail industry
A few previous literatures have studied the issue of (Choi et al., 2004; Kim et al., 2002) and health
online e-Service quality from a multi-industry industry (Van Riel et al., 2001), and security or
perspective, or have collected data across several privacy is the antecedent across e-retail industry
industry sectors. According to review of 19 (Kim et al., 2002; Janda, S., Trocchia, P. J. and
literatures, we discovered overall web site quality is Gwinner, K. P., 2002) and books industry (Lin, H.
the most frequent antecedent across general F., 2007).
e-Service/ e-retail industry (Kim et al., 2002; Palmer, For banking and travel industry, aesthetic appeal is
2002), online banking/ financial services industry the driver of e-Service quality (Montoya-Weiss et
(Chen and Hitt, 2002; Montoya-Weiss et al., 2003; al., 2003; Van Riel et al., 2003), and comparative
Jayawardhena, 2004), travel industry (Harris and prices is also the driver across e-retail industry (Choi
Goode, 2004), books industry (Gefen, 2002) and et al., 2004) and banking industry (Balasubramanian
health industry (Van Riel et al., 2001) . The second et al., 2003). Download speed became the
frequent antecedent is trustworthiness or assurance antecedences across e-retail industry (Palmer, 2002)
across four industries of online banking services and travel industry (Van Riel et al., 2003).
industry (Balasubramanian et al.,2003; Usefulness is significant driver in books industry
Jayawardhena, 2004), travel industry (Harris and (Koufaris, 2002), health industry (Viitanen et al.,
Goode, 2004), books industry (Gefen, 2002; Lee, G. 2003). And on time delivery became the significant
G. & Lin H. F., 2005; Lin, H. F., 2007) and health driver in books industry (Gefen, 2002) and health
industry (Viitanen et al., 2003). industry (Viitanen et al., 2003). Online banking
We examined general e-Service industry, online industry and travel industry studies used navigation
banking services industry, travel industry and books as antecedent (Montoya-Weiss et al., 2003; Van Riel
industry, and found that security of environmental et al., 2003). Product quality viewed as predicator in
risk is a antecedence variable of general e-Service/ e-retail industry (Choi et al., 2004) and travel
e-retail industry (Kim et al., 2002; Zhang, X. & industry (Clemons et al., 2002), responsiveness used
Prybutok, V. R., 2005), online banking industry as predicator in books industry (Lee, G. G. & Lin H.
(Balasubramanian et al., 2003; Montoya-Weiss et al. F., 2005; Lin, H. F., 2007). However, web site
2003), books industry (Gefen, 2002) and in health design is significant association with dependent
industry (Viitanen et al., 2003). In addition to that, variables only in books industry (Lee, G. G. & Lin
we also found content or information is the H. F., 2005; Lin, H. F. 2007).
predicator of e-Service quality across industries of
e-retail industry (Kim et al., 2002; Palmer, 2002;
Janda, S., Trocchia, P. J. and Gwinner, K. P., 2002; 3 CONSEQUENCES
Liao, C., Palvia, P. and Lin H. N., 2006), online
banking industry (Montoya-Weiss et al., 2003) and We found satisfaction is the most major consequent
books industry (Lin, H. F., 2007). Furthermore, variables in the previous studies across general
personalization, customization or interactivity were e-Service/ e-retail industry (Kim et al, 2002; Palmer,
predicator across e-retail industry (Palmer, 2002), 2002; Janda, S., Trocchia, P. J. and Gwinner, K. P.,
online banking industry (Chen and Hitt, 2002; 2002; Zhang, X. & Prybutok, V. R., 2005), online
Jayawardhena, 2004), travel industry (Van Riel et banking/ financial services industry
al., 2003) and books industry (Lin, H. F., 2007). But (Balasubramanian et al. ,2003; Montoya-Weiss et al.,
technical quality or technical adequacy viewed as 2003);), travel industry (Van Riel et al. ,2003; Harris
antecedent only in e-retail industry (Choi et al., and Goode, 2004), books industry (Lee, G. G. & Lin
2004; Kim et al., 2002; Liao, C., Palvia, P. and Lin H. F., 2005; Lin, H. F., 2007) and health industry
H. N., 2006) and health industry (Viitanen et al., (Van Riel et al., 2001; Viitanen et al., 2003). And the
2003). second major consequences is customer loyalty
We found accessibility or e-Service convenience across four industries including general e-Service/
used as independent variable in studies of e-retail e-retail industry (Kim et al. ,2002; Palmer ,2002),
industry (Zhang, X. & Prybutok, V. R., 2005), online travel industry (Van Riel et al. ,2003; Harris and
banking industry (Jayawardhena, 2004) and travel Goode ,2004), books industry (Gefen, 2002;
industry (Van Riel et al., 2003). Fulfillment or Koufaris, 2002) and health industry (Van Riel et al..
reliability is the antecedent across travel industry 2001; Viitanen et al., 2003). Next consequent
(Van Riel et al., 2003) and books industry (Gefen, variable is quality or e-Service quality studied in

444
 ANTECEDENCES AND CONSEQUENCES OF E-SERVICE QUALITY ACROSS INDUSTRY SECTORS

three industries including online banking/ financial overall e-Service performance of various e-Service
services industry (Jayawardhena, 2004), travel quality dimensions in internet retailing. We analyzed
industry (Van Riel et al., 2003) and books industry the impact of e-Service quality dimensions of online
(Lee, G. G. & Lin H. F. ,2005). Moreover, value retailers along several sectors. We found that while
variable is studied across different industries general many dimensions of online service performance
e-Service/ e-retail industry (Choi et al., 2004), travel were similar in their impact across all sectors,
industry (Harris and Goode, 2004) and health several dimensions in particular exhibited
industry (Van Riel et al., 2001). sector-by-sector performance differences.
In addition, we examined online banking and This study makes several contributions to the
books industry, and found customer switching is the literature on e-Service performance specifically due
antecedent separately studied in Chen and Hitt to our sector-by-sector analysis. In prior studies,
(2002) and Gefen (2002). And Clemons et al. (2002) researchers have taken a scattered approach to
surveyed travel industry, price dispersion as identifying e-Service quality dimensions for their
antecedent. Otherwise, Janda, S., Trocchia, P. J. and studies. They have also focused on the issue from a
Gwinner, K. P., (2002) examined general e-Service/ single industry perspective. This research approach
e-retail industry and used likelihood of future has led the literature to a state of many quality
purchase, likelihood of complaining and dimensions having been identified, but little
word-of-mouth as dependent variables, but Liao, C., evidence that can be triangulated across sectors or
Palvia, P. and Lin H. N. (2006) used trust as across studies. In closing, we hope our paper will
dependent variables at the same industry. We may prove insightful to academics and managers
make up the situation of outcome variables were interested in managing e-Service operations. We
identified, the summary shows in Table IV. believe there is much interesting research left to be
done on such topics, and hope our paper serves to
motivate others to explore the emerging issues in this
4 METHOD interesting industry arena.

The a-priori theory underlying our analysis of the

relationship between individual e-Service quality REFERENCES
dimensions and overall e-Service performance is that
customers in each e-retailing sector will possess Balasubramanian, S., Konana, P. and Menon, N.M.
certain preferences about aspects of an online (2003), “Customer satisfaction in virtual
shopping system, and e-Service designers will strive environments: a study of online investing”,
to understand those sector-specific factors and will Management Science, Vol. 49 No. 7, pp. 871-89.
Berthon, P., Pitt, L.F. and Watson, R.T. (1996), "The
design their online service system to attempt to world wide web as an advertising medium", Journal of
satisfy customers in their sector. Across all online Advertising Research, Vol. 36 No. 1, pp. 43-54.
retailing sectors, a large body of e-Service quality Chen, P. and Hitt, L.M. (2002), “Measuring switching
dimensions will exist, subsets of which will be costs and the determinants of customer retention in
relevant for managing individual sectors. internet-enabled businesses: a study of the online
brokerage industry”, Information Systems Research,
Vol. 13 No. 3, pp. 255-74.
Choi, B., Lee, C., Lee, H. and Subramani, M. (2004),
5 CONCLUSIONS “Effects of web retail e-Service quality and product
categories on consumer behavior: a research model
Firms today need to understand better how different and empirical exploration”, Proceedings of the 37th
industry sectors and contexts lead to different Hawaii International Conference on System Sciences,
relationships between tactical activities such as Waikoloa, Big Island, Hawaii.
e-Service quality management and their overall Clemons, E.K., Hann, I-H. and Hitt, L.M. (2002), “Price
dispersion and differentiation in online travel: an
performance outcomes (Bolton et al., 2004). With
empirical investigation”, Management Science, Vol.
the phenomenal growth of e-Services, research must 48 No. 4, pp. 534-49.
aim to better understand the dimensions, Dhyani, D., Ng, W. and Bhowmick, S. (2002), "A survey
antecedences, and consequences of quality and of web metrics", ACM Computing Surveys, Vol. 34
customer satisfaction in the online domain across No. 4, pp. 469-503.
various industry sectors (Lauren et al, 2006). Gefen, D. (2002), “Customer loyalty in e-commerce”,
Therefore we focused on analyzing impacts on Journal of the Association for Information Systems,
Vol. 3, pp. 27-51.

445
ICE-B 2008 - International Conference on e-Business

Gronroos, C. (1984), "A e-Service quality model and its Palmer, J.W. (2002), “Web site usability, design, and
market implications", European Journal of Marketing, performance metrics”, Information Systems Research,
Vol. 18 No. 4, pp. 36-44. Vol. 13 No. 2, pp. 151-67.
Harris, L.C. and Goode, M.M.H. (2004), “The four levels Pollard, K. and Blyth, R. (1999), "User-centered design of
of loyalty and the pivotal role of trust: a study of web sites and the redesign of lineone", BT Technology
online service dynamics”, Journal of Retailing, Vol. Journal, Vol. 17 No. 1, pp. 69-75.
80, pp. 139-58. Riel, A.C.R., Liljander, V. and Jurriens, P. (2001),
Huffman, D.L. and Novak, T.P. (1995), Marketing in “Exploring consumer evaluations of e-Services: a
Hypermedia Computer-Mediated Environments: portal site”, International Journal of Service Industry
Conceptual Foundations, available at: Management, Vol. 12 No. 4, pp. 359-77.
www.2000.ogsm.vanderbilt.edu/ Riel, A.C.R., Semeijn, J. and Pauwels, P. (2003), “Online
cmepaper.revision.julyll.1995/cmepaper.html travel e-Service quality: the role of pre-transaction
(accessed 11 July). services”, MAXX Working Paper 2003-01, Maastricht
Janda, S., Trocchia, P. J. and Gwinner, K. P. (2002), Research School of Economics of Technology and
“Consumer Perceptions of Internet Retail E-Service Organization.
quality”, International Journal of Service Industry Roy, M., Dewit, O. and Aubert, B. (2001), "The impact of
Management, Vol. 13, No.5, pp.412-431. interface usability on trust in web retailers", Internet
Jayawardhena, C. (2004), “Measurement of e-Service Research: Ekctronic Networking Applications and
quality in internet banking: the development of an Policy, Vol. 11 No. 5, pp. 388-98.
instrument”, Journal of Marketing Management, Vol. Slatin, M.J. (2001), "The art of ALT: toward a more
20, pp. 185-207. accessible web", Computers and Composition, Vol. 18
Kim, J., Lee, J., Han, K. and Lee, M. (2002), “Businesses No. 1, pp. 73-81.
as buildings: metrics for the architectural quality of Thelwall, M. (2000), "Commercial web sites: lost in
internet businesses”, Information Systems Research, cyberspace?", Internet Research: Electronic
Vol. 13 No. 3, pp. 239-54. Networking Applications and Policy, Vol. 10 No. 2,
Koufaris, M. (2002), “Applying the technology acceptance pp. 150-9.
model and flow theory to online consumer behavior”, Viitanen, J., Liljander, V., Pura, M. and van Riel, A.
Information Systems Research, Vol. 13 No. 2, pp. (2003), “Customer loyalty to content-based web sites:
205-23. the case of an online health care service”, Working
Lauren M Trabold, Gregory R Heim, Joy M Field. Paper, Maastricht Academic Center for Research in
,Comparing e-Service performance across industry Services.
sectors: Drivers of overall satisfaction in online Weinberg, D.B. (2000), "Don't keep your internet
retailing, International Journal of Retail & Distribution customers waiting too long at the (virtual) front door",
Management. Bradford: 2006. Vol. 34, Iss. 4/5; p. 240 Journal of Interactive Marketing, Vol. 14 No. 1, pp.
(18 pages) 30-9.
Lee, G. G. & Lin H. F. (2005), “Customer Perceptions of Wolfinbarger, M. and Gilly, M.C. (2003), “eTailQ:
E-Service quality in Online Shopping”, International dimensionalizing, measuring, and predicting etail
Journal of Retail & Distribution Management, Vol. 33, quality”, Journal of Retailing, Vol. 79, pp. 183-198.
No, 2, pp.161-176. Zhang, X. and Prybutok, V. R (2005), “A Consumer
Liao, C., Palvia, P. and Lin, H. N. (2006), “The Roles of Perspective of E-Service quality”, IEEE Transactions
Habit and Web Site Quality in E-commerce”, on Engineering Management, Vol. 52, No. 4, pp.
International Journal of Information Management, 461-477.
Vol. 26, pp.469–483.
Lin, C.J. and Lu, H. (2000), "Towards an understanding of
the behavioral intention to use a web site",
International Journal of Information Management,
Vol. 20 No. 3, pp. 197-208.
Lin, H. F. (2007), “The Impact of Website Quality
Dimensions on Customer Satisfaction in the B2C
E-commerce Context”, Total Quality Management and
Business Excellence; Vol. 18 No 4, pp. 363-378.
Liu, C. and Arnett, P.K. (2000), "Exploring the factors
associated with web site success in the context of
electronic commerce", Information & Management,
Vol. 38 No. 1, pp. 23-33.
Montoya-Weiss, M.M., Voss, G.B. and Grewal, D. (2003),
“Determinants of online channel use and overall
satisfaction with a relational, multichannel service
provider”, Journal of the Academy of Marketing
Science, Vol. 31 No. 4, pp. 448-58.

446
 RFID PASSWORD MANAGEMENT METHODS FOR
FALSIFICATION PREVENTION IN BOOKSTORE
MANAGEMENT USING SECURE RFID TAGS

Yuichi Kobayashi, Yoji Taniguchi

Hitachi, Ltd., Systems Development Laboratory, 1099 Ohzenji, Asao, Kawasaki, Kanagawa, 215-0013, Japan
[email protected], [email protected]

Toshiyuki Kuwana
Hitachi, Ltd., Tracing & Tracking Systems Division, 890 Kashimada, Saiwai, Kawasaki, Kanagawa, 212-8567, Japan
[email protected]

Masanori Akiyoshi
Graduate School of Information Science and Technology, Osaka University, 2-1
Yamada-oka, Suita, Osaka, 565-0871, Japan
[email protected]

Keywords: RFID, falsification prevention, risk analysis.

Abstract: The receipt data on radio frequency identification (RFID) tags attached to books may be used to prevent
shoplifting in bookstores. To protect the receipt data, it is important to manage the passwords of RFID tags.
We use Secure RFID tags, which protects data with an RFID password, for preventing falsification of RFID
data. We also propose ten methods that manage RFID passwords from the point of the phases of RFID
passwords and the ways a password is associated with a Secure RFID tag. We analyze and compare these
methods using fault tree analysis. We show that our proposed RFID passwords management methods are
effective in preventing falsification for resale.

1 INTRODUCTION Secure RFID tags are low cost RFIDs. This type
of RFID tag protects data using a simple password
The shoplifting of books is a serious problem in authentication method. We chose to use Secure
Japan. A report says that there are many shoplifted RFID tags because these tags are smaller, cost less,
books for resale. The publishing industry in Japan is and suitable for attaching to books. It is very
examining a method for attaching an RFID tag to important to manage an RFID password when using
books, which records receipt data on the RFID tag Secure RFID tags. Therefore we propose ten
memory. However, there is a risk of someone methods for managing RFID passwords. We analyze
reselling a shoplifted book to a second-hand their security, and show that they are effective in
bookstore, after illegally overwriting the receipt data preventing falsification.
in the RFID tag. So a method for protecting the data
in the RFID tag is needed.
There are several methods that mount hash logic 2 BOOK MANAGEMENT
on an RFID tag and authenticate the user to protect SYSTEM WITH SECURE RFID
the RFID data (Weis 2003; Engberg, Harning &
Jensen 2004; Tripathy & Nandi 2006). However, TAGS
these methods are not realistic in a situation in which
an RFID tag is attached to a book because mounting In this chapter, we describe the book management
advanced calculation logic on an RFID tag requires system with Secure RFID tags in a bookstore
a larger IC chip, which increases the cost of the environment.
RFID tag.

447
ICE-B 2008 - International Conference on e-Business

Secure RFID memory

Publisher
common
common data
data publisher
publisher data
data distributor
distributordata
data bookstore
bookstore data
data
ISBN
ISBN and
and ・
・・
・
・・ ・
・・
・
・・
Serial
Serial No.
No.

Write Lock

Reverse Logistics
Distributor common
common data
data publisher
publisher data
data distributor
distributordata
data bookstore
bookstore data
data
ISBN
ISBN and
and ・
・・
・
・・ ・
・・
・
・・

Forward Logistics
Serial
Serial No.
No.
Delete
Write Lock “Stock” when
Write “Stock” at the the book is returned.
time of arrival.
Bookstore
common
common data
data publisher
publisher data
data distributor
distributordata
data bookstore
bookstore data
data
ISBN
ISBN and
and ・
・・
・
・・ ・
・・
・
・・ Stock
Stock
Serial
Serial No.
No.

Write “Sold” at the Write Lock

Consumer
time of sale.

common
common data
data publisher
publisher data
data distributor
distributordata
data bookstore
bookstore data
data
ISBN
ISBN and
and ・
・・
・
・・ ・
・・
・
・・ Sold
Sold
Serial
Serial No.
No.
Secondhand
Bookstore Write Lock
Check for “Sold”
when buying the book.

Figure 1: Shoplifter prevention method.

2.1 Book Distribution Procedure for information should be done with an access control
Preventing Shoplifting function with a password of the Secure RFID tag.

