Pattern Discovery

To uncover qualitative and quantitative patterns in a data set is a challenging task for research in the area of machine learning and data analysis. Due to the complexity of real-world data, high-order (polythetic) patterns or event associations, in addition to first-order class-dependent relationships, have to be acquired. Once the patterns of different orders are found, they should be represented in a form appropriate for further analysis and interpretation. In this paper, we propose a novel method to discover qualitative and quantitative patterns (or event associations) inherent in a data set. It uses the adjusted residual analysis in statistics to test the significance of the occurrence of a pattern candidate against its expectation. To avoid exhaustive search of all possible combinations of primary events, techniques of eliminating the impossible pattern candidates are developed. The detected patterns of different orders are then represented in an attributed hypergraph which is lucid for pattern interpretation and analysis. Test results on artificial and real-world data are discussed toward the end of the paper.

The increased usage of World Wide Web (WWW) becomes a vast data repository related to the users’ interaction with the websites which is unstructured, unlabeled, high redundant and less reliable recorded in weblog. In addition, the existence of non linearity, incompleteness, heterogeneous and transient nature of data makes the weblog further complex. This situation creates inevitably increasing challenges in extracting desired knowledge from web log. Web usage mining is a methodology that blends traditional mining techniques with sophisticated algorithms to capture, model and analyze the behavioral patterns from weblog. The knowledge derived from such patterns creates a great value addition to any organization as they make use in decision making process. Thus, it is necessary to empower the web usage mining techniques that are aptly compatible to incremental nature of weblog. These techniques promote the prerequisite of applying the new approach at all stages of web usage mining comprehensively, to completely investigate the web usage mining effectively. To design and develop comprehensive model for investigating the web usage mining, the authors in the present paper conducts an extensive literature survey on various research activities in the era of web usage mining.

We describe a model for strings of characters that is loosely based on the Lempel Ziv model with the addition that a repeated substring can be an approximate match to the original substring; this is close to the situation of DNA, for example. Typically there are many explanations for a given string under the model, some optimal and many suboptimal. Rather than commit to one optimal explanation, we sum the probabilities over all explanations under the model because this gives the probability of the data under the model. The model has a small number of parameters and these can be estimated from the given string by an expectation-maximization (EM) algorithm. Each iteration of the EM algorithm takes O(n2) time and a few iterations are typically sufficient. O(n2) complexity is impractical for strings of more than a few tens of thousands of characters and a faster approximation algorithm is also given. The model is further extended to include approximate reverse complementary repeats when...

Recommender systems are helpful tools which provide an adaptive Web environment for Web users. Recently, a number of Web page recommender systems have been developed to extract the user behavior from the user's navigational path and predict the next request as he/she visits Web pages. Web Usage Mining (WUM) is a kind of data mining method that can be used to discover this behavior of user and his/her access patterns from Web log data. This paper first presents an overview of the used concepts and techniques of WUM to design Web recommender systems. Then it is shown that how WUM can be applied to Web server logs for discovering access patterns. Afterward, we analyze some o f the problems and challenges in deploying recommender systems. Finally, we propose the solutions which address these problems.

The detection of frequently occurring patterns, also called motifs, in data streams has been recognized as an important task. To find these motifs, we use an advanced event encoding and pattern discovery algorithm. As a large time series can contain hundreds of motifs, there is a need to support interactive analysis and exploration. In addition, for certain applications, such as

This paper uses disaggregated export data to explore the relationship between economic discovery and economic development. We find that discoveries, or episodes when countries begin exporting a new product, are not limited to so-called 'dynamic' industries, rather they also occur in traditional sectors such as agriculture. In addition, the data suggest discovery is a component of the stages of productive diversification that occur with development, following a consistent pattern: discovery activity peaks at the lower-middle income level and then declines. Based on this pattern, we show that discovery in the 1990s occurred with a higher than expected frequency in Eastern Europe and Central Asia, and lower than expected frequency in Sub-Saharan Africa. Discovery is not found to be a product of structural transformation based on changing factor endowments across income levels. Beyond export growth, population, and development, there are no significant and positive relationships between the expected drivers of entrepreneurship and the frequency of discovery. Combined with the finding that higher absorptive capacity and lower barriers to entry are associated with a reduction in discovery, this suggests that market failures arising from imitation and free-riding may be inhibiting the emergence of new export products in developing countries.

