Information retrieval and the user's profile: critical aspects

Proceedings of the 8th WSEAS Int. Conf. on ARTIFICIAL INTELLIGENCE, KNOWLEDGE ENGINEERING & DATA BASES (AIKED '09) Information retrieval and the user's profile. Critical aspects. MARIA RODICA VOLOVICI; DANIEL VOLOVICI University Library Lucian Blaga University of Sibiu Victoriei Blvd. No.10 ROMÂNIA [email protected] http://bcu.ulbsibiu.ro Abstract: - The paper presents the quality services improvement for every type of user in the university library and their utility. The users’ profile contains information about the users, which is connected with its informational necessities. The profiles were used as models for possible types of questions (queries) since the appearance of the systems for the redeeming of information. Many efforts have been done recently for the integration of the profiles used and questions in more satisfying systems. Many systems of current awareness are based on the easy profiles. This type of profile is easier to compare with the new documents than any other complex profile; in the same time it clears away the developing of a sophisticated profile. Key Words: - user, system, quality, information, knowledge connected to the user, seen as a person and less to its informational necessities. This type of profile is concerned with the way in which the person uses the information and on the stocking of those data about the user, which can be used in order to see which documents have to be included in the recovered package. These data can be found out about by a reporter when interviewing his customers. In conclusion, a complex profile can contain information about: • The level of education. A high school pupil and a Doctor of Science will probably want different packages of documents which could be the answer of a similar question. • The acquaintance with a demanded field. A researcher, familiar to a subject and who wants to be permanently informed about it, will search for a different reference material, while a person who can hardly approach the field won’t. • Linguistic abilities. Documents written in a foreign language have a limited utility if he, who looks for them, can’t read in that language. • Getting subscriptions to newspapers. The newspapers the user is a subscriber to, are easily accessible and thus, they are great sources of reference. • Customs connected to informing by reading. If the user reads a newspaper regularly, he may have information about the most important articles in that newspaper. Such references should be mentioned at the end of the list, and more, only to point out the fact that the user has already seen them. If the user doesn’t read the newspapers he is subscribed to, they become major sources of references. 1 Easy profiles An easy profile is similar to a question. It contains a set of key terms which, usually, have different weights. Such profiles were first used in systems for currently awareness (CA) for selective extent of information (SDI in English).These systems are usually used when there is a relatively fixed group of users who have an interest that lasts in time. For example, there are many organizations which use these systems in order to help the researchers to find out about the innovations in their fields of researching. An investor or stockbroker could use such a system to supply the stock exchange with the newest information about the companies on the market. When using these systems, the user is usually asked to fill in a formulary (a profile) of necessities or interests, which he has to periodically renew, at about six months or even more often. In the meantime, the profile works as a standard question which is rolled to the data base as often as it has to be. The term routing query (repeated question) has been recently introduced in order to describe a persistent question, such as a profile, which is different from an ad hoc query (an ad hoc question). This is usually asked just once (Harman [Har93]). Due to its nature, the easy profile is relatively easy to roll on in the data base. Nevertheless, it has a limited capacity to show which the documents that can be used by the user, are, because it restores only words and key- phrases which appear in a text. 2 Complex profiles A complex profile contains information which is difficult to correlate with the documents but, however, it can influence the recovering of data. The information included in such a profile is ISSN: 1790-5109 46 ISBN: 978-960-474-051-2 Proceedings of the 8th WSEAS Int. Conf. on ARTIFICIAL INTELLIGENCE, KNOWLEDGE ENGINEERING & DATA BASES (AIKED '09) • Special references. The user might want to access some writers’ works or to have certain preferences when talking about newspapers. At the same time, there is a category of materials which the user doesn’t want to see, such as the works written by him. Such information cannot be compared to the ones found in books or in the printed newspapers. found in the medicine field or some other situations in which time is an important factor, because the literature of specialty contains new information, applicable in current contexts any time. They even suggested the introduction of the term petrieval = prerecovering for a system which could alert the doctors very quickly about the most important documents for the cases they would work (Wagner[Wag95]). That is why many of them can’t be used in the recovering process, but they must be applied to the recovered package in order to organize it in such a way, so that it will correspond to the unused documents. As the user is usually a passive participator regarding the system of current awareness, its efficacy is pretty difficult to evaluate. If the user gets the results on line (for example on the internet), then one can make – with his approval – a monitoring of the way in which these results are used. However, this action has some limits, because the user can copy his documents and only after he an use them, without necessarily being connected to the network. The value of this monitoring becomes visible when the person wants to check the documents while he or she is connected to the network and then, he or she makes a selection regarding the documents which will be saved and those which won’t. if the results won’t be presented on-line, the users need to cooperate to make the system as efficient as possible. The only action which has good results remains, however, the demand the user makes to get certain document. 