The publishing industry in Japan is considering a 2.2 An Existing RFID Passwords

method for preventing shoplifting by using Secure Management Method
RFID tags. Figure 1 shows the procedure for
preventing shoplifting in a book distribution. First, a Figure 2 shows the RFID password management
bookstore writes in “Stock” onto the Secure RFID system for a bookstore.
tag when a book arrives from a distributor. This
Operation 1: When the book arrives, the book’s
denotes the state before being sold. Next, when a
code written in the Secure RFID tag is read, and the
consumer buys the book, the bookstore changes
pre-arrival data is checked. After it is checked, the
“Stock” to “Sold” in the Secure RFID tag. At the
book code is input into the server with the arrival
entrance of the store, the security system reads the
data. Then, the date of arrival, bookstore code,
data on the Secure RFID tags to check if the book
“Stock”, etc... are written onto the Secure RFID tag.
was purchased. Finally, when a second-hand
An RFID password is then set up for the Secure
bookstore buys a book from a consumer, it checks
RFID tag, and the memory is locked using that
whether the book’s Secure RFID tag reads “Sold” or
password.
“Stock”. Not only bookstores, but second-hand
bookstores cooperate with the publishing industry. Operation 2: When the book is sold, the book code
All the companies relevant to book publishing and in the Secure RFID is read, the price is checked, and
selling can examine the system to counter the the sale information is input into the server. At this
shoplifting problem. time, the memory of the Secure RFID tag is
In order for this system to work, the data of the unlocked by using the RFID password, and “Sold” is
Secure RFID tag must not be able to be easily rewritten onto the Secure RFID tag. Finally, the
rewritten from “Sold” to “Stock”. Rewriting memory is locked again by using the password.

448
 RFID PASSWORD MANAGEMENT METHODS FOR FALSIFICATION PREVENTION IN BOOKSTORE
MANAGEMENT USING SECURE RFID TAGS

BOOKSTORE
BOOKSTORE
1 Arrival of books
Handy
Interrogator read data and 2 Sale of books
set up a tag
password read data and Fixed
write new data Interrogator
using the tag Gate
password

Handy
Interrogator 3 Surveillance

4 Returned books Fixed

only read data
Interrogator

only read data

Figure 2: System using the Secure RFID tags in the bookstore.

effective to divide RFID passwords into groups to

Operation 3: A gate type interrogator system at the
manage different RFID passwords. For example, the
entrance is set up, and the interrogator checks if
group includes an inspection interrogator, a book,
“Sold” or “Stock” is written on the Secure RFID tag.
etc… Moreover, because it is usually impossible to
Operation 4: If the book is returned, the memory of read the RFID password set in the Secure RFID tag,
the Secure RFID tag is unlocked by using the RFID it is necessary to set the RFID tag with an RFID
password, and all data from the bookstore and the password beforehand. Therefore, we present ten
password are deleted. Afterwards, the returned methods which can be used to manage RFID
book’s information is input into the server. passwords from the point of a phase of RFID
password management and the ways in which a
The publishing industry in Japan tries to manage password is associated with a Secure RFID tag.
RFID passwords by using identical passwords in all Figure 3 shows the relationship between the phase of
bookstores, and applies the RFID password RFID passwords, and the ways passwords are used
management method for each bookstore. This for a Secure RFID. There are four phases of RFID
method has the advantage of being less expensive; password management, (1) each bookstore, (2) each
however, if one RFID password is stolen, all Secure inspection interrogator system, (3) each arrival, and
RFID tags are compromised. The possibility of (4) each book. On the other hand, there are (a) the
information being falsified on the Secure RFID tags way of using a conversion table, (b) the way of using
attached to books stocked in all bookstores is high. operation logic, and (c) the way of updating a key.
Therefore, the distinction between a shoplifted book We define the combination of phases of RFID
and a legally bought one becomes difficult in a passwords and the ways a password is used for a
second-hand bookstore, and the effect of preventing Secure RFID tag as the RFID password management
theft by using the Secure RFID tag weakens. It is methods.
necessary to manage RFID passwords more
carefully.
3.1 Phases for Managing RFID
Passwords
3 RFID PASSWORDS A bookstore that manages RFID passwords
MANAGEMENT METHODS separately by each group lessens the risk of RFID
tag falsification more than a bookstore that manages
We propose ten methods for managing not identical identical RFID passwords. We explain the phases
RFID passwords but different passwords so that the for managing RFID passwords as follows.
risk after a password is stolen will be lessened. It is

449
ICE-B 2008 - International Conference on e-Business

existing method→ identical

identical

( １) each
each
bookshop
bookshop

( ａ) ( ｂ) ( ｃ)
( ２) each
each conversion operation update
interrogator
interrogator table logic electronic key
( ａ) ( ｂ) ( ｃ)
( ３) each
each conversion operation update
arrival
arrival table logic electronic key

( ａ) ( ｂ) ( ｃ)
( ４) each
each conversion operation update
book
book table logic electronic key

Figure 3: Classification of Password Management Methods.

(1) Each bookshop: Manage the different RFID 3.2 Ways in which RFID Passwords
passwords in each bookstore. Each bookstore are Associated with a Secure RFID
decides and manages an RFID password. Tag
(2) Each interrogator system: Manage the
different RFID passwords in each interrogator When a different RFID password is set to a Secure
system in a bookstore when the book arrives. RFID tag, the system should associate that password
Each interrogator sets up an RFID password with that Secure RFID tag. We explain the ways in
and an interrogator’s ID to the Secure RFID which this is accomplished.
tag attached to the newly arrived book. (a) Conversion table: The system generates the
RFID password at random, and associates the
(3) Each interrogator system and arrival: Manage RFID password with the interrogator’s ID,
the different RFID passwords in each arrival date, or a unique ID written to the
interrogator system in a bookstore when the Secure RFID tag to identify the phase of RFID
book arrives and update the RFID passwords password management.
each time books arrive. Each interrogator
system sets up an RFID password, an (b) Operation logic: Use operation logic with an
interrogator’s ID, and the arrival date to the electronic key. The system generates an RFID
Secure RFID tag. password by calculating data, such as the
interrogator’s ID, arrival date, or a unique ID
(4) Each book: Manage the different RFID written to the Secure RFID tag (Kobayashi,
passwords in each Secure RFID tag attached Kuwana, Taniguchi & Komoda 2007).
to the book. The bookstore associates the (c) Update electronic key: Periodically update the
RFID password with a unique ID that electronic key used in (b).
identifies the Secure RFID tag.
It is possible to manage the different RFID
passwords in each title of the book. However, the 4 SECURITY EVALUATION OF
titles with only one book in stock accounts for 70% RFID PASSWORD
of the inventory in most bookstores. Therefore, we
omit the explanation for each title here because each MANAGEMENT METHODS
title and each book are almost the same.
In this chapter, we show the results form a fault tree
analysis (FTA) of the RFID password management
methods, and compare those analysis results.

450
 RFID PASSWORD MANAGEMENT METHODS FOR FALSIFICATION PREVENTION IN BOOKSTORE
MANAGEMENT USING SECURE RFID TAGS

RFID data will be falsified,

and resale of the book will
be successful
A
B C

(B-2) Can falsify (C-1) An (C-2) Can falsify

B-1 sales data using a interrogator sales data using the
stolen password will be stolen stolen interrogator
0.3%
(B-11) Can falsify sales (B-12) A password
data using a password will be decoded by
stolen at another bookstore hacking
4.1%
Figure 4: FTA of shoplifting a book.

4.1 Fault Tree Analysis of RFID (B-12): This event is the probability that an RFID
Passwords Management Methods password will be decoded in one month by
hacking into a Secure RFID tag. This
We analyzed the security performance of the RFID probability PB-12 was obtained from the
password management method, which combined the following expression.

PB −12 = 1month E (Time ) = 4.1%

phase of an RFID password and the ways in which
(1)
the password is associated with a Secure RFID tag
described above, using FTA. E (Time) is an expected time spent on the
We know that the probability of a successful hacking attack. The time that a Secure RFID
falsification of RFID data for resale is low; the RFID tag is checked whether one password is right
password management method will have a high using one interrogator is assumed to be about
security performance. Therefore, let a top event of 30 milliseconds and the length of the Secure
FTA be the probability that the RFID data of a RFID password is assumed to be 32 bits
shoplifted book will be falsified, and resale of the according to the specification of Secure RFID.
book will be successful in one month. Moreover, it
is necessary to decipher the RFID password of the (B-2): This event is the probability of falsifying
Secure RFID tag so that someone may falsify the sales data on a Secure RFID tag in one month
RFID data. There are two effective attacks for using a common interrogator. This probability
deciphering an RFID password. The first is that depends on the number of books that can be
someone illegally uses a common interrogator and falsified using a decoded RFID password.
hacks into the Secure RFID tag like brute force This probability PB-2 is 100% in the case of the
attack. The second attack is that someone steals and existing method or “(1) each bookstore”
illegally uses the handy interrogator of the bookstore. because all RFID passwords in the bookstore
The FTA is shown in Figure 4. The probability of are the same. In the case of “(2) each
each event of FTA is explained below. interrogator”, this probability PB-2 is 33%
when there are three interrogators for
(B-11): This event is the probability of falsifying the inspection is three. In the case of “(3) each
RFID data of a bookstore using the RFID interrogator and arrival”, this probability PB-2
password stolen at another bookstore. In the was obtained by the following expression.
existing RFID password management method,
this probability PB-11 is 100% because an PB − 2 = 1 IN × ABN BN = 27% , (2)
RFID password is identical at all the
bookstores. In the proposed RFID password where IN denotes the number of the
management method, this probability PB-11 is interrogators for inspection, ABN denotes the
0% because the RFID passwords at least differ number of applicable stocked books, and BN
for each bookstore. denotes the number of stocked books. The
applicable stocked book means the books that
remain unsold for one month because the

451
ICE-B 2008 - International Conference on e-Business

100
conversion table
operation logic
update electronic key

falsification risk (%)

10

0.1 each
each each interrogator each
identical bookstore interrogator & arrival book
Figure 5: Relationship between the management phases and the falsification risk.

decoded RFID password is associated with because the merchandise turnover is about
those books. IN is assumed to be three and 17% and the update frequency of the RFID
ABN/BN is assumed to be 0.79 because the passwords is once a day. In the case of “(b)
merchandise turnover is about 17%, and the operation logic”, this probability PC-2 is 100%
update frequency of the RFID passwords is because the operation logic is contained in the
once a week. In the case of “(4) each book”, interrogator of the bookstore. In the case of
this probability PB-2 is 1% when the number of “(c) update electronic key”, this probability
books stolen in one month in one store is 100, can be estimated at 80% when changing the
because the number of books that can be key once a week.
falsified by using the decoded RFID password
The result of FTA for each RFID password
is only one.
management method is shown in Table 1. The RFID
(C-1): This event is the probability that an password management methods below from “(1)
interrogator will be stolen from a bookstore in each bookstore” is classified into the phases of RFID
one month. Here, this probability PC-1 is password management and the ways in which an
assumed to be 0.3%. RFID password is associated with a Secure RFID
tag. The falsification risk in Table 1 is the
(C-2): This event is the probability of falsifying
probability that falsification of the RFID data for
sales data on a Secure RFID tag in one month
resale will be successful. This falsification risk is
using the stolen interrogator. This probability
low because the security of the method is high.
depends on the ways in which an RFID
password is associated with a Secure RFID tag.
In the case of “(a) conversion table”, this
4.2 Comments of the Results of FTA
probability depends on the managing phase of
Figure 5 shows the probability that a falsification
RFID passwords because the conversion table
will be successful for each RFID password
is contained in the interrogator of the
management method. This figure also shows that the
bookstore. This probability PC-2 is 100% in the
difference between the phases of RFID password
case of “(1) each bookstore”, 33% in the case
management is larger than the difference between
of “(2) each interrogator” or 27% in the case
the ways in which a password is associated with a
of “(3) each interrogator and arrival”, as well
Secure RFID tag. We found that it is better to
as in the case of the (B-2) event. In the case of
manage RFID passwords in several phase. Moreover,
“(4) each book” this probability PC-2 is 76%

452
 RFID PASSWORD MANAGEMENT METHODS FOR FALSIFICATION PREVENTION IN BOOKSTORE
MANAGEMENT USING SECURE RFID TAGS

all the RFID password management methods are and Industry) for the eight months from August
effective in preventing falsification for resale 2006 to March 2007.
because the highest probability that falsification will
be successful for “(1) each bookstore” is 4.25% for
all the methods. REFERENCES
Table 1: Comparison of the security risk for each RFID Weis, S 2003, ‘Security and Privacy in Radio-Frequency
password management method.
Identification Devices’, Masters Thesis, Massachusetts
RFID passwords Falsification Institute of Technology.
management method Risk Engberg, SJ, Harning, MB & Jensen, C 2004, ‘Zero-
Identical 100.00 % knowledge device authentication: Privacy and security
enhanced RFID preserving business value and
(1) Each bookstore 4.25 % consumer convenience’, in the Second Annual
Conference on Privacy, Security and Trust (PST'04),
(a) Conversion (2) Each interrogator 1.42 % pp.89-101
table Tripathy, S & Nandi, S 2006, ‘Robust Mutual
Authentication for Low-cost RFID Systems’, in 4th Int.
(3) Each interrogator 1.13 %
IEEE Conf. Industrial Informatics (INDIN’06),
& arrival
pp.949-954.
(4) Each book 0.24 % Kobayashi, Y, Kuwana, T, Taniguchi, Y & Komoda, N
2007, ‘Group Management System of RFID
(b) Operation (2) Each interrogator 1.58 % Passwords for Item Life Cycle’, in Emerging
logic Technologies and Factory Automation (ETFA 2007),
(3) Each interrogator 1.31 % pp.884-887.
& arrival
(4) Each book 0.30 %

(c) Update (2) Each interrogator 1.53 %

electronic key
(3) Each interrogator 1.26 %
& arrival
(4) Each book 0.25 %

5 CONCLUSIONS
We proposed methods that use Secure RFID tags
and RFID password management for preventing
falsification of RFID data in book distribution.
These proposed methods were explained by the
point of phases of RFID passwords, and the ways in
which a password is associated with a Secure RFID
tag. We showed that these methods decreased the
probability of falsification to about 5% or less in this
situation. These results are useful for RFID
password management in bookstores.

ACKNOWLEDGEMENTS
This paper is based on the achievement of a
Japanese National Research and Development
Project, the “Secure RFID Project” that was
conducted by METI (Ministry of Economy, Trade,

453
 WORK LISTS FOR THE TRANSPORT OF PATIENTS
A Case for Mobile Applications in Health Care

Andreas Holzinger, Jürgen Trauner

Institute of Medical Informatics, Statistics and Documentation (IMI), Research Unit HCI4MED
Graz University Hospital, Auenbruggerplatz 2/V, A-8036 Graz, Austria
[email protected]

Stefan Biffl
Complex Systems Engineering Lab, Institute of Software Technology and Interactive Systems (IFS)
Vienna Univeristy of Technology, Favoritenstrasse 11, A-1040 Wien, Austria
[email protected]

Keywords: Mobile User Interfaces, Medical Workflow optimization, Mobile Computing in Hospitals.

Abstract: In many hospitals, the workflow involved in transporting patients is supported by interactive work lists on
desktop computers that provide the status of a work order and detailed information on request. However, the
nurses, who organize and conduct the transportation, need to go to the desktop in order to update the work
list status; which has been found very time consuming. In this paper we report on the development of a
mobile solution prototype that provides interactive transport work lists on a PDA and we discuss some
design and implementation issues. Despite the limited screen size, end-users were able to use the PDA with
no more problems than their usual desktop user interface as the PDA user interface can be customized with
property files to fit exactly the requirements of the end-users in the hospital.

1 INTRODUCTION work with two lists. The list contains the name and
the date of birth of the patient, for identification; the
The transport of patients within hospitals is often station where the patient lies; which type and when
supported by work lists on desktop computers (Gale the checkup is planned; and whether the patient has
& Gale, 2000). The central information is the status to lie in bed or can walk. Usually, the nurse does not
of a work order. Whilst such a non-mobile system is transport the patients in the order given on the list.
useful for work orders that do not change, the highly Each transport is arranged with the assistant at the
dynamic nature of hospital work often results to medical unit. Some changes in order are made by the
changing situations (Holzinger & Errath, 2007). nurse, for example, when more than one patient is at
Consequently, the nurses, who organize such the same station or waiting at the station where
transportations, are required to go to the desktop another patient was returned. Then they may be
computer in order to update the work list status – fetched in order to save time. However, nurses are
only to find out that their next patient is waiting not automatically notified when patients are added
where they just came from. In this context, a desktop during their shift, so either the nurses are informed
solution is time consuming, and consequently wastes about the new additions by the assistant or, less
personal resources. In our project example, at the often, by telephone.
Danube Hospital Vienna, nurses have the core
responsibility of transporting patients between the
Radiology department and other medical 2 MATERIAL AND METHODS
departments: 14 different rooms, each with a
separate list of appointments. In order to ensure platform independency of our
At the beginning of their shift, the nurse prints out mobile prototype, our first choice was Java.
the list of patients to be transported on that day. Fortunately, there is a free Virtual Machine for every
Approximately 12 nurses are responsible for Windows Mobile 2003 device (and later versions),
transportation per shift, which means that some must called Mysaifu JVM (Freebeans, 2008).