This study presents an unsupervised feature selection approach for the discovery of significant patterns in seismic wavefields. We iteratively reduce the number of features generated from seismic time series by first considering significance of individual features. Significance testing is done by assessing the randomness of the time series with the Wald-Wolfowitz runs test and by comparing observed and theoretical variability of features. In a second step the in-between feature dependencies are assessed based on correlation hunting in feature subsets using Self-Organizing Maps (SOMs). We show the improved discriminative power of our procedure compared to manually selected feature subsets by cross-validation applied to synthetic seismic wavefield data. Furthermore, we apply the method to real-world data with the aim to define suitable features for earthquake detection and seismic phase classification in seismic recordings.

With the rapid development of information technology, the World Wide Web has been widely used in various applications, such as search engines, online learning and electronic commerce. These applications are used by a diverse population of users with heterogeneous backgrounds, in terms of their knowledge, skills, and needs. Therefore, human factors are key issues for the development of web-based application and research. This paper first identifies reviews from different authorsand then examines the three important human factors: gender differences, prior knowledge, and cognitive styles. The review result is not significantly correct; a new model is proposed that will access the data (log data) and show the human access behavior. The proposed model has two stages: web intelligence and navigation pattern. Stage 1(web intelligence system) captures data from different server and converts in the form of table (data store). Stage 2 uses the N-gram algorithm which assumes that the last N-pages browsed affect the probability of the next page to be visited, and user navigation sessions are modelled as a hypertext probabilistic grammar whose higher probability strings correspond to the user's preferred trails.In this paper web caching and prefetching are two important approaches used to reduce the noticeable response time perceived by users.The model improves the navigation pattern of users and find the users behavior ( gender difference and user type) that finding is used by site designer and researchers and also used for detecting and avoiding the terror threats caused by terrorists all over the world.The paper is organized into five different parts, first part contain introduction, second part contain different type of web mining third part contain usage mining on the web and forth part contain analysis of human factor and evaluation technique,fifth part contain propose methodology and last part contains application, limitation, conclusion and further work.

Knowing patterns of relationship in a social network is very useful for law enforcement agencies to investigate collaborations among criminals, for businesses to exploit relationships to sell products, or for individuals who wish to network with others. After all, it is not just what you know, but also whom you know, that matters. However, finding out who is related to whom on a large scale is a complex problem. Asking every single individual would be impractical, given the huge number of individuals and the changing dynamics of relationships. Recent advancement in technology has allowed more data about activities of individuals to be collected. Such data may be mined to reveal associations between these individuals. Specifically, we focus on data having space and time elements, such as logs of people's movement over various locations or of their Internet activities at various cyber locations. Reasoning that individuals who are frequently found together are likely to be associated with each other, we mine from the data instances where several actors co-occur in space and time, presumably due to an underlying interaction. We call these spatio-temporal co-occurrences events, which we use to establish relationships between pairs of individuals. In this paper, we propose a model for constructing a social network from events, and provide an algorithm that mines these events from the data. Experiments on a real-life data tracking people's accesses to cyber locations have also yielded encouraging results.

Motivation: Transcription factor binding sites often differ significantly in their primary sequence and can hardly be aligned. Often one set of sites can contain several subsets of sequences that follow not just one but several different patterns. There is a need for sensitive methods to reveal multiple patterns in unaligned sets of sequences. Results: We developed a novel method for analysis of unaligned sets of sequences based on kernel estimation. The method is able to reveal 'multiple local patterns'-a set of weight matrices. Every weight matrix characterizes a pattern that can be found in a significant subset of sequences under analysis. The method developed has been compared with several other methods of pattern discovery such as Gibbs sampling, MEME, CONSENSUS, MULTIPROFILER and PROJECTION. The kernel method showed the best performance in terms of how close the revealed weight matrices are to the original ones. We applied the kernel method to analyze three samples of promoters (cell-cycle, T-cells and muscle-specific). We compared the multiple patterns revealed with the TRANSFAC library of weight matrices and found a strong similarity to several weight matrices for transcription factors known to be involved in the mentioned specific gene regulation.