3 The systems of current awareness A system of current awareness is made to keep the users informed about the fields of interest. So, a major premise, from its beginning is that the user is aware of the existence of the works from the past and that he only needs to know about the modern developments. The system is made in such a way so that it could take a responsibility, letting thus the user use his time to search or elaborate new works. In this context, the system usually works automatically, without a human intervention. There are two conditions that are needed to build a system of current awareness: 1) One has to look only for the current literature. 2) The search must be done automatically and the user doesn’t need to use questions to activate the system. Thus, the user is asked to make a profile of his interests which is stocked. The new documents which arrive in the library or other systems of information are registered in a short time, about a week or two. Then all the users’ profiles are turned to routing queries for this documentation and the results are sent to the right users. The users have more options, starting from the possibilities of asking completed copies of some documents just through filling in the notifications they get and ending with ignoring the results. The users have the possibility to remake their profile and there are many systems which often remind them to readapt this profile. Fig.1 Many systems of current awareness are based on the easy profiles. This type of profile is easier to compare with the new documents than any other complex profile; in the same time it clears away the developing of a sophisticated profile. The system of current awareness is efficient when the documentation is continually changing. This happens in the laboratories of research and in the organizations of business, finances and government and politic exchanges. A similar situation can be ISSN: 1790-5109 47 ISBN: 978-960-474-051-2 Proceedings of the 8th WSEAS Int. Conf. on ARTIFICIAL INTELLIGENCE, KNOWLEDGE ENGINEERING & DATA BASES (AIKED '09) In the framework of the Library of “Lucian Blaga” of Sibiu the users are invited to fill in a questionnaire which permit to the collective of the library to improve theirs usually activities. In the schedule (annexes) there are represented the results of the questionnaire perform. Do you prefer and use the on-line database ? 70 70 60 50 Fig.2 42.5 40 4 Systems for retrospective search The systems for retrospective search are destined to look for the whole existing documentation as an answer to an ad hoc question. A typical system for retrospective search contains a relatively extended and stable stock of documents and also a set of questions that are always changing, questions that are renewed whenever the user accesses the system. In opposition to it, the system of current awareness has an extended and stable stock of profiles of the users and a set of documents which change themselves continually starting with the stocking of the new documents such as the numbers of newspaper or the technique reports. While a system of retrospective search is said to examine the documents of data base to look for the answer of a question, a system of current awareness is said to examine a data base which contains the users’ profiles to find out which of them suits to the existing documents. 30 27.5 30 PREFERA UTILIZEAZA NU UTILIZEAZA 20 15 15 10 0 PREFERA UTILIZEAZA NU UTILIZEAZA DA 70 27.5 42.5 NU 30 15 15 Do yoy know the structure and the services of the library? Are they satisfactory ? 7000.00% 6250.00% 6000.00% 5000.00% 37.5 4000.00% 3750.00% CUNOSC 3000.00% SUFICIENTE 25 INSUFICIENTE 2000.00% 1000.00% 0.00% Normally, all the systems for retrospective search when are used, without those in the phase of experimenting, treat the users in the same way. This means that, for a system, all people are alike and no effort is done to differentiate the users concerning their past or their capacity to use the information they got. Although it is easier to elaborate and handle a system in this way, the result is that such systems are limited concerning their power to satisfy the people’s informational necessities. At the same time the problem gets deeper if the group of users is bigger and more heterogeneous. However, an informational system which answers to a homogeneous group of users can work on the base of a ‘standard user’ with very good results. But, generally, there is no standard user to be used in the systems, which should treat the people by their level of education, experience and special interests. ISSN: 1790-5109 CUNOSC SUFICIENTE INSUFICIENTE DA NU 6250.00% 3750.00% 37.5 25 References : [1] Harman, Dorma K: - The First Text Retrieval Conference (Trec 1). Washington, D. C., Nist Special Publication 500-2007, 1993; [2] Volovici, R. M. – Cercetări asupra creşterii fiabilităţii sistemului informatic şi al informaţiei. Referat de doctorat, Sibiu 2003; [3] Wagner, Michael M. – Decision-theoretic clinical information systems. Ph. D. Kiss, University of Pittsbourg, 1995. 48 ISBN: 978-960-474-051-2