454
 WORK LISTS FOR THE TRANSPORT OF PATIENTS - A Case for Mobile Applications in Health Care

Although this is not a Java 2 Micro Edition, it is provided by SASTransferObjects. These made it
a Java 2 Standard Edition (J2SE) conformant Virtual necessary to add a function to transport them
Machine, which is built as a Source Forge project without alteration. Example: the logon data is stored
under the GNU Public License Version 2 (GPLv2). within an object but the function takes the username
For this project we used version 0.2. This version and the password as its arguments, therefore it is
supports both the Abstract Windowing Toolkit necessary to prevent direct access to this data.
(AWT) and Swing. We applied small sample On the other hand, there is the client. The newly
programs in order to ensure full functionality; the built communication method is also sub-classed
use of a standard edition JVM has, in our opinion, from the SASServerProxy. This makes it possible to
some benefits: first, it is very simple to test the change the method of communication just by
application before copying the files to the PDA. changing the properties file. The first idea was to use
Second, J2ME is not just J2ME; which means that an object stream for both directions; however there
there are many different versions highly dependant was some difficulty with this method. Generally, the
on the performance of the device (Sun, 2008), communication with objects from the client to the
(Knyziak & Winiecki, 2005). Because of these relay does not work – because it is far more
issues and of other restrictions of J2ME, the J2SE complicated as it normally is on PCs, due to the
was the best way to implement this application, in different JVMs. As already mentioned, the Mysaifu
order to provide compatibility with most of the JVM is not yet fully compatible with the Java
current systems – which are usually developed in defaults, since it calculates the version numbers and
Java (Bruno, 2005). The general idea was to use an does not allow for these to be fixed, as implemented
already available connection to the existing server. in the source code. Every class which has a fixed
The main connection type is the Bright Side version number causes an error at the relay when it
Framework (BSF) (Brightsidefactory, 2008), which is de-serialized. The InvalidClassException, states
transports XML content over the HTTP protocol that the serial versions are different but in reality
with dynamic instantiation and invocation of they are the same and only one number has been
available interfaces. However, in such specialized wrongly calculated. The solution for this is to
circumstances the JVM for the PDA is not fully change the version number before de-serialization.
compatible with the version for the PCs. The other Any data which is sent through a stream is sent as a
connection was based on Remote Method Invocation series of bytes. In an object stream, this is done with
(RMI), facing the problem that dynamic interfaces a special protocol which is publicly available (Sun,
were not supported on the PDA. This required us to 2004). There is an exact description of how the class
develop another connection for the PDA. Due to the and the serialVersionUID are printed out. Some
communication protocol should not be changed in general information is followed by the class and
the first version we built a relay application. super class information, which is succeeded by the
class name printed in UTF. Directly after the name,
the version number is written to the next eight bytes,
3 BACKEND LESSONS followed by some flags and the variables, which of
course can also be classes.
LEARNED Once the object serialization stream protocol is
known, this ceases to be a problem, since the right
A new concept was necessary, employing a relay, in version numbers are known by the relay.
order to establish the connection to the server. On After the serialVersionUID of any class defined
the server side the connection was built using RMI in the properties file is changed, the objects from the
or BSF (Sims, 2004), while, on the PDA side the received byte stream can be regenerated and it is
communication is based on an object stream. The then possible to send a SASTranferObject, including
BSF and the RMI connections were given the means classes, from the PDA client to the relay. However,
to access the data on the server. Both in the other direction, the communication from the
communication connections had to be adapted to fit relay back to the PDA client is less simple.
the requirements of this application. The The reason for this is mainly that the JVM on the
communication is no longer made over a static PDA throws an InvalidClassException for each
connection; this is due to the fact that the relay has class. Theoretically, it would also be possible to
to be able to open many connections. On top of this, change the version numbers on all the classes
some additional functions were necessary. The most transferred on the PDA, practically, there are too
interesting change is the function which returns the many and the cost would be too high; therefore it
sessionID. The communication process between the was decided to send strings as a series of bytes. A
relay and server, and relay and PDA client, is

455
ICE-B 2008 - International Conference on e-Business

very simple protocol was invented where the string is 32 bytes, the maximum supported by this
is divided into fields with field terminators. The algorithm. This is acceptable because, although the
PDA splits up the field and rebuilds the resulting set. calculation time needed for the encryption,
This method has proved successful. The decryption and the data lookup on the server is up to
disadvantage for the relay, was the necessity of 5 seconds, the old list is left on display until the new
rethinking the data encryption technique, however, one is loaded. It is possible that the customer might
this actually turned out to be an advantage. Since the decide that no encryption on top of WPA encryption
communication is built up with such easy methods, would be needed and change this in the property file
the necessary number of libraries decreased and now to save time. It is also possible to change the whole
requires less than 100 KB. algorithm, because this part was built to support
The WPA encryption, which is supported by the other algorithms. The test with non-Standard English
WLAN hardware, is generally sufficient to secure characters, such as the German “äöüß”, was also
the transmitted data. However, since the client on interesting. These characters are encoded differently
the PDA is used within a hospital and includes the on the PDA. For example, on the PC the “ü” is
transmission of sensible patient data (Weippl, represented by a byte with the value of -4. On the
Holzinger & Tjoa, 2006), (Miller, 2004), PDA, the value for the same character is -127. Other
(Mupparapu & Arora, 2004) we decided to include languages with additional character tables will have
additional encryption. The first technique we similar problems. To add support for these
considered was an asymmetric method: its security languages, an adjustment of the characters can be
has remained unbroken up until now. However, as made after the transfer. Which changes have to be
we practically expected, a performance test of the performed can again be specified in the property file
RSA algorithm showed that this was not possible in which contains all the general settings. This makes it
our case. For the test, a very short stream of about 50 simple to adopt the system for many different
bytes was used. The decryption of these few bytes character sets.
took almost a second. Within the real application,
the amount of data sent is often more than one kilo
byte. The time delay caused by this encryption and 4 FRONTEND LESSONS
decryption would be unacceptable. Other
asymmetric algorithms are as slow as the RSA LEARNED
algorithm, which also made them unsuitable for our
client application on the PDA (Salomaa, Rozenberg At first the nurse logs on; the server then checks the
& Brauer, 1996). An additional, very fast, symmetric logon information. The BSF communication
encryption, which has not been patented, was framework provides the necessary security. After
implemented in order to ensure against a possible logon, the transport list is shown with the default
deactivation of the WPA encryption. There are many filter. In order to provide the ability to change this,
symmetric encryption algorithms. The One-Time two option panels are available. Our main design
Pad would have been best, but it is not practicable. criteria was to build the mobile interface as similar
DES and 3DES are old and also quite slow. IDEA, as possible to the existing non-mobile interface on
RC5, RC5a and RC6 are protected through patents. the PC, which the nurses are used. We followed
A5 is already broken. The result of this is that only previous experiences on mobile interface design
RC4, Blowfish, Twofish and AES were suitable. Of (Holzinger, Sammer & Hofmann-Wellenhof, 2006),
course, there are some other (proprietary) (Holzinger, Searle & Nischelwitzer, 2007),
algorithms; however, AES is a common and very (Nischelwitzer et al., 2007) and general experiences
fast algorithm. on usability engineering (Holzinger, 2005) in order
This new standard algorithm has already been to ensure an end-user centered user interface. On
analyzed for security vulnerabilities, which mobile interfaces, generally the keyboard size
practically ensures that there is no chance of problem refers to the fact that using the on-screen
cracking the algorithm within the next few years. keyboard significantly reduces the remaining screen
Therefore, the AES algorithm was chosen. The size. The log on and the Options panel use Scalable
implementation of the algorithm was not done Vector Graphics (SVG). The layout manager expects
within this project. A free implementation of the the whole screen to be available for the widgets.
ACME Laboratories was chosen to encrypt the data. Constraints specify the place and the position as a
In this project, it is used with a 16 byte (128 bit) key percentage of the screen. This generally worked, due
which is calculated from a pass phrase which has to to the independence from the physical screen size.
be specified in the property file. The used block size However, when the keyboard is selected, the
program is notified to refresh its screen – to fit the

456
 WORK LISTS FOR THE TRANSPORT OF PATIENTS - A Case for Mobile Applications in Health Care

new, smaller screen; this new rendering causes each used for the okay and cancel buttons. An X in the
component to become smaller, which makes the title bar means cancel and the symbol for okay is
options panel very difficult to read. After the simply represented by the two characters ‘OK’.
removal of the keyboard, the PDA must again adjust Because of this, the cancel button generally will not
the size of all components. This takes approximately be used very often; and therefore could be removed
3 seconds, slowing down the overall application to the title bar. Experienced users will be acquainted
time considerably, which was perceived as with this. Also, inexperienced users will not have
extremely unpleasant by our test users. This problem many problems because they normally only need the
can be solved by using another layout manager, OK button which is same as on the PC. When the
however, the start up problem still has to be solved, cancel button is needed, a short glance over the rest
where we followed a three step approach: of the screen will solve any difficulty that may arise.
The first step is to review the needed libraries. The options scroll is really one of the bigger
Maybe not all classes of the library are needed. A problems. This view shows the different options
new library should be built, which only contains the which the nurse can set up. As already described in
necessary classes. The reason for this is simple, to the options paragraph above, every time the user
shrink the size of the libraries which have to be activates a scrolling action the whole window must
loaded, thereby decreasing load time. This is not be redrawn. If there are only a few components on
very difficult, since the class files contain the screen, then the PDA’s processor is sufficient to
information of their super class and interface, the perform the necessary actions of the layout manager.
dependency within a package can be discovered and Therefore, it is better to split the panel up into two
the methods’ argument types can also be seen. windows, or panels, so that the scrollbar is no longer
The second step is the loading process itself. needed. A button is necessary to switch between the
This prototype loads all elements which are shown two panels. This makes sense because in this way it
on the display at start-up. This mean that all three is possible to put the more important widgets on one
panels, the logon, the transport list and the options, panel and the remaining components on the other
are all loaded before anything is painted to the panel, which may not need loading as often. Another
screen. A better solution would be to load only the possibility is to reduce the number of possible
first panel, where the user enters their username and options to a minimum by removing the less
password. To do this, the user needs a few seconds important ones and therefore the necessity for a
to enter their account details: meanwhile the second panel. Other known errors refer to usability
transport list can be loaded in the background. The difficulties rather than real problems, avoiding these
benefit of this is simply that the user can start to makes it easier for the user to work with the
work after a shorter loading time. From the user’s program. A typical example includes the information
point view of while using the application, this does displayed about the last update or a logout button.
not make a big difference as the next panel is The last update is necessary because if the nurse
already loaded when it is needed. The same strategy is somewhere in the hospital without a wireless
can be used for the options panel. While the user connection to the server it is not possible to update
looks at the screen and uses the functions of the the transport list. And because the nurse usually gets
transport list panel, the options panel can be loaded. an update of the list every five minutes it is possible
The only disadvantage of this technique is from the that they may think that the program does not work
programmer’s perspective: it is more work to load anymore because the list had not changed for more
something in the background than it is to load than ten minutes. In this case, the time of the last
everything at once. In this case, it is necessary to use update indicates a connectivity problem, which
separate threads to control which panels will be requests that the nurse move towards an area with a
needed next and to check whether it is already wireless connection. This will mainly be an issue at
loaded, or to load it before it is needed, in order to the beginning, later support will be expanded to the
avoid delay. We had to ensure that the system waits entire hospital. The logout button is a time saving
until the required panel is loaded; in real-life this service for the nurses to enable them to share PDAs
will happen very rarely. without the necessity of restarting the PDA
The third step is to use separate frames for each application each time it changes hands.
panel. This helps to reduce the required memory, When the nurse goes off duty they just log off
because after the user is logged in, the logon frame and the next nurse can login immediately. Another
can simply be removed from memory. thing which is not directly an error is the cancel
The not default mistake means that it is default button. Because this button has been moved to the
on the PDA that the close button in the title bar is title bar it is no longer a real widget which needs

457
ICE-B 2008 - International Conference on e-Business

space within a panel or window. So the okay button the application. In the top left of the window, there
should be moved to the middle or to the right of the is a button to select the direction. The title of the
available space to fit the user’s expectations. Our button shows the currently selected direction.
last idea was to use different panels for the different Possible values are “To Radiology” and “To
screens and to perform any cancel actions using the Station”. By pressing it, the direction is changed. On
“X” in the title bar. However, this was not possible. the right of this button, there is a label which tells
The reason for this is simple, that the default action the user when the last update of the list was made.
on the PDA is to hide the window and to show the This is necessary because of the automatic update of
next one which is below the current window. On the the list. The time interval of the update is specified
PC it is possible to execute own code when the in a property file of the client. On starting, the list
closing button is pressed. On the PDA this is not consists of the patients name, the time to transport,
possible because no event is generated to perform an from where and if the patient can walk, which are
action which would prevent the window from being displayed in two rows. After the selection of one of
hidden. So some changes were necessary. The first the items, the item pops out and shows additional
was that every panel gets its own window, except information, such as the patient’s date of the birth.
the two option panels, which share one. The options When one patient is transported, the nurse double-
window is the only panel which has a default cancel clicks on the patient in the list to mark this as
operation, performed by pressing the “X” on the title transported. This is shown by the light gray
bar. This is possible because then the transport list background. After selecting another item the marked
window, which is the window below, is shown. The item pops in and the color of the font of the selected
problem that switching the windows takes longer item turns dark gray. By the next update of the list,
than switching the panels could also be eliminated the update will have been completed on the server
by preloading the windows. If a window is needed, it and the item is removed from the list. It is possible
is just shown on the screen because it is already in to declare a minimal time in the property file, during
the memory, which prevents the shimmer. The other which an item is not updated to prevent mistakes, it
two windows no longer have the close button in the is only necessary to make a second double-click on
title bar to prevent user mistakes. Standard buttons the marked item in order to change the item state
are used to perform the needed actions. This means back to normal.
that the exit button, which was just removed, is
again added to the logon panel. Also the logout
button within the transport list panel can not be
removed. In the transport list panel, one very
important button is absent. There is no way to
change the direction of the transportation. The
patients have to be transported to the radiology and
they must also be transported back to their stations.
This means that two lists must be displayed. To save
space on the screen the button is added instead of
using one options button. The loss is that there is
now no way to get directly to each of the options
panels. To minimize the effects of this disadvantage
Figure 1: Transport List: Before and after performing a
the options dialog always shows the panel which transport.
was used the last time. A next/back button is used to
switch between the two options panels. Other
When the button “Filter” in the transport list
modifications which are needed are the logout and
panel is pressed, the first options panel is shown.
direction buttons and the change of the last update
The nurse can specify different options for the filter;
time string. The change of the two buttons has two
e.g. the transport type or the examination date. The
reasons. The first is that the application should be
transport list button behaves as a back button to the
consistent. In the options dialog, the position of the
transport list panel; when pressed the expected panel
OK button to go back to the transport list is at the
is shown and the contained list is updated to fit the
bottom left. The logout button is the same within the
current filter settings. The “Next” button is used to
transport list window. The second reason is that the
get to the second options panel. The “X” is abort.
direction is like a headline. And so its logical
The second panel is built as the first one. The whole
position is at the top of the window. An extra
available space is used for components which are
window is used for the list, which is the main part of
used to specify different filter criteria. Examples for

458
 WORK LISTS FOR THE TRANSPORT OF PATIENTS - A Case for Mobile Applications in Health Care

filter options on this panel are the patient name or factory.org/components/remotingDoc/architecture.htm

the handling of already completed transports. At the l, last access: 2008-06-10
bottom of the panel there is again a button to return Bruno, E. J. (2005) NetBeans 4.1 & Eclipse 3.1 -
to the transport list and update it with the specified Development platforms for J2SE, J2EE, J2ME. Dr
Dobbs Journal, 30, 8, 14-23.
options to match the selected filters. The “Back” Freebeans (2008), Mysaifu JVM. Online available:
button switches to the first options panel. http://www2s.biglobe.ne.jp/~dat/java/project/jvm/inde
x_en.html, last access: 2008-06-10
Gale, M. E. & Gale, D. R. (2000) DICOM modality
5 CONCLUSIONS worklist: An essential component in a PACS
environment. J of Digital Imaging, 13, 3, 101-108.
Holzinger, A. (2005) Usability Engineering for Software
It is not trivial to develop both a useful and usable Developers. Comm of the ACM, 48, 1, 71-74.
mobile application. The general idea to use a J2SE Holzinger, A. & Errath, M. (2007) Mobile computer Web-
Virtual Machine was abandoned, due to lack of application design in medicine: research based
compatibility. The server communication could not guidelines. Universal Access in the Information
be done trough the available methods. This made it Society International Journal, 6, 1, 31-41.
necessary to build a relay to forward the inquiries to Holzinger, A., Sammer, P. & Hofmann-Wellenhof, R.
the server. That made it possible to convert the result (2006) Mobile Computing in Medicine: Designing
Mobile Questionnaires for Elderly and Partially
to a string which is sent as a byte stream. Due to this
Sighted People. Springer LNCS 4061. Berlin, New
sensible data, encryption was necessary. Primarily York, 732-739.
the WPA encryption is used, which is secure and is Holzinger, A., Searle, G. & Nischelwitzer, A. (2007) On
supported by the PDA’s hardware. Additionally, the some Aspects of Improving Mobile Applications for
data is encrypted by the application. After a the Elderly. Springer LNCS 4554. Berlin, Heidelberg,
comparison the AES implementation of the ACME New York, 923-932.
Laboratories was deemed the optimal choice. Due to Knyziak, T. & Winiecki, W. (2005) The new prospects of
the fact that it is the new standard algorithm for distributed measurement systems using Java (TM) 2
symmetric data encryption, the algorithm can be Micro Edition mobile phone. Computer Standards &
Interfaces, 28, 2, 183-193.
graded as secure. The user interface of the PDA
Miller, A. (2004) PDA security concerns. Network
client also caused some troubles. It is not possible to Security, 2004, 7, 8-10.
build a slightly more complicated interface when Mupparapu, M. & Arora, S. (2004) Wireless networking
using the default layout managers. It was necessary for the dental office: current wireless standards and
to program a custom interface to fit all our needs security protocols. J Contemp Dent Pract, 5, 4, 155-
within an adequate time. Another problem was that 162.
the PDA had a different window handling. It is not Nischelwitzer, A., Pintoffl, K., Loss, C. & Holzinger, A.
possible to receive any window events within a Java (2007) Design and Development of a Mobile Medical
application and that must also be considered. Application for the Management of Chronic Diseases.
In: Springer LNCS 4799. Heidelberg, Berlin, New
Another big problem on the PDA, which is not an
York, 119–132.
issue on the PC, is the scrolling of complicated Salomaa, A., Rozenberg, G. & Brauer, W. (1996) Public-
interfaces. The scrolling time is much too long Key Cryptography.
because a simple scroll request required a few Sims, B. (2004) Moving from liability to viability.
seconds of time to perform. Everything had to be Hospitals, health plans and physician practices can
optimized to save as much computing time as outsmart hackers with policy, a comprehensive
possible. However, some of the challenges of security infrastructure and wireless monitoring. Health
migrating desktop applications to mobile technology Management Technology, 25, 2, 32-35.
are the limited screen size and the originality of the Sun (2004), Object Serialization Stream Protocol.
http://java.sun.com/j2se/1.5.0/docs/guide/serialization/
technologies that need to work together correctly for
spec/protocol.html, last access: 2008-06-10
a useful and usable solution. In conclusion, the PDA Sun (2008), CLDC HotSpot™ Virtual Machine.
prototype can reduce the time needed for transports http://java.sun.com/j2me/docs/pdf/CLDC-
and the end-users were able to use the PDA similarly HI_whitepaper-February_2005.pdf, 2008-06-10
to their non-mobile interfaces. Weippl, E., Holzinger, A. & Tjoa, A. M. (2006) Security
aspects of ubiquitous computing in health care.
Springer Elektrotechnik & Informationstechnik, e&i,
123, 4, 156-162.
REFERENCES
Brightsidefactory (2008), Bright Side Framework
Overview. Online available: http://www.bs-

459
 TECHNOLOGY VS BUSINESS NEEDS IN BUSINESS
INTELLIGENCE PROJECTS

Ana-Ramona Bologa, Razvan Bologa

Faculty of Economics Cybernetics, Statistics and Informatics, Academy of Economic Studies, Bucharest, Romania
[email protected], [email protected]

Adela Bara
Faculty of Economics Cybernetics, Statistics and Informatics, Academy of Economic Studies, Bucharest, Romania
[email protected]

Keywords: Decision Support Systems, Business Intelligence Projects, ERP systems, organization changes, critical
success factors, evaluation criteria.