A very large percentage of business and academic data is stored in textual format. With the exception of metadata, such as author, date, title and publisher, this data is not overtly structured like the standard, mainly numerical, data in relational databases. Parallel to data mining, which finds new patterns and trends in numerical data, text mining is the process aimed at discovering unknown patterns in free text. Owing to the importance of competitive and scientific knowledge that can be exploited from these texts, “text mining has become an increasingly popular and essential theme in data mining” (Han & Kamber, 2001, p. 428). Text mining is an evolving field and its relatively short history goes hand in hand with the recent explosion in availability of electronic textual information. Chen (2001, p. vi) remarks that “text mining is an emerging technical area that is relatively unknown to IT professions”. This explains the fact that despite the value of text mining, most research and development efforts still focus on data mining using structured data (Fan et al., 2006). In the next section, the background and need for text mining will be discussed after which the various uses and techniques of text mining are described. The importance of visualisation and some critical issues will then be discussed followed by some suggestions for future research topics.

Application of data mining techniques to the World Wide Web, referred to as Web mining, has been the focus of several recent research projects and papers. However, there is no established vocabulary, leading to confusion when comparing research e orts. The term Web mining has been used in two distinct ways. The rst, called Web content mining in this paper, is the process of information discovery from sources across the World Wide Web. The second, called Web usage mining, is the process of mining for user browsing and access patterns. In this paper we de ne Web mining and present an overview of the various research issues, techniques, and development e orts. We brie y describe WEBMINER, a system for Web usage mining, and conclude this paper by listing research issues.

Decision support nowadays is more and more targeted to large scale complicated systems and domains. The success of a decision support system relies mainly on its capability of processing large amounts of data and efficiently extracting useful knowledge from the data, especially knowledge which is previously unknown to the decision makers. With a large scale system, traditional knowledge acquisition models become inefficient and/or more biased, due to the subjectivity of the experts or the pre-assumptions of certain ideas or algorithmic procedures. Today, with the rapid development of computer technologies, the capability of collecting data has been greatly advanced. Data becomes the most valuable resource for an organization. This paper presents a fundamental framework toward intelligent decision support by analyzing a large amount of mixed-mode data (data with a mixture of continuous and categorical values) in order to bridge the subjectivity and objectivity of a decision support process. By considering significant associations of artifacts (events) inherent in the data as patterns, we define patterns as statistically significant associations among feature values represented by joint events or hypercells in the feature space. We then present an algorithm which automatically discovers statistically significant hypercells (patterns) based on: 1) a residual analysis, which tests the significance of the deviation when the occurrence of a hypercell differs from its expectation, and 2) an optimization formulation to enable recursive discovery. By discovering patterns from data sets based on such an objective measure, the nature of the problem domain will be revealed. The patterns can then be applied to solve specific problems as being interpreted or inferred with. Classifiers can be developed, probabilistic density functions can be estimated, association rules can be generated and pattern based data query can be formulated. It presents a data driven system and a process from data to patterns and from patterns to applicable rules/models for decision support.

This paper provides a comprehensive idea about the pattern discovery of Web usage mining. Web site designers should have clear understanding of user's profile and site objectives, as well as an emphasized knowledge of the way users will browse Web pages. The designers can study the visitor's activities through the Web analysis and find patterns of the visitor's behavior. This Web analysis involves the transformation and interpretation of the Web log records to find out the hidden information or predictive pattern by the data mining and knowledge discovery technique. This result offer a great prospect coupled with the Web warehousing.

Process discovery is the automated construction of structured process models from information system event logs. Such event logs often contain positive examples only. Without negative examples, it is a challenge to strike the right balance between recall and specificity, and to deal with problems such as expressiveness, noise, incomplete event logs, or the inclusion of prior knowledge. In this paper, we present a configurable technique that deals with these challenges by representing process discovery as a multi-relational classification problem on event logs supplemented with Artificially Generated Negative Events (AGNEs). This problem formulation allows using learning algorithms and evaluation techniques that are well-know in the machine learning community. Moreover, it allows users to have a declarative control over the inductive bias and language bias.

Discovery of interesting or frequently appearing time series patterns is one of the important tasks in various time series data mining applications. However, recent research criticized that discovering subsequence patterns in time series using clustering approaches is meaningless. It is due to the presence of trivial matched subsequences in the formation of the time series subsequences using sliding window method. The objective of this paper is to propose a threshold-free approach to improve the method for segmenting long stock time series into subsequences using sliding window. The proposed approach filters the trivial matched subsequences by changing Perceptually Important Point (PIP) detection and reduced the dimension by PIP identification.

Discovering patterns in graphs has long been an area of interest. In most contemporary approaches to such pattern discovery either quantitative anomalies or frequency of substructure is used to measure the interestingness of a pattern. In this paper we address the issue of discovering informative subgraphs within RDF graphs. We motivate our work with an example related to Semantic Search. A user might pose a question of the form: ' What are the most relevant ways in which entity X is related to entity Y?' the response to which is a subgraph connecting X to Y. Relevance of the discovered subgraph therefore will depend on the amount of useful information conveyed to the user. This in turn depends on the meaning of the edges in the subgraph. We introduce heuristics that guide a discovery algorithm away from banal paths towards more informative ones. This guidance is based on weighting mechanisms (driven by edge semantics) for the edges in the RDF graph. We present an analysis of the quality of the subgraphs generated with respect to path ranking metrics. We then conclude presenting intuitions about which of our weighting schemes and heuristics produce higher quality subgraphs.

Organizations are taking advantage of ''data-mining'' techniques to leverage the vast amounts of data captured as they process routine transactions. Data mining is the process of discovering hidden structure or patterns in data. However, several of the pattern discovery methods in data-mining systems have the drawbacks that they discover too many obvious or irrelevant patterns and that they do not leverage to a full extent valuable prior domain knowledge that managers have. This research addresses these drawbacks by developing ways to generate interesting patterns by incorporating managers' prior knowledge in the process of searching for patterns in data. Specifically, we focus on providing methods that generate unexpected patterns with respect to managerial intuition by eliciting managers' beliefs about the domain and using these beliefs to seed the search for unexpected patterns in data. Our approach should lead to the development of decision-support systems that provide managers with more relevant patterns from data and aid in effective decision making. q 1999 Elsevier Science B.V. All rights reserved.

One of the popular trends in computer science has been development of intelligent web-based systems. Demand for such systems forces designers to make use of knowledge discovery techniques on web server logs. Web usage mining has become a major area of knowledge discovery on World Wide Web. Frequent pattern discovery is one of the main issues in web usage mining. These frequent patterns constitute the basic information source for intelligent web-based systems. In this paper; frequent pattern mining algorithms for web log data and their performance comparisons are examined. Our study is mainly focused on finding suitable pattern mining algorithms for web server logs.

A major source of revenue shrink in retail stores is the intentional or unintentional failure of proper checking out of items by the cashier. More recently, a few automated surveillance systems have been developed to monitor cashier lanes and detect non-compliant activities such as fake item checkouts or scans done with the intention of deriving monetary benefit. These systems use data from surveillance video cameras and transaction logs (TLog) recorded at the Point-of-Sale (POS). In this paper, we present a pattern discovery based approach to detect fraudulent events at the POS. Our approach is based on mining time-ordered text streams, representing retail transactions, formed from a combination of visually detected checkout related activities called primitives and barcodes from TLog data. Patterns representing single item checkouts, i.e. anchored around a single barcode, are discovered from these text streams using an efficient pattern discovery technique called Teiresias. The discovered patterns are used to build models for true and fake item scans by retaining or discarding the anchoring barcodes in those patterns respectively. A pattern matching and classification scheme is designed to robustly detect non-compliant cashier activities in the presence of noise in either the TLog or the video data. Different weighting schemes for quantifying the relative importance of the discovered patterns are explored: Frequency, Support Vector Machine (SVM) and Frequency+SVM. Using a large scale dataset recorded from retail stores, our approach discovers semantically meaningful cashier scan patterns. Our experiments also suggest that different weighting schemes result in varied false and true positive performances on the task of fake scan detection.