Abstract: This article presents the development of a real industry business intelligence project, in a company that used
an ERP system. We have focused on the main factors that influenced and affected the project development
and it analyses the system evolution from technical point of view, the solutions found to fulfill the changing
requirements, but also the impact of the company reorganization on the project. This particular experience
can be of use to all those who are involved in building business intelligence solutions to reveal some of
success factors, to prevent or to solve some of the inherent problems related to this type of projects.

1 INTRODUCTION finally we’ll formulate some lessons learned of use

for future implementations.
Many organizations use modern ERP systems for
transaction processing and reporting. Information
from applications within an ERP system is managed 2 BI ENVIRONMENT: THE ERP
by a relational software database. Managers and SYSTEM AND THE BUSINESS
executives need information for strategic and tactical
decision that often requires the combination of data REQUIREMENTS
from ERP and non-ERP application sources. The
ERP’s database complex schema is not adequate for In (Larissa T. Moss, 2001) we can find out that
data analysis and reporting. Business Intelligence 60%-70% of business intelligence (BI) applications
technologies can provide the best solution for fail and these failures are not related only to the
extracting, transforming and presenting the huge technology but to organizational, cultural, and
information from within the database. infrastructure issues.
The specific aspects may come from the way of The ERP project apparently started on a solid
implementing the BI project. In our case, the BI ground, with Oracle Financial (FI) module
project was implemented by the same developer that implementation at the central unit. The ERP project
implemented the ERP system, and it was strongly then continued by implementing Oracle FI in all
related to the ERP project implementation, which company branches, sectors and working points.
continued to be extended along with our BI project. The implementation generated the need for
This paper describes a BI project experience processes reengineering and revealed important
developed in the context of implementing Oracle flaws of the company structure and management. In
ERP solution into a big company activating in order to complete these tasks, many changes had to
Romanian oil industry services, with a complex be made. These changes affected company structure,
organizational structure, having over 15000 business information and material flows and the way
employees. We’ll analyze the difficulties and the they were registered by accountability, management
problems met as the project was developed, and

460
 TECHNOLOGY VS BUSINESS NEEDS IN BUSINESS INTELLIGENCE PROJECTS

and reporting hierarchy. It was also considered management tool. The entire structure of incomes
implementing the Logistics and Work over modules. and outcomes categories and their corresponding
In this context, the BI project started when the accounts were changed in the middle of the project.
financial module had already been implemented in In the following sections we will focus on the
the central quarters. In the beginning, the project technical and organizational structure aspects.
seemed to be a classical one, and we followed the
classical phases described in (Larissa T. Moss,
Shaku Atre, 2003): Planning, Business analysis, 3 BIS’S DEVELOPMENT PHASES
Design, Construction and Deployment.
During project stages we took into account some The system described in this article includes the use
specific in modeling BIS (Business Intelligence of the following technologies: data warehouse
Systems), such as:
realized with aggregate data and virtual model,
• BIS are oriented on business opportunities OLAP technology, data mining facilities, analytic
rather than transactional needs; SQL functions. The interface is based on Oracle
• BIS have to implement tactical and Business Intelligence Portal that integrates all these
strategically decisions, not only technologies and offers direct access to the system
departmental or operational decisions; from anywhere, anytime.
• BIS analysis is focused on business needs.
This stage is the most important of the First Phase
process; The cost factor was decisive when deciding
• Development process is cyclical, focused on which of two options to adopt:
evaluation and improvement of successive • A traditional data warehouse, having aggregate
versions, not only building and major data, involving an extended data volume, both
delivering of a singular final version. elementary and aggregated.
• A virtual data warehouse, which worked
In order to respond to all business needs, the directly with data from the ERP database, much
phases of the project were completed almost entirely easier to implement.
twice: for the initial company structure and for the For this phase, we used Oracle Discoverer to
modified company structure. During this process, build a virtual data warehouse over the database.
the BI team faced three important types of problems: Multi-dimension structures like dimensions and fact
1. Technical Problems – the initially allocated tables were transformed and mapped from relational
resources were reduced, as the initial estimations of sources. We started with low data volumes,
the data volume were overcome as soon as the BI generated by the use of the FI module at the central
module was implemented in the entire company. It quarters. Oracle Discoverer Desktop was installed
was estimated that a virtual data warehouse would on the computers of the main decision makers that
be sufficient, but the reporting and analysis needs were going to use it. We developed reports for
were higher, so the performances of the initial analyzing Cash Flows, Budgets, Costs, Expenses
solution had to be somehow improved. and Incomes only for the central head quarters. The
2. Communication Problems – there was a lack number of rows involved in these reports was in
of availability from the company personnel, as they average 100000/month, so we did not need a large
were already overwhelmed by tasks related to the amount of resources.
ERP implementation and by activities related to the
consultancy companies’ cooperation. Second Phase
3. Organization changes Problems – the BI When the financial module was installed over
project was from the very beginning under the sign the entire company, we faced the problem of having
of change. Beside the changes due to technical to go to each of the sectors to install the product and
aspects, drastic structure changes in the company instruct the managers how to use it and some other
negatively influenced the relevance of the initial IT persons how to ensure the maintenance and future
business analysis, and often made some of the development. Inn order to minimize the effort, we
developed reports irrelevant or inapplicable. integrated Oracle Discoverer with Oracle
The work was performed with gaps, as there was Application Server Portal 10g Release 2, and the
always some top management decision to make that solution became available on-line, allowing
affected the BI project. The initial plan was revised everybody to access information according to their
over and over, so it somehow lost its power as established privileges. But, this decision had a
negative impact on performances: the working speed

461
ICE-B 2008 - International Conference on e-Business

decreased as the required memory space increased designed one central data warehouse with sub-data
from 32 GB server shared as Discoverer needed, to marts for financial, production, purchasing orders
64 GB server shared as Oracle Portal needed. and order entry modules. The final solution is
We didn’t manage to get a dedicated server for described Figure 1. The number of rows involved
Oracle Application Server because of high hardware was now over 3 millions/month, and the
resource cost. In this phase, we had to develop performances were finally satisfying.
reports for analyzing Cash Flows, Budgets, Costs,
Expenses and Incomes for the central head quarters
and ten sectors with different area of production. So, 4 EVALUATION OF THE FINAL
the number of rows involved in these reports was
over 1, 5 million/month. Under the circumstances, SOLUTION
with low allocated resources and a large amount of
data per month we needed to find another solution to 4.1 Organization Changes Impact
increase the speed and reduce the loading time for
reports. Major business reorganization is one reason BI
project typically get hosed (Mitchell A. Ocampo,
2007). In this project we faced an important business
restructuring. More, the initial scope estimation
proved to be inexact, and it had to be re-evaluated,
so the project gained new dimensions.
Most of the analysis results had to be re-
ORACLE PORTAL ORACLE APPLICATION evaluated, and database design and repository design
SERVER had to be revised and seriously extended, having an
BI REPORTS
ORACLE PORTAL
important impact over technical and human resource
requirements. The BI team had to be enlarged and it
ORACLE DATA
was itself a problem, as there were no other BI
WAREHOUSE AND
specialists available, and other two new persons had
DATAWARE HOUSE ORACLE to be employed and integrated.
DISCOVERER Many of the reports developed in the first phase
DATA MINING were inapplicable, generating important time loss
and inefficiency; it often happened that after getting
complete specifications, data gathering, and
ORACLE E-BUSINESS completing of the report design, during the
ORACLE DB SUITE
validation meeting we were communicated the
whole perspective was changed. This had impact not
Figure 1: The system’s architecture in the final phase. only directly upon efficiency, but our team members
got to be confused or demoralized.
Third Phase But, there was a bright side in all those changes.
When the other ERP modules started to be As the consultancy company progressed in
implemented, we were facing serious problems. We elaborating the new business view, they inevitably
decided to construct an enterprise data warehouse brought out the reporting needs. Our team was
integrating Oracle Data Warehouse Builder and invited for discussions, and we had the chance of
Oracle Data Mining into Oracle Portal. Unlike proving the great advantages of BI over the existing
Oracle Discoverer, in Warehouse Builder data are reports. A demonstration based on the customized
physically stored into a special repository, build on reports had a relevant impact.
the database. The role of Oracle Data Mining was to There was also a change in the attitude of end-
supply additional analysis power. users, previously reluctant to changing their old
This solution was a success as data were tools (Microsoft Excel and other Office applications)
aggregated in the central data warehouse, the ETL and to learning how to use a new one. They were
(Extract, Transform and Load) process was built off- now eager and motivated as the old reports were no
line, not affecting the reporting analytical time. We longer valid. And the truth is that they come to love
built new types of reports related to production it in very short time, as the new reports were very
(costs and allocated resources), work over, flexible and easy to use and they interfaced directly
purchasing orders, transactions per account, etc. We with the ERP application and other data sources.

462
 TECHNOLOGY VS BUSINESS NEEDS IN BUSINESS INTELLIGENCE PROJECTS

The end-user involvement is a very important 5. Friendly user Interface

factor in developing functional BI solutions. As BIS should be designed to provide managers a fast,
project requirements and scope are set by users, they easy, and understandable way to navigate into data
know exactly the data they need and they will test and identify trends and patterns. BIS should allow
and use the result of the project. There are voices the user interface to accommodate different degrees
claiming there was a shift of authority and of technical knowledge.
ownership in BI projects, from IT to business side We notice these criteria are related not only to
(Larissa T. Moss, Shaku Atre, 2003). technical aspects, but also to business aspects, as
managers offer the most important feedback to the
quality of business intelligence systems.
4.2 Criteria for Evaluating Business
Intelligence systems
5 CONCLUSIONS
Deploying BIS involves many risks: system design,
data quality, and technology obsolescence. Large The most important conclusion of this real project
budgets and strategic information are at steak in experience is that, more than technical reasons,
deploying BIS systems, a good reason to establish organizational and infrastructure dysfunction
rigorous criteria for evaluating this type of system. endanger the success of a project. When there are no
These criteria are discussed below. supporting business decisions, the BI project is
1. Decisions based on Business Process paralyzed. So, the main factor in conducting such a
BIS should not be viewed only as a data repository project is business rather than IT. Following this
or a large set of data. Instead, system’s direction, after our project implementation was
implementation should be concern with ready, the maintenance and further development of
conceptualizing new data models, processes, and BI reports were entrusted to some end-users,
indicators that form the content of the BIS. specially trained (non IT).
Only combining BI technologies on the top of an
2. Performance
Enterprise Resource Planning system will never
It reports to the response time the system provides
provide a viable response to the changing
to its users, depending on the complexity of the
management needs. A competitive advantage can be
database and the queries being requested. Most
obtained only by adding business-oriented
responses should range from few seconds to a
personalization so that the information provided can
maximum of 30 seconds for routine queries.
fulfill the particular needs of end-users and empower
3. Flexibility and Scalability dynamic analysis and decisions.
Flexibility determines whether a BIS solution can
adapt to changing business conditions after the
system’s delivery. BIS should be able to
accommodate changes in any type of business
REFERENCES
processes and functions, as well as in new laws and
Kaniclides A., Kimble, C. 1995. “A Development
regulations, requiring the capture of different types Framework for Executive Information Systems”,
of data. BIS should be expandable to accommodate Proceedings of GRONICS '95, Groningen, the
data growth and changes in organizational structure, Netherlands, Ed T LOURENS;
without a slowdown in performance. Larissa T. Moss, Shaku Atre, 2003. Business Intelligence
Roadmap. The Complete Project Lifecycle for
4. Integration Decision Support Applications, Addison-Wesley
Integration involves two types of issues: data Professional
integration and system integration. BIS will be Larissa T. Moss, Organizational Barriers to Business
particularly effective if it can overcome the Intelligence (Part 2), http://www.businessintelligence
challenge of information fragmentation, allowing .com /ex/asp/code.135/pagenum.1/xe/article.htm
managers to measure features of business processes Larissa T. Moss, 2001. Organizational and Cultural
that involve information from inside and outside of Barriers to Business Intelligence, Cutter Consortium,
the organization. System integration involves two Executive Reports, http://www.cutter.com
Mitchell A. Ocampo, 2007. Overcoming the Barriers to
aspects: the ability to extent the BIS software with Business Intelligence Success, EMC Corporation
new capabilities and modules and the system’s Publications, http://www.emc.com/solutions/microsoft
ability to coexist with other enterprise solutions. /pdf/H2740_overcom_barriers_bus_intell_success_ep_ldv.
pdf

463
 WEBSITE INTERACTIVITY
e-Commerce Usability Perspectives in Indonesia

Vincent Didiek Wiet Aryanto

Graduate School of Business Management Soegijapranata Catholic University, Semarang, Indonesia
[email protected]

Keywords: Website, interactivity, usability.

Abstract: Usability is a key factor of website that is commercially successful. Interactivity has been linked to website
usability. This research confirms that improving website interactivity promotes website usability. In
addition, the extension of previous research such as usability sub construct such as learn ability,
memorability, efficiency, user error and self-efficacy are validated to improve the body of knowledge
particularly on e-commerce research.

1 INTRODUCTION When the consumers is interested, he or she

clicks on one of these advertisings and begin to
The use of e-commerce alleviates Indonesian textile control the communication process, hence,
industries nowadays that are beset by economic advertising exposure is able to active (Chaterjee et
crisis. As a matter of fact, using e-commerce al, 2003). Websites, therefore, represent the most
increases firm’s revenue up to twenty percent. important form of interactive advertising. Websites
Consequently, it increases firm’s efficiencies in are based on information and communication
production and marketing as well (Kompas, 2004). technologies enable easy and quick interaction
The impact of internet on marketing and/or between consumers and marketers (Coyle &
advertising has been very important in current years, Thorson, 2001). In a website, individuals can
and it continues to proliferate (Ha, 2003; Macias interact with the medium itself, which is called
2003). The internet exposes to advertising, giving “machine interactivity”. The machine interactivity
consumers the discretion to participate to particular allows consumers to control what information will
messages within the medium (Klein, 2003). be presented, in what order, and for how long.
Marketing has changed dramatically due to the (Ariely, 2000).
development of the internet (Strauss and Frost, Website interactivity is a design factor that has
1999) and since its inception; online marketing been indirectly linked to website usability.
including e-commerce has seen rapid change Furthermore, interactivity is defined as mutual
(Rowley, 2001). Growing online competition and communication between at least two entities, with
maturation of internet technology have underpinned website interactivity generally involving
the expansion of website factors, beyond extensive communication between a user and the website (Liu
product offerings, customer convenience, ease of & Shrum, 2002). Research on website interactivity
navigation, and security, that affect online marketing is still in its preliminary stages, and there is a
success (Choate, 2000; Liu et al, 2000; Lohse et al; theoretical gap exists in describing and foreseeing
2000). the linkage between usability and interactivity.
Unfortunately, this is not true for most formats of Seemingly, it may be intuitive that websites with a
internet advertising, advertising formats such as high extent of interactivity are more usable than less
buttons, spams, banners, hypertexts, pop-ups; bars extent of interactivity, there is little theoretical and
are not requested by the receiver of the messages. empirical proof to support this notion. So far,
These advertisements are commonly incorporated in research indicates that interactivity can be useful,
portals, search engines, or media websites, reach value-neutral, or detrimental to a website, depending
large audiences (Ha, 2003). on the situations. So, understanding how
interactivity affects usability is essential to the

464
 WEBSITE INTERACTIVITY - e-Commerce Usability Perspectives in Indonesia

progress of research on usability in e-commerce communication. This assertion is particularly

websites. (Liu, 2003). relevant when considering work that takes place
Usability literature provides guidelines for over the WWW (World Wide Web) and related
improving website usability; however, many of technologies. In the technology point of view,
these guidelines are not supported by theoretical and interactivity concept includes both human beings
empirical evidence causing them incorrect, and and computers as parties in an interaction.
inapplicable in some contexts. Accordingly, many
usability guidelines largely ignore interactivity 2.2 Website Interactivity
aspects of user engagement. Thus, practitioner-based
usability guidelines will not necessarily provide Website interactivity is pondered to be a subset of
appropriate levels of website interactivity. (Chen & interactivity that focuses on the interaction between
Yen, 2004). websites and users. In this regard, website
In response to the increasing demand for e- interactivity is broadly defined as any action a user
commerce, it becomes more important to promote or a website takes a time, while a user is
recent usability literature with research based on predetermined end-user goal with the website, such
validated theory and empirical data. This research is as accessing information-whether users are seeking
based on the improvement of the previous research specific information or just browsing. Other goals
undertaken by Lowry et al; 2006. That suggested to may include business transactions or even
take into account on usability sub construct such as entertainment (Stomer & Galley, 2000).
learn ability, memorability, efficiency, error Interactivity of a website offers facilitated
prevention and self efficacy as suggested by Liu and communications, customization of presented
Shrum (2002), aside from satisfaction, this research. information, image manipulation, and entertainment
In response to the existing gaps in interactivity for the customer. (Mathwick, 2002). The interactive
literature we set forth some research questions: nature of websites has been credited with increasing
• Should perceived measures of two way the desire to browse and purchase online (Li et al;
communication, active control and 2001).
synchronicity highly correlated with each According to Liu and Shrum (2002) a two-way
other? communication, active control and synchronicity
• Is there any positive correlation between mold the basis of a powerful, yet simple. Further
website interactivity and website usability? development in two way communication refers to
• Is there any positive correlation between bi-directional flow of communication between
website interactivity and website usability sub communicators. (Liu, 2003). Active control is
construct satisfaction? present when a user is given the ability to choose
• Is there any positive correlation between information and guide the interaction.
website interactivity and sub construct self Synchronicity indicates the timing of information
efficacy of user? exchange; a more concurrent or simultaneous
exchange between two entities is a more
synchronous interaction than an exchange with a
longer time lag between responses. These three sub-
2 ITERATURE REVIEW constructs of interactivity form the basis for
exploring the relationship between website
2.1 Interactivity interactivity and website usability.
The effects of fulfilled and unfulfilled
Most of the definitions of interactivity particularly expectations and desires of interactivity in websites
refer to two-way communication and interchange in are preliminary supported by website interactivity
which two or more parties send related message to research. Wu (1999) found out a strong, positive
each other (Liu & Shrum, 2007). Pavlik (1996) correlation between positive perceptions toward a
defines interactivity as a process of reciprocal website and interactivity. Liu and Shrum (2002)
influence (p.135), while Alba et al; (1997) proposed provide a literature review and theoretical
that two dimension of interactivity generally focuses proposition to support the notion that two way
on face-to-face interpersonal communication, communication, active control and synchronicity
however, Ha & James (1998) contended that the improve interactivity and communication with
definition of interactivity based on interpersonal websites. Lowry et al; (2006) research’s to students
communication is too restrictive to fit technical taking Management Information System

465
ICE-B 2008 - International Conference on e-Business

coursework, supported Liu and Shrum’s theoretical H3 There is positive correlation between website
proposition that two way communication, active interactivity and website usability sub construct
control and synchronicity have strong positive satisfaction.
correlation with website interactiveness. H4 There is positive correlation between website
interactivity and sub construct self efficacy of user.
2.3 Website Usability
The more usable websites, the more positive attitude 3 RESEARCH PURPOSE
towards online marketing or e-commerce and
increase repeated visit rates, while websites with low This research entitled: Website interactivity: E-
usability have the opposite effect. Nielsen (2003) commerce Usability Perspectives aim at exploring
defines usability in terms of learn ability, efficiency, the following:
memorability, user errors and satisfaction.
• The perceived measures of two way
Concerning website usability, Lowry et al; (2006)
communication, active control and
applied on his research, Nielsen’s theoretical
synchronicity correlate with each other.
framework, but only for the satisfaction variable,
• The correlation between website interactivity
since the researcher believed that satisfaction
and website usability.
triggered website usability and ultimately, satisfied
users revisited designated websites. However, • The correlation between website interactivity
Lowry et al (2006) suggested for future research to and website usability sub construct satisfaction.
explore usability in terms of learns ability, • The correlation between website interactivity
efficiency, memorability, user errors. Lowry et al; and sub construct self-efficacy of user.
(2006) used modified measure of satisfaction as
developed by McKinney et al (2002) and found out
that there was strong positive correlation between 4 RESEARCH SIGNIFICANCE
website interactivity and satisfaction. The attribute
of subjective satisfaction used by Lowry et al (2006) This study contributes to the theoretical
refers to how pleasant a user finds it to use computer understanding of how website interactivity relates e-
application or websites. commerce usability and its sub construct such as
satisfaction and self-efficacy. Because user
Self Efficacy influences people’s choice of
satisfaction, learn ability, memorability, efficiency,
activities, how much effort they will expend, and
user error, and self-efficacy have been identified in
how long they will sustain effort in dealing with
defining website usability. Poor website usability
stressful situations. Self efficacy expectations
may lead to the misgivings in e-commerce
toward the emerging traditional computer-based
(Schneiderman, 2000). Rising interactivity shows
learning systems are likely to influence how
potential method of promoting consumer trust in e-
participants use the systems (Bandura, 1996). Users
commerce websites via improved usability. A
with high self-efficacy tend to be more persistent in
deeper understanding of the relationship between
their learning and more confident in their ability to
website interactivity and website usability provides
use the systems. Disorientation is one of the
several theoretical contributions by validating the
problems that novice explorers tend to have while
theoretical models of Khalifa and Liu (2003), Liu
navigating within a cyberspace (Dias et al; 1999).
and Shrum (2002), and extending research done by
Lowry et al; (2006) suggested for future
Lowry et al; (2006).
empirical research the use of self-efficacy to
measure website usability. Previous empirical
research done by Liu and Shrum (2002) also
suggested exploring users’ self-efficacy. We now 5 RESEARCH METHOD
operationalize hypotheses to test various elements of
our theoretical extension of website interactivity to We tested the website interactivity and website
e-commerce website usability as follows: usability of website owned by Indonesian
H1 The perceived measures of two way Association of Textile and Apparel Industries
communication, active control and synchronicity (A.P.I/Asosiasi Pertekstilan Indonesia) composed of
should highly correlate with each other. twenty nine company members. The website can be
H2 There is positive correlation between website found at http://www.textile.web.id, this association
interactivity and website usability. is an open membership association, willing to

466
 WEBSITE INTERACTIVITY - e-Commerce Usability Perspectives in Indonesia

bolster joint e-commerce on one website. So far, To test H2 - there is positive correlation between
API website has been inviting new members to join. website interactiveness and website usability. The
Consumers are able to surf various products offered result was positively correlated between
by the association interactively. This website was interactiveness and usability (0.656). Correlation
chosen for three reasons (1) Textile and Apparel was significant at the 0.01 level (2-tailed) which
Industries is one of the backbone industries fully supported H2.
employing workers and generating fifth biggest To test H3 - there is positive correlation between
income nationally (2) Website participants from website interactiveness and website usability sub
both parties (consumers and producers) require construct satisfaction. The result was positively
interaction with the website to furnish the intended correlated (0.271) but in a less sturdy sense and less
task (3) This website is unique since it is a joint- statistically significant.
membership industries or industries association. The To test H4 - there is positive correlation between
experimental design attributed website usability as website interactiveness and website usability subs
dependent variable and website interactivity as the construct self-efficacy. The result was positively
independent variable. correlated (0.466). Correlation was significant at the
The participants were 31 students from a level 0.01 (2-tailed), which fully support H4. We
Management Information System class at performed Npar test, and the distribution was
Soegijapranata Catholic University in Semarang normal.
Indonesia. Participants were sophomore-level
economics and business majors; 80% were female 6.1 Discussion
and 20% were male. All major protocols were
followed e.g., informed consent. They were not This research extended e-commerce website
asked to mention their name at the questionnaire in interactivity theory by testing the relationships with
order to fulfill objectivity. website usability. We found that perceived
The participants were assigned to surf and interactivity sub constructs of two-way
explore the intended website at the campus communication, active control and synchronicity
computer laboratory for three consecutive days (4 were correlated with each other. This finding
hours times 3 days = 12 hours). They were assigned bolstered the previous research done by Khalifa and
as “ghost shopper” or acting consumers that pretend Liu (2003); Liu and Shrum (2002) and Lowry et al;
to transact with the companies. Participants were (2006). We also provided extension of usability sub-
asked to what they expected and decided from API constructs in terms of learn ability, memorability,
or Indonesian Textile website. Each question was efficiency, user error and self-efficacy that was
rated on 1 to 5 Likert-like scale. recommended for further study by previous
researchers. We captured these ideas to carry out in
our research and eventually, the result was website
6 RESEARCH FINDINGS interactivity, significantly correlated with usability.
However, contrary to the previous research done by
To test H1 - the perceived measures of two-way Lowry et al; (2006), we found that website
communication, active control and synchronicity interactivity was not correlated with usability sub
should be highly correlated with each other-we used construct satisfactions.
Pearson’s correlations. These correlations provided These phenomena can be explained that
support for H1 that was positively correlated to each technically speaking, network performance was
other. However, two-way communication was lousy in the campus area and its surrounding; this
correlated with active control slightly (0.231), caused slower internet speed especially during office
whereas two-way communication correlated with hours. Users encountered with problems for wasting
synchronicity (0.477) significant at the 0.01 level (2- time. Dissatisfaction also occurred when the images
tailed). Active control correlated positively with appeared on textile website was not many and
synchronicity 0.444 significant at the level 0.05 attractive. Apparently, the website designer
levels (2-tailed). The weak correlation between two- purposely designed the minimum images and
way communication and synchronicity can be pictures to be posted in the website to enable faster
described that users faced a major obstacle download and upload time. Fiore and Jin (2003)
particularly the degree of synchronicity with respect found out that fewer images appeared on textile and
to internet speed and the responsiveness of website apparel website e.g., 3 dimensions of virtual images,
officers/masters to deal with the e-commerce users. images of apparel examples dissatisfied the users.

467
ICE-B 2008 - International Conference on e-Business

Slow response from the webmaster or persons in that perceived interactivity sub constructs of two-
charge to serve and entertain the users in fact, has way communication, active control and
dissatisfied them. synchronicity were correlated with each other.
We came up in the study that website Furthermore, result of the research was website
interactivity was strongly correlated with website interactivity, significantly correlated with usability.
usability sub construct self-efficacy. The However, contrary to the previous research done by
interactivity of this textile website was related to the Lowry et al; (2006), we found that website
users confident to surf, obtain information and interactivity was not correlated with usability sub
transact business on the website. construct satisfactions.

7 RECOMMENDATIONS REFERENCES
& CONCLUSIONS
Alba, J; Lynch, Weitz, B; Haniszewski, C. 1997.
Interactive home shopping: consumer, retailer, and
7.1 Limitations & Future Research manufacturer incentives to participate in electronic
marketplaces. Journal of Marketing, vol. 61, pp. 38-53.
Limitation of this research is that although the Ariely, D 2000. Controlling the information flow: Effects
increased interactivity benefits in our sample on consumers’ decision-making and preferences.
website, however, high interactivity is not Journal of consumer research. Vol. 27 pp. 233 – 248.
necessarily always good. Sundar et al (2003) have Bandura A.1996. Multifaceted impact of self-efficacy
suggested that there is point interactivity affects beliefs on academic functioning. Child Develop-
users negatively; this point may be based on the ment, 67, 3. 1206-22.
Chatterjee, P; Hoffman, D.L; Novak, T.P.2003. Modeling
character of the individual. Future research shall the clickstream: Implication web-based advertising
explore the effects of high levels of interactivity on efforts. Marketing Science, 22 (4), 520-541.
website users. Coyle, JR and Thorson, E.2001. The effects of progressive
Another limitation is that our research only levels of interactivity and vividness of in web
focused on textile website, with respondents who marketing sites. Journal of Advertising. Vol. 30 (3), pp
were college or university students. Future research 13-28.
should incorporate travel agents, flight agents, blogs Chen, K and Yen, DC. 2004. Improving the quality of
and other types of websites etc and should also use online presence through interactivity. Information &
research samples from populations other than Management, Vol. 42. pp.217-226.
Dias, P, Gomes, M J and Correia, AP. 1999 Disorientation
university students. in Hypermedia: Environments: mechanisms to support
Since we determined the designated purposely navigation, Journal of Educational Computing
textile website for the respondents or participants in Research, 20, 2 93-117.
the interactivity research, for the future research we Fiore, AM and Jin, HJ. 2003. Influence of image
recommend to open the participants in opting their interactivity on approach responses towards and online
websites. retailer. Internet Research: Electronic Networking
The shortcomings such as internet networks that Applications and Policy. Vol. 13 (1), pp.38-48.
decelerate the internet speed, lack of attractive Ha, L and James E. 1998. Interactivity re-examined: A
images and pictures, unresponsive behavior of baseline analysis of early business websites. Journal
of Broadcasting and Electronic Media, Vol.42 pp.457-74.
officer in charge should be taken into account; Kompas 2004. Industri tekstil Indonesia bisa bangkit lagi
otherwise, user distrust will be prevalence to use e- dengan e-business. (Indonesian Textile Industries can
commerce. be revitalized again by e-business) downloaded at
Limited number of participants or respondents www.kompas.co.id on September 12nd, 2007.
was also the weakness of this research; therefore, Khalifa, M and Liu, V. 2003. The role of expectations and
future research should involve various internet users. desires. International Journal of Electronic
Commerce, vol.7 pp. 31-49
7.2 Conclusions Klein, L. 2003. Creating virtual product experiences: The
role of tele-presence. Journal of Interactive
Marketing. Vol.17 (1), pp. 41-55.
This research confirms that improving website Lohse, G.L., Bellman, S and Johnson, E.J., 2000.
interactivity promotes website usability. It extended Consumer buying behavior on the internet: findings
e-commerce website interactivity theory by testing from panel data. Journal of Interactive Marketing, vol.
the relationships with website usability. We found 14 (1) pp.15-29.

468
 WEBSITE INTERACTIVITY - e-Commerce Usability Perspectives in Indonesia

Li, H., Daugherty, T and Biocca, F. 2001. Characteristics

of virtual experience in electronic commerce research:
a protocol analysis. Journal of Interactive Marketing.
Vol.15 (3). Pp13-30.
Liu, Y. and L.J Shrum. 2002. What is interactivity and is it
always such good thing? Implication s of definition,
person and situation for the influence of interactivity
or advertising influence. Journal of Advertising, Vol.
31 pp. 53-64.
Liu, Y. 2003. Developing a scale to measure the
interactivity of websites. Journal of Advertising vol.
42, pp. 207-216.
Lowry, P.B; Spaulding, T; Wells, T; Moody, G; Moffit, K
and Madariga, S. 2006. A Theoretical model and
empirical result linking website interactivity and
usability satisfaction. Proceedings of the 39th Hawaii
International conference on system sciences. Hawaii.
Macias, W. 2003. A preliminary structural equation model
of comprehension and persuasion of interactive
advertising brand websites. Journal of Interactive
Aadvertising. Vol 3 (2). Available at http://jiad.org.
Mathwick, C. 2002. Understanding the online consumer: a
typology of online relational norms and behavior.
Journal of Interactive Marketing. Vol. 16(1). Pp. 40-55.
McKinney, V., Kanghyun, and Zahedi, FM. 2002.
Measurement of web-customer satisfaction: An
expectation and disconfirmation approach.
Information System Research. Vol. 13 pp. 296-315.
Nielsen, J. 2001. Alertbox: usability metrics. Vol. 2004,
Nielsen Norman. Group.
Pavlik 1996. New media technology: cultural and
commercial perspectives. Boston MA, USA: Allyn
and Bacon.
Stomer, J and Galley. 2000. Online interaction and why
candidates avoid it. Journal of Communication. Vol 50
pp.111-132.
Sundar, SS; Kalyanaraman and Brown, J. 2003.
Communication Research. Vol. 30 pp. 30-59.
Strauss, J and Frost, R. 1999. Marketing in the Internet:
principles of online marketing. Prentice-Hall Upper
Saddle River, NJ.
Shneiderman, B. 2000. Designing trust into online
experience. Communication of ACM. Vol. 43, pp. 57-59.
Wu, G 1999. Perceived interactivity and attribute toward
website. Presented at Annual conference of American
Academy of Advertising. Albuquerque. NM USA.

469
THE HONG KONG GOVERNMENT AUTOMATED PASSENGER
CLEARANCE SYSTEM (E-CHANNEL)
A Study of Channel Management Strategies

Tak Ming Lam

Hong Kong Polytechnic University, Kowloon, Hong Kong
[email protected]

Keywords: e-Government, e-channel, channel management.

Abstract: This paper studies the operation and IT applications in Hong Kong e-channel, including the network,
authentication, and database management, the overall benefits and costs, the evaluation of the IT investment
justification, followed by potential problems of e-channel and possible solutions, finally, a comparison will be
made between Hong Kong e-channel and US e-channel, with further recommendations, followed by a brief
conclusion. E-channel is an Automated Passenger Clearance System which was launched on 16 December
2004, in order to enhance the efficiency of the clearance system.

1 INTRODUCTION is encrypted by user’s fingerprint. When obtaining

the fingerprint after the user inserting the smart ID
Each terminal is connected to the central card, data of the fingerprint will decode the private
management and database server to transfer the in/out key and check whether the pair of private key and the
record and report any abnormal situation. For data public key is valid, so that it will be able to identify
flow, as those in/out records are confidential, a good the authentication. The database of Hong Kong
protection to prevent from leaking information is Immigration Department is one of the most
necessary. Wired connection is the best way as the confidential and important databases in the
information can hardly be unauthorized accessed. government. A large amount of in/out records has to
The Hong Kong Immigration Department also has be processed every day.
the authority to use dedicated line to build WAN
connecting among different locations. That is, using
CAT5 or optical fiber to connect from different 2 BENEFITS
passenger control points such as Lo Wu, Lok Ma
Chau to the central database server, located in Wai First, the implementation of e-channel can save labor
Chai. Apart from that, there is a wireless network cost. The traditional counter needs one inspector for
(WiFi) connecting to palms of inspectors who each counter to monitor people who go through the
monitor the operation of e-channel. As the customs and check their identities. However, in order
information transferred through the wireless network to help those people who have difficulty in passing
is not that confidential, it is justifiable to use it. In fact, through the e-channel, in practice, one inspector is
WiFi has certain kinds of protections for data needed to monitor 4 to 5 e-channels. Therefore, the
transmission. number of inspectors to be employed can be reduced,
In terms of the authentication, the most important part which helps to save much labor cost.
of the e-channel, a double identification is The second benefit is that immigration inspection
implemented, which contains verification of smart ID with e-channel is more reliable than using the
card and biometric verifying using fingerprint. A traditional counter. In the traditional counter, the
Public Key Infrastructure (PKI) is applied in the inspector will just check whether the ID card is valid
e-channel system. Both the private key and public or not, and then check whether the photo on the ID
key are saved in the smart ID card and the private key card is similar to the user’s face. They seldom check

470
 THE HONG KONG GOVERNMENT AUTOMATED PASSENGER CLEARANCE SYSTEM (E-CHANNEL) - A Study
of Channel Management Strategies

in detail whether the user is the owner of the ID card. 4 COMPARISON

The reason is that the validity of your ID card will be
checked before you can enter the e-channel, and it US e-channel is designed for the application of the
will then check your fingerprint to ensure that you are US e-passport, which was newly introduced. The
the owner of the ID card. Besides, in the e-channel United States requires that travelers entering the
system, there is an infrared ray scan. United States under the Visa Waiver Program must
The third main benefit of implementing e-channel is have an e-passport if their passports were issued on or
that it can save the space and time for the customs after October 26, 2006. According to the relevant
service. Some space is needed for placing the information provided on the website of United States
computer, some is for the inspectors to sit and some is Department of State, the US e-passport contains a
left for the corridor. However, the e-channel does not contact-less integrated circuit, which is a 64 Kbit
need any computer or inspectors sitting there, but RFID chip.
only a corridor for people to go through. Compared to the technology used in Hong Kong
Last but not least, the implementation of e-channel smart ID card and the chip reader in e-channel, the
can improve the image of Hong Kong. Since not application of RFID technology in the US e-channel
many countries have adopted similar systems, having helps to speed up immigration inspections. However,
the automatic e-channel implemented may give Hong there are several security concerns about the
Kong a superior representation as a more application of RFID in e-channel, such as skimming
internationalized city, so that the image of Hong the data in e-passport, eavesdropping
Kong can be improved. communications between the chip and reader,
tracking user of the e-passport, and cloning the
passport chip in order to facilitate identity theft
3 PROBLEMS crimes.
Basic Access Control (BAC) is implemented to
The first problem is about the data security. In the minimize the risk of skimming and eavesdropping. A
traditional ID card, only basic information is pair of secret cryptographic keys is stored in the chip
provided. However, in order to match the new embedded in the e-passport. When the reader
e-channel system and other new functions of Hong attempts to read the information in the chip, it
Kong smart ID card, extra information which is more engages in a challenge-response protocol that proves
important and confidential has been added it. Thus, knowledge of the pair of keys and derives a session
private information of citizens may leak out if the key key. Only if authentication is successful can the RFID
is exposed. reader access the data stored in the chip, so that Basic
The second problem is concerning about the Access Control reduces the possibility of
fingerprint identification used in e-channel. unauthorized access to the data.
Sometimes, a person may be blocked in the e-channel. However, the Unique Identifier (UID) can still be
The most possible reason is that some people may communicated with the reader in this process, which
have problems in their fingerprints could theoretically allow the passport user to be
The final problem is the chance of chip damage. tracked. The United States Department of State uses a
Theoretically, the card reader retrieves information Random Unique Identifier (RUID) to prevent the use
by detecting the yellow chip in the ID card. If there of UID for tracking. Each time the chip is accessed,
are any damages in the chip, the card reader may not the e-passport presents a different UID which is not
read the information properly. associated with the UID used in sessions that precede
To cope with this problem, Hong Kong government or follow the current session.
has already provided a cover for the card. Cloning is another security concern that someone
Furthermore, a landing contact of the card reader may copy the information in one chip and store it in
instead of the friction contact is adopted. For card another fake chip. The simplest way to mitigate this
readers of friction contact, the contact part is fixed. action is to verify that the data in the chip match the
data presented in the e-passport, by checking the
photos and biographical data, etc. Additionally,
Public Key Infrastructure (PKI) has been introduced
to automatically confirm that the identity of the

471
ICE-B 2008 - International Conference on e-Business

person presenting the e-passport matches the data Kong government may constantly revise the system
stored in the chip and shown on the passport. and learn from other countries for further
Even though there have been many security improvement. E-channel is only one of the IT
protections in the application of the US e-passport applications of Hong Kong smart ID card. More
and e-channel, the technology is still not mature functions, such as public library service, can be
enough. In August 2006, a security researcher Lukas performed with the smart ID card. The development
Grunwald demonstrated the cloning of a European of e-government has offered citizens and businesses
Union e-passport at the Black Hat and DEFCON quicker and more convenient access to government
security conferences in Las Vegas. (Martin, 2006) information and public services. There is a future
The EU e-passport uses similar RFID technology to trend to achieve a more accessible, accountable and
the US e-passport. However, Randy Vanderhoof, efficient government for Hong Kong as a leading
executive director of the Smart Card Alliance, digital city.
claimed that the data encoded in the chip is digitally
signed and locked by the issuing nation, and could
not be altered even if the chip was cloned. (O'Connor, REFERENCES
2006) Besides, these data are only basic information
presented on the passport data page, and a digital Corcoran, D. et, al. (1999). Smart Cards and Biometrics:
photo. Even if the chip is cloned, there will not be Your Key to PKI. Retrieved April 2, 2008 from
serious problems except that the photo may be used http://www.linuxjournal.com/article/3013
for other purposes. (Reid, 2006) Nevertheless, people E-channel (2003). Retrieved March 25, 2008 from Hong
still worry about the security of e-passport and the Kong Immigration Department http://
RFID technology used in e-channel. www.immd.gov.hk/ehtml/20041216.htm
As mentioned above, US e-channel adopts the RFID E-government in Hong Kong (2008). Retrieved April 10,
2008 from http://www.info.gov.hk/digital21/e-gov/
technology so that data in the chip can be accessed
eng/index.htm
about 10 centimeters away from the RFID reader.
Kanellos.M. (2004). E-passports to put new face on old
Using Hong Kong e-channel, people need to insert documents. Retrieved March 25, 2008 from
smart ID card, and the reader will contact the chip to http://www.zdnetasia.com/news/hardware/0,39042972
read data. This makes the immigration inspection in ,39190596,00.htm
Hong Kong slower than that in US. However, since Martin. K. (2006). U.S. deploys first e-Passport readers.
RFID technology has more security problems, Retrieved March 25,2008 from http://
protections for the US e-passport and e-channel are www.securityfocus.com/brief/315
more than those in Hong Kong. National Database and Registration Authority.
In terms of the biometric identification, US e-channel (2005).Multi-Biometric E-Passport. Retrieved March
uses face recognition while Hong Kong e-channel 25, 2008 from http://www.nadra.gov.pk/site/410/
default.aspx
uses fingerprint. Comparatively, fingerprint is more
NXP. (2006).U.S. State Department Advances NXP
accurate according to security experts (Kanellos,
Technology for ePassport Program. Retrieved March
2004). 25, 2008 from http://www.nxp.com/news/content/
Last but not least, electronic visa may also be file_1257.html
introduced for further convenience. When the visa is O'Connor. C. M. (2006). Industry Group Says E-Passport
stored in the chip in electronic form, it will save time Clone Poses Little Risk. Retrieved March 25, 2008 from
for the Customs officers to check. Costs especially http://www.rfidjournal.com/article/articleview/2559/1/1/
labor cost can be saved. However, it seems not Reid. D. (2006). ePassports 'at risk' from cloning.
appropriate to introduce electronic visa until security Retrieved March 25, 2008 from http://
protection technologies become mature enough. newsvote.bbc.co.uk/mpapps/pagetools/print/news.bbc.
co.uk/2/hi/programmes/click_online/6182207.stm
Stone. M. (2003) E-government Challenges and the Hong
Kong Case Study of Smart Identity Card. Retrieved
5 CONCLUSIONS April 2, 2008 from http://www.info.gov.hk/
digital21/e-gov/eng/press/doc/20030716s.pdf
From the above, it is clear that various IT applications The US Department of State. The U.S. Electronic Passport
in Hong Kong e-channel have brought lots of benefits Frequently Asked Questions. Retrieved March 25,2008
to citizens and the Hong Kong government. However, from http://travel.state.gov/passport/eppt/eppt_2788.html
problems and concerns still exist. Therefore, Hong http://www.immd.gov.hk/zhtml/docs/facts/SB-c2_2007.pdf

472
 THE PRIVATE AND PUBLIC PARTNERSHIP STRATEGY
IN E-GOVERNMENT ESDLIFE IN HONG KONG

Tak Ming Lam

Hong Kong Polytechnic University, Kowloon, Hong Kong
[email protected]

Keywords: Partnership strategy, e-government, ESDlife.

Abstract: Increasingly e-governments are turning to public and private partnerships as an alternative option to fund IT
projects. Also The Hong Kong Government developed ESDlife.com as a one-stop portal to deliver public
and private information and services. The portal’s brand name “ESDLife” best demonstrates the vision to
make “life” simpler and easier for the Hong Kong community. This paper is to discuss whether this business
model would deliver a competitive advantage to customers using the case of “ESDLife. E-government, like
e-commerce, will not be automatically successful by investing resources without clear strategic
understanding of the critical success factors.

1 INTRODUCTION public libraries; or smartly designed public kiosks

installed throughout the city in train and subway
The ESD Scheme was launched in December 2000 stations, shopping centers, supermarkets, cultural
as a flagship e-government project and its portal is and exhibition centers and Government offices.
one of the world’s first bilingual (English and
Chinese) one-stop portals providing integrated
electronic public and commercial services. The main 2 ESDLIFE BUSINESS MODEL
aims of ESD are: to deliver high quality public
services to the community in an innovative manner; The private sector operator is responsible for the
to improve the efficiency and reduce the cost of development, operation and management of the ESD
delivery of public services; and to foster the information infrastructure. The Government has not
development of electronic commerce in Hong Kong. paid for the capital investment nor the recurrent
Providing an open and common information operational and maintenance cost. It will only start
infrastructure for the delivery of public and to pay for each chargeable transaction made through
commercial services, the Scheme drives the the system after the number of cumulative
development and adoption of e-commerce by the chargeable transactions conducted has reached an
business sector and the general community. agreed volume. The operator is allowed to make use
Users can search for services through three of the system to provide advertisements and
mega-channels – “People”, “Business” and “City”; revenue-generating private sector e-commerce
through nine service types “Transport”, services via the same portal web site. This provides
“Citizenship”, “Education”, “Employment”, an additional income stream to the operator.
“Finance”, “Household”, “Leisure”, “Business” and To the public and government, it can embark on such
“Tourist”; through the list of departments and partnerships for the greater good, especially if the
agencies providing the services or through agency has the mandate to promote ICT adoption to
interactive search. Services are available 24 hours a domain industry partners, through leadership in ICT
day and seven days a week. Services can be exploitation, to raise the overall competitiveness of
accessed via personal computers with Internet industry or community specific services. Also the
connection; public computer facilities installed in public agencies are no longer required to front
district offices, community halls, post offices and substantial capital development and yet co-share in

473
ICE-B 2008 - International Conference on e-Business

e-services revenue collected. In addition if the public commercial services, the project also promotes the
agency is not a dominant player to other parties in development of e-commerce in Hong Kong. The
the value chain, then the agency could collaborate e-government services and the commercial services
with other agencies so that the government as a can drive the overall adoption of e-commerce in the
whole becomes a dominant player. business sector and the wider community.
From the private service companies views, we can
see that the motivation lies on the potential 3.4 Quick Receive and Response to
additional profit stream and an opportunity to gain Customer Feedback
market share and domain knowledge of the public
agency’s user base. The private sector is able to get Apart from that, the ESD operator has carried out
the incentives to create more innovative and value customer surveys to obtain customer feedback on the
adding services for the end users. And it has greater usage of ESD services. Such feedback is also
chance to build up goodwill in front of the citizens important to improving the services under the
and the government so as to keep a long-term Scheme.
business relationship with the government.
Probably the most important success factor in
implementing e-government in Hong Kong is the 4 DISADVANTAGE OF THIS
vision of the HKSAR Government. The
MODEL
e-Government Coordination Office was established
to plan and oversee the e-government policies and
initiatives, including ESDLife project.
4.1 Registration Problem
Although the e-government initiative offered
convenience to the public, there remained some
3 ADVANTAGE OF THIS MODEL inconvenience, in particular in relation to those
electronic services that required an electronic
3.1 Reduce Risk of Government certificate to allow the user to register on-line.

Under this partnership approach, the business risk to 4.2 Third Party Software Needed
Government will be kept to the minimum. The
financial charging model also provides an incentive In addition to the electronic certificate, users needed
for the operator to drive the public in accessing to download and install special software in their
e-government services, continuously to upgrade the computer. Two programs were required, one for
system with the most advanced technologies and to security issues and another for the use of Chinese
deliver private sector electronic services characters.

3.2 New Platform for Commercial 4.3 Connection Problem

Opportunity
Although ESD was made available in a variety of
More electronic business opportunities are provided ways – in kiosks at around underground train
by the government, and it could then enhance the stations and other public places – the system only
e-business environment in Hong Kong. The scope of efficiently supported users with broadband Internet
ESDLife requires the exploitation of new hook-up, which meant that users who were
technologies, and it would then strength the connected to the Internet through telephone lines
community for digital exploitation. faced long delays in transmission.

3.3 Integration of Public and 4.4 Complexity of Process

Commercial Services in the Same
Portal For example, renewing a driver’s license through the
Internet required scanning and transmitting an
This enables the public to enjoy electronic services Identity Card and other documents, a process that
in a seamless and convenient manner. By allowing took several minutes and equipments to complete.
the ESD information infrastructure to carry

474
 THE PRIVATE AND PUBLIC PARTNERSHIP STRATEGY IN E-GOVERNMENT ESDLIFE IN HONG KONG

This process was not considered simple, particularly Digital Signature Fee. Since digital signature is one
for non-technical people. of the methods used to provide data integrity, most
of transactions occurring in the ESDLife need to
4.5 Privacy and Security Issues adopt the digital signature technology to verify the
users. However, in most of the situations, citizens
From the citizens’ point of view, the major concern are needed to charge $50 per year for the use of the
for them towards the involvement of the commercial digital signature.
sector in providing e-government services is
infringement of the personal data. It is true that more User Interface. Firstly, ESDLife system should
use of the personal information means more privacy always inform the users what is going on. It is very
concerns. It is possible that in collecting, transferring important in payment status. Secondly, It is
and accessing citizens’ data through the ESDLife important to lead users to have more control and
systems for registration or transaction, the private freedom in using the system.
sector may be inappropriate to use and even
disseminate them without citizens’ previous
consents. And lack of adequate control mechanism REFERENCES
in reviewing private institutes would even bring this
problem to a large extent. On the other hand, privacy Apple Computer (1987). Human Interface Guidelines: The
will also surface as it is possible for the government Apple Desktop Interface, Addison-Wesley.
to track online activities like frequently accessed Poon, S. and Huang X., (2002) Success at E-Governing: A
products, the length of time spent on each page, and Case study of ESDLife in Hong Kong. Electronics
the length of time spent searching. While this Markets 12(4), 270-280
tracking information can be used toward improving ESDLife Launches Electronic Service Delivery,Hutchison
the website and its offerings, the temptation to sell Global Crossing (2000) http://www.hgc.com.hk/eng/
this information to external parties may also exist aboutus_latestnews_11dec00.html
since a close relationship with each other. ESD Services limited www.esdlife.com
Information Technology and Broadcasting Bureau,
Government of the Hong Kong Special Administrative
5 FURTHER IMPROVEMENT Region, August 2001. “Hong Kong Report on Portals”
SUGGESTIONS Jessie Ting, Deputy Secretary of Information Technology
and Broadcasting, Government of the Hong Kong
To develop a consolidated mix government portal, Special Administrative Region, December 10, 1999.
and to keep improving and well prepared for the “Impact of ESD on e-Commerce Development in
challenge in future, we proposed the following Hong Kong”
suggestions: “StockHolm Challenge: Hong Kong – Electronic Service
Delivery (ESD) Scheme (“ESDlife”) http://
Smooth Transformation from Traditional www.stockholmchallenge.se/projectdata.asp?id=1&pr
Processes to Modern Technology. Despite that the ojectid=2892
Scheme has offered a substantially more efficient
and effective alternative means of service delivery, it
is not a easy task to use the Scheme to replace the
existing forms of providing services to the public
over the counter or over the phone.

Privacy Enhancement. To alleviate the concern for

the customers over the public and private sector
collaboration in delivering government services, a
clear and transparent guideline should be drawn to
illustrate the liabilities and obligations between both
parties.

475
 A WEB-BASED ARCHITECTURE FOR
E-GOV APPLICATION DEVELOPMENT

Marcelo Tilli, André M. Panhan, Osman Lima and Leonardo S. Mendes

Communication Networks Laboratory, Communications Department, Eletrical and Computer Engineering Faculty
University of Campinas, UNICAMP- Cidade Universitária "Zeferino Vaz"- Barão Geraldot, Campinas, SP, Brazil
[email protected], [email protected], [email protected], [email protected]

Keywords: Electronic Goverment, e-gov, web-based system, MVC, Hibernate, Struts, Service-oriented.

Abstract: In this paper we present a web-based architecture for e-gov application development. We propose to study
and characterize a software development environment of “electronic governance in cities”, with emphasis
on: legacy system integration, heterogeneous data base integration and development of an integrated web-
based environment for the provision of e-gov services. This architecture is based on a 4-layer MVC model,
the typical MVC (Model, View, Controller) plus a data layer, this architecture imposes few requirements to
its clients, mainly a HTML web-browser, which allows this architecture to be platform-independent.

1 INTRODUCTION information processing platforms, the development

of interoperability platforms and legacy system
A great deal of the development effort of software convergence, are some of the many research areas
systems is centered on modeling, characterization that the study of advanced techniques of software
and development of business models for the solution engineering, computation engineering and
of specific problems. However, when you have to communication engineering can contribute to solve
deal with the development of e-gov systems, several difficulties that faces e-gov system’s
particularly when the systems are built in an software developers.
integrated network environment, there is a great In this paper we present a web-based architecture
dependency among all the systems that compose the for e-gov application development. We propose to
general solution of e-gov applications. study and characterize a software development
Systems of people registration, services, taxes, environment of “electronic governance in cities”,
protocol, finance, are common on a government with emphasis on: legacy system integration,
environment. The problem is that usually, several heterogeneous data base integration and
different vendors, which use different platforms and development of an integrated web-based
different development tools, often build these environment for the provision of e-gov services.
applications. Then we end up with several
heterogeneous and independent data bases, which
are often incompatible. This scenario is very 2 PREVIOUS WORKS
common on e-gov systems around the world.
On the last ten years, several countries have The work of (Peristeras, Tarabanis, 2004). This
sponsored e-gov systems and projects of system paper summarizes a broad research modeling effort,
interoperability. Initially, these architectures were which aims at developing a domain description for
created to access specific problems and not with the the overall governance system. The authors propose
objective of being a generic model that could be the Governance Enterprise Architecture (GEA) as a
universally accepted. set of domain models that serve as a top-level
There are many research areas on the topic of enterprise architecture. Namely, they present the
electronic governance systems (Peristeras, mega-process model, the interaction model, the
Tarabanis, 2004), (Narasimha, Kumar, 2003), public policy formulation object model, the service
(Murthy, Kumar, 2003), (Song Gang, 2005), (Lenk, provision object model and the latest development of
Traunmüller, 2001). The study of semantic the object model for the overall governance system.

476
 A WEB-BASED ARCHITECTURE FOR E-GOV APPLICATION DEVELOPMENT

The work of (Narasimha, Kumar, 2003) propose This 4-layer model allowed us to reach some
an architecture for model development, for the objectives: Facilitates the development of a
planning, design and implementation of e-gov distributed system, with security and with high
systems. The proposed model has four conceptual availability; Enhances the automatization capacity of
layers: Business Process Architecture; Data the city’s public administration sectors, offering
Architecture; Application Architecture; Technology online services to the citizens, expediting its
Infrastructure. execution and accomplishing greater overall quality.
This architecture imposes few requirements to its
clients, mainly a HTML web-browser, which allows
3 SYSTEM ARCHITECTURE this architecture to be platform-independent. With
this architecture, there’s no need to install/update the
We propose an architecture for e-gov systems for application software on the user’s computer.
cities, that can access the general conditions of such Modifications of the system’s software are made
systems, such as: the existence of legacy software, directly on the main server, which is transparent for
redundant, heterogeneous data bases and the the user. This is an improvement over the 2-layer
paradigm shift from client-server platforms to web- model, where all update would have to be physically
based, integrated system for e-Gov services. installed in the user’s computer (Fowler, 2003).
The 4-layer model is as follows:
3.1 Architecture 3.1.1 Model (Business)
For the architecture development, we used a This layer contains all the business logic and
modified MVC model (Model-View-Controller). manages this logic through the use of EJB
In pursuit of greater security and flexibility, we components. Therefore, this layer provides a
separated the client’s communication layer from the structure for the implementation of distributed
business logic; hence we created an application layer applications, aiming for the separation of concepts
responsible for dispatching requests and controlling and to improve on quality factors, such as
its flows. The 4-layer architecture (Model-View- modularity, extensibility and reusability (Alur,
Controller-Data) is show on Figure 1 (Fowler, 2003). Crupi, Malks, 2001).

3.1.2 View (Presentation)

This layer renders the model into a user interface
element. It access enterprise data through the Model
and specifies how that data should be presented. We
are developing this layer with JSP (Java Server
Pages) and JSF (Java Server Faces)

3.1.3 Controller (Application)

This layer processes and responds to events,
typically user actions, and may invoke changes on
the model. Based on the user’s interactions and the
outcome of the model actions, the controller
responds by selecting an appropriate view. To
facilitate the model’s utilization, this layer should be
able to interact with default Data-Base access
technologies, such as JDBC and EJB and also be
able to interact with third-parties technologies such
as Hibernate, iBATIS, or Object Relational Bridge.

3.1.4 Data
This layer is the domain-specific representation of
the information on which the application operates. It
Figure 1: 4-Layer architecture.

477
ICE-B 2008 - International Conference on e-Business

is responsible for the DAO (Data Access Objects), As this technology works with data object’s
or the direct Data-Base access. This model allows concept, the conventional relational data base BDUC
for a transparent persistence of the objects models. was created to facilitate the implementation and take
The data persistence allows us to aggregate advantage of all of the Hibernate resources.
several features to the proposed architecture, like the Therefore, we created a data base under two
following: Transitivity persistence; transparent perspectives: object-oriented and relational, thus, it
retrieval of data-objects (on a given object graph); is possible to represent any data evolved with e-gov
The purpose of the data-layer is to allow for an and its legacy systems.
intelligent way to store object-oriented data, This modeling, as in object oriented system,
allowing for the convergence of heterogeneous data- allows us to reuse data for creation of new tables,
bases. without affecting the system operation.
In particular for this use case, we used the
Microsoft Windows 2003 Server operational system
4 AN E-GOV SYSTEM FOR and the Oracle DBSM, but we can also apply the
same use case to others platforms.
CITIES The SIGM can be deployed in any city, even
those that have legacy application in use; since this
On this section, we will present a use case for a
kind of integration was planned by the adopted
complete system for cities management, which can
architecture and can be done in several ways:
manage all services, citizen records, processes
- SIGM control the legacy applications through
management and relevant data for a city’s
its functionalities and the BDUC data base.
administration. The main purpose of this system is to
- The legacy applications are incorporated by the
provide consistent information to the manager to
SIGM.
make the right decisions. From this prerogative,
- The legacy applications can use the business
emerged the SIGM (Integrated System for
rules already implemented in the SIGM´s EJB
Municipal e-Gov).
layer.
The application integration under a same domain
After deployment, the SIGM can be accessed
was possible with the development of a single data
from any computer in any city location, by any
base, covering, for example, citizens, business,
person (citizen or government employee) registered
social data, departments, processes, internal services,
in the system. These users can request certain
online services, systems users, addresses and others.
services through the Internet, without the need to
Therefore, all data are concentrated and correlated
move to the city hall or any other service point.
under the same concept “Citizen Single Data Base”
This architecture also offers the possibility for
(BDUC).
application distribution on multiple servers, not only
in the system execution, but also on its development.
By distributing the development of a system of this
magnitude, we obtain independence between the
developer’s teams and we can control more
effectively each module implementation. By
separating the business layer from the web layer, it
is possible that others systems can use its
functionalities without affecting the SIGM´s
operation. This can be done by direct access of the
EJB containers which are on the business layer.
Figure 2: SIGM System Modules. To keep in order with the proposed architecture,
on the web layer we adopted the Struts framework,
Through this concept, we achieved the which is an open source framework that works on
interconnection between different modules of an e- MVC architecture and provides several control
gov system in an efficient and easy way, as show in components to create dynamic Java web applications
Figure 2. We have adopted the Hibernate (Husted, Dumoulin, Franciscus, Winterfeldt, 2003).
technology for data base communication, which This architecture also provided the development of
allows the independence of the DBSM (Data base an authentication system, which allows the SIGM to
System Management). manage the permissions and profile control of the
SIGM´s users and users of any other legacy

478
 A WEB-BASED ARCHITECTURE FOR E-GOV APPLICATION DEVELOPMENT

application already in operation, as show on Figure 3. By using this single data base, all the data are
In these systems, all users and applications are converged and correlated under the concept of the
registered through SIGM, and each one has its BDUC, which data access is made through
profile registered in different tables, so we can create Hibernate, which allows the independence from a
different permissions profiles to different users for DBSM. SIGM can also control legacy applications
different applications. To control a legacy through its functionalities and the BDUC data base.
application, we just need to register the profile of
this application and their permissions.
REFERENCES
Peristeras, V., Tarabanis., 2004. Governance Enterprise
Architecture (GEA): Domain Models for E-
Gorvernance. In ICEC’04, Sixth International
Conference on Electronic Commerce, ACM.
Narasimha M., Kumar P. R.V. 2003. Software
architectural design model for e-Governance systems.
In, TENCON 2003. Conference on Convergent
Technologies for Asia-Pacific Region.
Murthy, N., Kumar P. R.V., 2003. Software architectural
design model for e-Governance systems. In TENCON
2003. Conference on Convergent Technologies for
Asia-Pacific Region.
Figure 3: Legacy System Integration. Gang S., 2005 Transcending e-Government: a Case of
Mobile Government in Beijing, The First European
This use case was useful to demonstrate the Conference on Mobile Government, Brighton, July
efficiency of the proposed architecture and the 2005
technologies chosen for its implementation. In Sánchez-Nielsen E., Chávez-Gutiérrez F., 2006. Smart
Diffusion of Government Contents. In ICE-B –
regard to this, we can consider that the use case was
International Conference on E-Business.
successful in its implementation. Lenk K., Traunmüller, R., 2001. Broadening the Concept
The first version of the system was implemented of Electronic Government. In Designing E-
and it is currently in use in the Brazilian cities of São Government, Prins J.E.J. (ed.), Kluwer Law
José do Rio Preto and Campinas. We continue the International, pp. 63-74.i
system development, with new functionalities and Fowler, M., 2003. Patterns of Enterprise Application
features being added frequently, without hindering Architecture. Addison-Wesley, Inc.
the system’s operation. Alur, D., Crupi, J., Malks, D., 2001. Core J2EE Patterns,
Sun Microsystems, Inc.
Husted, T., Dumoulin, C., Franciscus, G., Winterfeldt, D.,
2003. Struts in Action. Manning Publications Co.
5 CONCLUSIONS Bauer, C., King, G., 2005. Hibernate in Action. Manning
Publications Co.
This paper presented an architecture (SIGM) for the
development of e-gov systems, taking in
consideration government environment such as: the
existence of legacy systems, redundant and
heterogeneous data bases, the paradigm shift toward
web-based platforms, the necessity of constant
services development by the government, etc. The
architecture is based on the MVC model with de
addition of a data layer.
The SIGM purpose is to integrate heterogeneous
applications under a distinct domain, to develop a
single data base, that encompass data from all the
citizens, enterprises, social, government
departments, government procedures, internal
services, online services, system’s users, address,
etc.

479
 PROSPECTS OF GRID IN THE CURRENT VOTER
REGISTRATION SCENARIO OF BANGLADESH

Sazia Mahfuz
Department of CSE, East West University, 43 Mohakhali C/A, Dhaka-1212, Bangladesh
[email protected]

Keywords: Grid approach to voter registration, Distributed election process.

Abstract: Bangladesh is a developing country which has gained independence in 1971. In the past, voter registration
was done by going from house to house and then filling up the required forms and counting them in. So any
type of computerized information system was not used. Currently, the government has taken initiatives to
collect voter information with photograph to effectively execute voter registration. This is definitely a
positive initiative from the part of the government. In this paper, to effectively limit the time duration and
complexity of the ongoing process, grid computing infrastructure implementation is proposed. Grid
infrastructure will not only accelerate the voter registration process, but also speed up the election process
on the whole.

1 INTRODUCTION 2 VOTER REGISTRATION IN

THE PAST
The bangladesh election commission (bec) is a
constitutional body responsible primarily for the task After the independence, Parliamentary elections
of conducting elections for the parliament and the were held for the first time under the new
local bodies (http://www.ecs.gov.bd). A significant constitution on 7 March 1973. Later in the year,
part of the task is the preparation, maintenance and elections to local bodies were also held
periodic updating of an electoral roll conforming to (“Banglapedia”,n.d.). According to the article no.
the norms of the structure. 118 of the constituency, there shall be an Election
Recently the BEC faced significant difficulties in Commission for Bangladesh supervising the whole
discharging its responsibilities due largely to the process of elections. Up until now, eight
process adopted in preparing the electoral roll. The parliamentary elections were held. In all of the
9th Parliamentary Election was supposed to be held election processes, the voter registration process was
on 22 January 2007 following the dissolution of somewhat manual in the sense that the listing was
Parliament in October 2006. But, in the days leading done by going from door to door and keeping track
to the elections, violence became widespread with of the voters in a written list. There the enumerators
deterioration in overall law and order situation of the designated by the Election Commission take account
country. Eventually, the elections could not be held. of all the voters under his area. He hands down the
A new caretaker government assumed charge and list to his supervisor, who after further scrutiny
vowed to the nation its intention of holding a free, submits it to the Assistant Registration Officer
fair and credible election. (ARO). Then the list is counted and the proper list is
One of the main reasons for the deferment of the maintained in every polling centre.
January 2007 election was a faulty and grossly This process had several drawbacks:
inaccurate electoral roll that was not acceptable to • A voter can cast multiple votes in different
the opposition parties. To improve the situation, the polling centres under different names as there
newly reformed BEC took steps for preparing a weren’t any biometric identification.
credible electoral roll with photograph. The attempt • Fraudulent voter entry can be created due to the
has been appreciated to all parts of the people of the lack of proper identification.
country. In this paper, an approach to speed up the
current process further has been proposed.

480
 PROSPECTS OF GRID IN THE CURRENT VOTER REGISTRATION SCENARIO OF BANGLADESH

As a result, outcome of the election process had 3.2 “Second Phase: Re-scrutinizing the
become unreliable and fraudulent. General people Forms”
had lost trust in the election process due to the
inefficient techniques adopted for the voter The forms maintained at the registration centres are
registration process. This led to the unrest in scrutinized again and then entries are created in
January, 2007. logbooks after serializing the forms in laptops.

3.3 “Third Phase: Data Collection and

3 VOTER REGISTRATION IN Identity Scrutiny at Registration
THE CURRENT SCENARIO Centres”

An improved system has been achieved by Voters provide their information to the data entry
developing a system whereby the computerized operators after scrutinizers have verified them. The
database is generated and is proposed to be entries are created if required and completed after
maintained through annual update using a process of taking the fingerprints, photos and signatures from
addition, deletion and transfer. This exercise the voters. Then the voters are given receipt.
involves the capture of photographs and the The information thus procured is again handed
integration of these photographs with data in the over to the team leaders using laptops and they are
computers. The whole process can be summarized transferred to the area wise server stations.
through the following phases
(http://www.ecs.gov.bd/English/): 3.4 “Fourth Phase: Registration of
Physically Challenged. Jail
3.1 “First Phase: Data Collection” Inmates, Missed-out Voters and
Ailing People”
Enumerators are provided with the appropriate
forms and required items using which they collect After collecting data from the regular voters, all the
data from the prospective voters. They have to keep other prospective voters who have missed out the
records of the mentally or physically challenged registration have to be taken into account. They have
people as well. The information is then handed over to be counted for following the steps described in the
to the supervisors and after scrutinizing the data, it is earlier phases. The registration for the physically ill
handed over to the Assistant Registration Officer people and the jail inmates have to be completed as
(ARO). The data is compiled according to the areas well, under the required arrangements.
after scrutiny by the ARO.
The voter registration centres and area wise 3.5 “Fifth Phase: Data Processing at
server stations are then set up with assistance from Area Servers”
army. The registration forms are provided to the
team leaders by the AROs. The team leaders then As all data have been collected for the voters, the
inform AROs of the data, time and numbers for information has to be processed in the respective
photograph taking. The information is then again area servers. The duplicate voters have to be
passed on to the enumerators from the AROs identified and the team leaders are informed. Then
through the supervisors. The enumerators inform the the errors are corrected as required. The national ID
prospective voters about the time and date on which cards are processed for distribution after identifying
they have to be available in the voter registration and correcting the errors in the exhibited draft
centre for photograph taking. The necessary national ID cards.
preparation is taken in the voter registration centre,
i.e. the laptops, webcams, fingerprint scanners, 3.6 “Sixth Phase: Draft Voters List
photocopy machines and generators are set up for
Printing and Finalizing”
the operations to be performed for data collection.
Finally, the draft voters’ list is prepared and
exhibited. After hearing the objections on the list,
corrections are made by revising the proper
authority. Then the final voters’ list is prepared and
authenticated by the registration officer.

481
ICE-B 2008 - International Conference on e-Business

The electoral roll with photograph thus prepared wise server and data processing required in those
is proposed to be merged into a central database by server centres.
incorporating the fields necessary for voter Moreover, it will pave ways for many more
registration. This initiative has been laudable opportunities for e-government related services.
compared to the earlier system that had always been
used for voter registration. But will it be successful Voter
or not, that is yet to be seen. A better approach to Registration Center
implementing this scenario will be grid computing at Khulna
infrastructure. Grid
Voter Infrastructure
Voter Registration Area Server Registration Center
Physical Movement of

Center at Khulna at Khulna at Narayanganj

Information

Voter Registration Area Server Voter

Center at at Registration Center
Narayanganj Narayanganj at Barisal

Voter Registration Area Server Figure 2: Proposed scenario for Voter Registration in grid
Center at Barisal at Barisal implemented approach.

Figure 1: Physical movement of information in the fifth

phase of the existing scenario.
5 ANALYSIS
According to the roadmap (http://www.ecs.gov.bd)
4 VOTER REGISTRATION IN provided by the Bangladesh Election commission,
THE PROPOSED SCENARIO the process of enumeration, data entry, printing,
posting and correction of the draft list will start from
The ongoing voter registration can be further 7th August 2007 and finish on 16th August 2008. So
improved with respect to the required time. The this phase of the process will have an approximate
proposed plan is to implement grid infrastructure for time span of one year. This extended time duration
the voter registration process. Currently, the voters’ is due to the lengthy process of data collection and
information has to be transferred to the area servers then transferring the information to the area wise
from the voter registration centres for data server for re-evaluation.
processing as described in the fifth phase. Our This time span will be substantially reduced if
proposed plan suggests a change of the architecture grid infrastructure is implemented. A simple online
used in this phase. interface can be developed to use the distributed
The voter registration centres already have access points. The voter registration centres
laptops, fingerprint scanners, webcams for taking distributed all over the country can be used as the
information from the voters. Every voter registration access points for the grid. The components required
center is also connected to the existing public to maintain the grid can be distributed in any number
switched telephone network (PSTN) provided by of servers located at the different government bodies
BTTB (Bangladesh Telegraph and Telephone secured from outside intervention. The need for
Board) (http://www.bttb.net.bd/). So if there is an manpower in local area wise server maintenance can
online interface for connection to the grid computing be removed through this approach.
system (Foster, 2006), then the need for the transfer
of information from the registration centres to the
area wise servers will be eliminated. All the voter 6 RISK ASSESSMENT
information taken in all the voter registration centres
can be directly fed into the proposed grid
“A gap exists for all e-government projects between
infrastructure.
the design assumptions/requirements and the reality
This approach will obviously limit the time
of the client public agency. The larger this gap
duration required for the data transfer into the area
between design and reality, the greater the risk that
the project will fail (“DRG”,n.d.)”. The technique

482
 PROSPECTS OF GRID IN THE CURRENT VOTER REGISTRATION SCENARIO OF BANGLADESH

followed in design-reality gaps model (“DRG”,n.d.) maintenance compared to the existing system. So a
presents a set of questions which will attempt to rate rating of 2 can be assigned.
the gaps. The total rating for the ITPOSMO dimensions is
The DRG model involves questions relating to a added up to 13. According to the table I, there is a
series of seven 'ITPOSMO' dimensions - high possibility of success in the proposed approach.
information, technology, processes, objectives &
values, staffing & skills, management systems and Table 1: Overall rating analysis.
structures, and other resources - with attached rating Overall
numbers (“DRG”,n.d.). Using each of the seven Likely Outcome
Rating
dimensions, two things have to be observed. One is The e-government project will almost
the organizational reality that exists at the time of 57 – 70 certainly fail unless action is taken to
analysis and the other one is the requirements within close design-reality gaps.
the design of the proposed e-government project. The e-government project may well fail
For each one of the dimensions, a numerical rating 43 – 56 unless action is taken to close design-
has to be allocated to indicate the size of the design- reality gaps.
reality gap on that dimension. The rating for each The e-government might fail totally, or
might well be a partial failure unless
dimension's gap can be anywhere on a scale from 29 – 42
action is taken to close design-reality
zero to ten. gaps.
After the rating, the rating numbers for all seven The e-government project might be a
ITPOSMO dimensions are added up and can be 15 – 28 partial failure unless action is taken to
interpreted according to the table 1 (“DRG”,n.d.). close design-reality gaps.
The seven dimensions are applied to the The e-government project may well
0 – 14
proposed approach and the corresponding rating is succeed.
discussed:
• Information: The information currently being
used for voter registration is the same as required by 7 CONCLUSIONS
the approach. So no change indicates 0 rating.
• Technology: The required software for the Grid computing has become the popular solution to
development of the grid computing environment distributed problems. Election process is a very
would be slightly different than the setup currently critical distributed problem faced in any
being used. So a rating of 4 can be allocated to this government. In this paper, a better approach is
dimension. proposed to maintain the database of the voter
• Processes: The work processes undertaken in registration, so that the distributed election process
the current scenario need not be increased for the can effectively be handled with the least possible use
requirement of the proposed scenario. So a rating of of centralization.
0 can be assigned to this dimension.
• Objectives & values: The objectives and values
required for the key stakeholders of the proposed
system have to be exactly the same as the objectives REFERENCES
and values of the current system. So a rating of 0
should be assigned corresponding to this dimension. Foster, I.(2006). Globus Toolkit Version 4: Software for
Service-Oriented Systems, IFIP International
• Staffing & skills: The staffing members and the
Conference on Network and Parallel Computing,
skill level or type will slightly differ in maintaining Springer-Verlag LNCS 3779, pp 2-13.
the control of the grid architecture. So a rating of 3 http://www.ecs.gov.bd/English/ (15th April 2008).
has to be assigned for these criteria. Retrieved: April 15, 2008.
• Management systems & structures: Banglapedia:history (n.d.). Retrieved: 17th April, 2008,
Management systems and structures required for the http://banglapedia.search.com.bd/HT/H_0136.htm.
implementation of the proposed approach will be http://www.bttb.net.bd/. Retrieved: 17th April, 2008.
slightly different than the exiting system. But the DRG Model
overall management will be the same. So a rating of http://www.egov4dev.org/success/techniques/risk_drg.
shtml . Retrieved: 27th May, 2008.
4 can be assigned for this dimension.
• Other resources: The time and money required
to implement the proposed system will be higher for
the initial setup. But it will definitely be lower for

483
 WEBSITE CREDIBILITY
A Proposal on an Evaluation Method for e-Commerce

Katsuya Watanabe, Masaya Ando and Noboru Sonehara

National Institute of Informatics, 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo, Japan
[email protected], [email protected], [email protected]

Keywords: Information credibility, website credibility, web trust.

Abstract: This paper introduces a research for clarifying the structure of the website credibility. The users with low
information literacy cannot have difficulty making use of e-Commerce services because they cannot judge
the credibility of websites appropriately. Conventional approaches to evaluate the credibility of a website
have been based on aspects like the design of the website or the usage of information security technologies.
However, it is not sufficient for users with low information literacy to distinguish ill-intentioned sites based
on the site design and security technologies alone. In this paper we examine a more comprehensive analysis
and evaluation method which is based not only on evidence internal to the site, but also on third-party
information about the site provider.

1 INTRODUCTION However, evaluating based only on design and

content is not able to identify ill-intentioned sites
Access to and use of the Internet has spread widely posing as trustworthy sites, such as phishing sites.
and swiftly in a decade. Especially, Consumer This research focuses on the verifiability of the
Generated Media (CGM) services such as Weblog information used to evaluate trustworthiness, and
and Social Networking Service (SNS) have been studies ways of evaluating the trustworthiness of a
getting more familiar to general internet users. As website more accurately by using third-party
users send messages actively on the web, the number information about the site information provider.
of cybercrimes such as a phishing has increased.
Information security services and technologies like
WebTrust for CA (Certification Authority) are 2 USER BEHAVIORS IN
provided to take measures against it. However all EVALUATING CREDIBILITY
users don’t necessarily know the meaning of these
information security services and technologies The authors studied specific credibility evaluation
(Dhamija et al., 2006). Therefore "information models used on e-Commerce (EC) and Non-Profit
credibility study" became the main controversial Organization (NPO) websites, based on
themes in the field of Human –Computer questionnaires and the results of experiments using
Interaction. real subjects (Watanabe et al., 2007). We observed
The first study on the credibility of web sites was the behaviour of subjects when evaluating credibility
done by Fogg et al. (Fogg et al., 2001a; 2001b; and explain it in terms of a three-step process.
2002; 2003). Through questionnaires and
experiments, they showed that the factors that 1) Check consistency with the basic message
effected a user's evaluation of the credibility of a site pattern
most strongly were the design and the information First the website is examined and evaluated as to
provided. The authors have also adopted an whether the basic information that one would
approach similar to Fogg et al., deriving a method naturally assume to be provided (the "basic message
which evaluates website credibility from elements pattern") is actually present in an appropriate form.
including design and information provided If there is inconsistency with the pattern the
(Watanabe et al., 2007). credibility evaluation drops considerably.

484
 WEBSITE CREDIBILITY - A Proposal on an Evaluation Method for e-Commerce

2) Evaluate production elements designed logos and icons were able to fool 90% of
In addition to the basic message, elements of the the subjects.
website such as the design, usability and detailed In other words, there are limitations to users'
information are evaluated. If the elements are ability to recognize phishing sites, and those with a
appropriate for the website, the site is deemed more design that simply looks trustworthy may often be
trustworthy, but if there are inadequacies, the successful.
evaluation suffers. The experiments also showed that no matter how
much the content of the site is analyzed, it will not
3) Check information reliability from outside
the website be possible to accurately evaluate the reliability of
The evaluation in 1) and 2) is based on the site.
information found in the site itself. Users also look- The goal of our research is not to identify
for and check third-party information outside of the phishing sites, but to more-accurately evaluate the
site in question to confirm and complement this credibility of websites, and study schemes to support
evaluation. External information tends to be the users' ability to make this judgment. In
consulted especially if the evaluation in 1) and 2) is particular, we expect to be able to support users with
not particularly good, but all users do not always particularly low Internet literacy in this way.
perform this step. Towards this goal, the authors considered the
Of the three steps above, the authors have following three approaches to determining the
derived an evaluation model corresponding to 1) and credibility of a website.
2) earlier, but did not consider the process of 1) Evaluate clues internal to the website
checking external indicators as in 3). The research This is also done by Watanabe et al. (2007). The
due to Fogg et al. also did not consider the factors in first thing the user sees is the website itself, so
step 3). evaluating it is essential to evaluating the credibility
The information on a website consists of the of the site.
site's own statements about itself, so actually, a
verifiably correct evaluation based only on this 2) Evaluation of the information provider
based on third-party information
information is not possible. In real society, we also
Examining the details of information provided on
do not evaluate credibility based on a person's own
third-party websites referencing the site in question
statements about themselves, but generally seek
should be helpful in evaluating the credibility of the
third-party information to verify it.
site. As mentioned earlier, however, the number of
However, in the experiments we have conducted
users checking third-party information is not
so far, there are in fact very few users that actually
particularly high.
use external information to check the credibility of a
As such, it should be helpful, particularly for
website.
users with low information literacy, if the system
can perform this type of evaluation and display the
results to the user on a regular basis.
3 CREDIBILITY EVALUATION
3) Evaluation based on hyperlink structures
FRAMEWORK Phishing sites often use the names of reliable
information providers while carrying on fraudulent
It is difficult to identify ill-intentioned websites, behavior, so it is difficult to correctly evaluate
such as phishing sites, when evaluating the credibility based on name alone.
credibility of the site based only on clues in the site According to a survey by the Anti-Phishing
itself. Working Group (APWG) in the USA, the average
In experiments in which subjects were shown a amount of time a phishing site exists is very short;
website and asked to determine whether it was a about four days (APWG, 2007), so it is not likely
phishing site (Dhamija et al., 2006), 23% of the that there will be any links from other sites to the
subjects only looked at the site content, and did not site. It may be possible to evaluate the credibility of
check other factors like the contents of the address a site by analyzing the structure of hyperlinks to the
bar or whether the SSL-lock icon was displayed. site.
Most of the subjects did not understand the meaning As mentioned in the definitions earlier,
of the SSL warning messages, and they reported credibility is something that the user him/herself
that, indeed, elaborate phishing sites with well must decide. There is a need to support better
decision making about credibility, and evaluating

485
ICE-B 2008 - International Conference on e-Business

websites based on the above three points and having can also be used. These are systems for certifying
the result displayed for the user should help even organizational activities, so they can also be a source
low-information-literacy users make appropriate of information to verify the credibility of the
decisions about the credibility of websites. information provider.
In this paper, we discuss evaluation based on
third-party information about the information 4.2 Preliminary Research
provider in particular detail.
EC sites are not necessarily operated by major
companies, and in fact, many are operated by small
4 WAYS TO EVALUATE and medium-sized businesses. Because of this, there
is some doubt about whether a given company's
CREDIBILITY OF THE name will be referenced on government or municipal
INFORMATION PROVIDER websites, so we performed a survey using a sample
of real companies.
User reviews have become a widely-used approach
to evaluating the credibility of websites using third- 4.2.1 Method
party information. Some EC websites incorporate
comments and ratings from users that have already We first selected an arbitrary 246 Japanese EC
made purchases through the site. Displaying websites, being sure to also include sites that are less
evaluations from existing customers is an indirect well-known.
way of expressing the credibility of the product or We then used a search engine (Google),
site operator itself, but there are still the problems of searching for the company's official name to
whether the provider of the comment or the examine the amount of information available on
comment itself is actually reliable. government sites (go.jp domain), regional
As an example, one could search the web for the municipality sites (le.jp, pref.*.jp, etc.), and
name of the business in order to gather third-party websites of public organizations (or.jp).
information about the operator of an EC website, but Then we looked at up to 100 results more
it is still difficult to determine which of the results closely, and classified them according to type of
are reliable. reference.
So, the authors focused on any public activities Note that domain names in these domains can
of the business or organization acting as the only be obtained by organizations of certain types,
information provider. In other words, we looked at and organizations must provide documentation that
references to the name of the organization in they qualify for the domain name.
information published on the websites of
organizations that are more public in nature. 4.2.2 Results
This investigation is still in progress, so the results
4.1 Validity of the Evaluation Method below represent only a partial survey.
Overall, 226 of the 246 companies (91.9%) were
For this study, we targeted businesses operating e-
referenced on other websites with public domain
Commerce (EC) websites. EC sites will have been
names, which is relatively high (see Table 1).
legally required to register or apply for various
permits and licenses, file reports, and have a history
Table 1: Rate of reference in public domain sites.
of affiliation with public institutions and business
associations. Businesses working with public Domain go.jp or.jp lg.jp pref. jp
institutions are also often required to go through an Rate of
76.4 89.4 46.7 67.5
investigation process. At minimum, insubstantial appearance (%)
companies and organizations are not likely to be Avg. no. of
able to work with public institutions. Much of this 598.3 372.3 18.4 73.9
references
sort of public activity is recorded on the websites of (N=246)
public institutions such as governments,
The content of the references were classified into
municipalities and other administrative
the 15 categories below (Table 2).
organizations.
As can be seen from this classification, even in
Other information like certification levels
references in websites with public domains there are
(ISO9000, ISO14000, Privacy Certification, etc.)
items related to credibility and others that are not. If

486
 WEBSITE CREDIBILITY - A Proposal on an Evaluation Method for e-Commerce

the evaluation can be done in consideration of this regarding whether the user's evaluation is correct or
type of difference in the references it should be not.
possible to further improve the accuracy. From this perspective, integrating third-party
information provides another indicator to support
Table 2: Avg. number of appearances per company. decisions about credibility for low-information-
literacy users.
Type Number Type Number
Organization
Permit 0.57 introduction
20.20

Registration 4.21 Committee 11.32 6 FUTURE WORK

Certification/
Authorization
3.92 Financial Reporting 2.67 In the future, we plan to develop a system which
Commendation/
2.59
Bankruptcy/
0.25
implements the three approaches described here and
Award Litigation
to evaluate the effectiveness of the methods.
Member of a public
agency
4.70 Recalls, etc. 3.81
Bidding/Contracts 5.74 others 6.11
Delivery/
1.09 outside object 22.76 REFERENCES
Provisioning
Participation in public
activities
14.60 (N=246) Fogg, B. J., Marshall, J., Laraki, O., Osipovich, A., Varma,
C., Fang, N., Paul, J., Rangnekar, A., Shon, J., Swani,
4.2.3 Challenges P., Treinen, M. ( 2001) What makes Web sites
credible?: a report on a large quantitative study,
Proceedings of the SIGCHI conference on Human
This method also has limitations. One limitation is
factors in computing systems, pp 61-68.
that the name of the organization may not Fogg, B. J., Kameda, T., Boyd, J., Marshall, J., Sethi, R.,
necessarily be unique in Japan, and it may be Sockol, M., Trowbridge, T. (2002) Stanford-Makovsky
difficult to distinguish between organizations in Web Credibility Study 2002: Investigating what makes
these cases. In this study, we were able to reduce the Web sites credible today, A Research Report by the
amount of this sort of confusion by using the official Stanford Persuasive Technology Lab & Makovsky &
name of the organization, but it will be necessary to Company. Stanford University.
study more-accurate and effective ways to resolve Watanabe, K., Hara, Y., Hasegawa, A. and Sonehara， N.
this difficulty. (2007) Evaluation model of Web page credibility,
Proceedings of the First Workshop on Information
Credibility on the Web (WICOW), pp49-56.
Dhamija, R., Tygar, J. D. and Hearst, M. (2006) Why
5 DISCUSSION phishing works, Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems,
In this paper we have proposed a method for ACM Press, New York, NY, USA, pp581-590.
evaluating the credibility of websites that uses third- Anti-Phishing Working Group (2007) Phishing Activity
party information to verify the credibility of the Trends: Report for the Month of February, 2007,
site's information provider in addition to the site Available at www.antiphishing.org
design and the information provided on the website
itself.
As discussed in the definitions section,
credibility is something that users must decide for
themselves, but as shown in this paper, we believe
that gathering appropriate third-party information
can help users make this sort of determination more
accurately.
More specifically, the user's evaluation of the
credibility of a website can be verified using third-
party information. If the evaluation is correct, the
user's confidence can be raised by the amount and
quality of the information from third-party sources.
In other words, if there is very little information
available from third-party sites, there is more risk

487
 AUTHOR INDEX

Abrahamsson, P....................................... 169 Davcev, D. ...............................................203

Abreu, P..................................................... 51 Decker, M. .......................................185, 255
Adebanjo, D. ............................................. 93 Domínguez, A............................................37
Aguilera, A. ............................................... 58 Duarte, M...................................................37
Akiyoshi, M............................................. 447 Eckert, K..................................................307
Alahuhta, P. ............................................. 169 Eriksson, N. .............................................115
Al-Bayari, O. ........................................... 217 Ervasti, M. ...............................................131
Alboaie, L.................................................. 64 Evens, T. ..................................................124
Ali, H....................................................... 209 Fernandes, V............................................149
Ali, S........................................................ 209 Ferreira, J. ................................................323
Al-Omar, K.............................................. 191 Florentina, C. .............................................77
Alotaibi, M. ............................................. 143 Freitas, C..................................................378
Ando, M. ................................................. 484 Gouveia, J. ...............................................161
Andonoff, E............................................. 267 Guido, A. .................................................101
Arba, R. ................................................... 437 Guo, J.........................................................81
Arsic, M................................................... 203 Hafeez, K. ................................................362
Aryanto, V............................................... 464 Haidegger, W...........................................107
Asensio, J. ............................................... 259 Häikiö, J...................................................247
Ayala, R................................................... 259 Hall, J...............................................307, 390
Baek, E. ................................................... 241 Hattori, T. ..................................................29
Baptista, F................................................ 225 Helaakoski, H. .........................................131
Bara, A. ................................................... 460 Hiramatsu, A............................................229
Barry, C. .................................................. 339 Höller, M. ................................................235
Bernardos, A............................................ 177 Holzinger, A. ...................................235, 454
Biffl, S. .................................................... 454 Hu, Z..........................................................81
Bloice, M................................................. 235 Hung, H. ..................................................279
Bologa, A. ....................................... 425, 460 Ilievski, D. ...............................................203
Bologa, R......................................... 425, 460 Ioana, B......................................................77
Bouaziz, R. .............................................. 267 Iribarne, L. ...............................................259
Bouzguenda, L. ....................................... 267 Ismail, H. ...................................................93
Braghetto, K. ........................................... 323 Iulian, P......................................................77
Brambring, T. .......................................... 155 Iwashita, M. ...............................................12
Bruno, G. ................................................. 315 Katai, O......................................................29
Buhl, M.................................................... 285 Kaufman, V. ............................................397
Bulander, R.............................................. 413 Kawakami, H. ............................................29
Bulo, I........................................................ 37 Keoy, K....................................................362
Butler, S..................................................... 93 Kittl, B. ....................................................285
Casar, J. ................................................... 177 Kobayashi, Y. ..........................................447
Chaâbane, M............................................ 267 Kodama, T. ................................................21
Charleton, D. ........................................... 339 Komoda, N...............................................349
Chen, D.................................................... 433 Koukoulas, S............................................390
Cho, V. ............................................ 279, 421 Kunii, T......................................................21
Cornacchia, M. ........................................ 315 Kuwana, T. ..............................................447
Cruz, P..................................................... 300 Lam, T. ............................................470, 473
Cunha, A.................................................. 149 Laukkanen, T. ..........................................300

489
ICE-B 2008 - International Conference on e-Business

AUTHOR INDEX (CONT.)

Lee, J. ...................................................... 241 Santin, A. .................................................378

Lewandowski, J....................................... 285 Santos, D....................................................89
Lieu, G..................................................... 421 Sapio, B. ..................................................315
Lima, O.................................................... 476 Saraiva, J..................................................225
Lin, C............................................... 384, 443 Schiefer, G. ..............................................255
Liu, W...................................................... 433 Schuurman, D. .........................................124
Livi, S. ..................................................... 315 Seki, Y. ......................................................21
Lukovic, S. .............................................. 197 Shimogawa, S. ...........................................12
Määttä, J. ................................................. 247 Shiose, T. ...................................................29
Mahfuz, S. ............................................... 480 Siddiqui, T. ..................................................5
Marcos, D. ............................................... 177 Silva, A. ...................................................225
Marez, L. ......................................... 124, 392 Sonehara, N. ............................................484
Marin, P..................................................... 77 Soto-Acosta, P. ........................................405
Martins, M............................................... 370 Stanisavljevic, M. ....................................197
Mendes, L.......................................... 89, 476 Stoica, F. ....................................................45
Mendes, P. ................................................. 51 Strandvik, P. ............................................115
Meroño-Cerdan, A. ................................. 405 Strangio, M. .............................................334
Moritsu, T................................................ 349 Su, Y. .......................................................443
Muntean, M. ............................................ 425 Subero, A. ..................................................58
Muresan, L. ............................................. 437 Sung, S.....................................................241
Natchetoi, Y. ........................................... 397 Susaki, S. .................................................429
Nicolò, E.................................................. 315 Takahashi, Y............................................429
Nishimatsu, K............................................ 12 Taniguchi, Y. ...........................................447
Nose, K.................................................... 229 Thissen, D................................................155
Nummiaho, A. ......................................... 169 Tilli, M.....................................................476
Oliveira, J. ............................................... 378 Torabi, T. .................................................209
Oliveira, P................................................ 161 Trauner, J. ................................................454
Oliveira, T. .............................................. 370 Urlesberger, B..........................................235
Padilla, N................................................. 259 Usugami, J. ..............................................357
Pah, I.......................................................... 45 Varajão, J. ........................................149, 161
Paiano, R. ................................................ 101 Verdegem, P. ...........................................292
Panhan, A. ......................................... 89, 476 Vinhas, V. ..................................................51
Papa, F..................................................... 315 Wallin, A. ................................................247
Pattinson, C. .......................................... 5, 73 Watanabe, K. ...........................................484
Podean, M................................................ 437 Wong, Y...................................................279
Ponomarev, V.......................................... 397 Yamada, T. ..............................................429
Popa, E. ..................................................... 45 Yamasaki, T.............................................229
Pu, C. ....................................................... 323 Zairi, M....................................................362
Pulimeno, E. ............................................ 101 Zhu, D..............................................384, 443
Puzovic, N. .............................................. 197
Rigas, D........................................... 143, 191
Robinson, L. .............................................. 73
Sabau, G. ................................................. 425
Sadoun, B. ............................................... 217

490