We are designing new data mining techniques on boolean contexts to identify a priori interesting concepts, i.e., closed sets of objects (or transactions) and associated closed sets of attributes (or items). We propose a new algorithm D-Miner for mining concepts under constraints. We provide an experimental comparison with previous algorithms and an application to an original microarray dataset for which D-Miner is the only one that can mine all the concepts.

Background: Pattern discovery in DNA sequences is one of the most fundamental problems in molecular biology with important applications in finding regulatory signals and transcription factor binding sites. An important task in this problem is to search (or predict) known binding sites in a new DNA sequence. For this reason, all subsequences of the given DNA sequence are scored based on an scoring function and the prediction is done by selecting the best score. By assuming no dependency between binding site base positions, most of the available tools for known binding site prediction are designed. Recently Tomovic and Oakeley investigated the statistical basis for either a claim of dependence or independence, to determine whether such a claim is generally true, and they presented a scoring function for binding site prediction based on the dependency between binding site base positions. Our primary objective is to investigate the scoring functions which can be used in known binding site prediction based on the assumption of dependency or independency in binding site base positions. Results: We propose a new scoring function based on the dependency between all positions in biding site base positions. This scoring function uses joint information content and mutual information as a measure of dependency between positions in transcription factor binding site. Our method for modeling dependencies is simply an extension of position independency methods. We evaluate our new scoring function on the real data sets extracted from JASPAR and TRANSFAC data bases, and compare the obtained results with two other well known scoring functions. Conclusion: The results demonstrate that the new approach improves known binding site discovery and show that the joint information content and mutual information provide a better and more general criterion to investigate the relationships between positions in the TFBS. Our scoring function is formulated by simple mathematical calculations. By implementing our method on several biological data sets, it can be induced that this method performs better than methods that do not consider dependencies.

Retail company's data may be geographically spread in different locations due to huge amount of data and rapid growth in transactions. But for decision making, knowledge workers need integrated data of all sites. Therefore the main challenge is to get generalized patterns or knowledge from the transactional data which is spread at various locations. Transporting data from those locations to server site increases the cost of transportation of data and at the same time finding patterns from huge data on the server increases the time and space complexity. Thus multi-database mining plays a vital role to extract knowledge from different data sources. Thus the technique proposed finds the patterns on various sites and instead of transporting the data, only the patterns from various locations get transported to the server to find final deliverable pattern. The technique uses the ranking algorithm to rank the items based on their profit, date of expiry and stock available at each location. Then association rule mining (ARM) is used to extract patterns based on ranking of items. Finally all the patterns discovered from various locations are merged using pattern merger algorithm. Proposed algorithm is implemented and experimental results are taken for both classical association rule mining on integrated data and for datasets at various sources. Finally all patterns are combined to discover actionable patterns using pattern merger algorithm given in section V.

We consider the interoperability of information systems within a distributed environment, such as across statistical organisations of the Member States of the European Union. Within a logical layer between the physical storage of the data (server) and the presentation of the data to the user (client), we introduce a conceptual data model that can facilitate interoperability between a number of different data models and structures in a distributed environment. For interoperability, we discuss a micro-macrometadata model that can interface readily with data warehouses and other cube object models and can exploit the accommodation of the operational metadata necessary for statistical query processing.

Application of data mining techniques to the World Wide Web, referred to as Web mining, has been the focus of several recent research projects and papers. However, there is no established vocabulary, leading to confusion when comparing research e orts. The term Web mining has been used in two distinct ways. The rst, called Web content mining in this paper, is the process of information discovery from sources across the World Wide Web. The second, called Web usage mining, is the process of mining for user browsing and access patterns. In this paper we de ne Web mining and present an overview of the various research issues, techniques, and development e orts. We brie y describe WEBMINER, a system for Web usage mining, and conclude this paper by listing research issues.

Scoring sentences in documents given abstract summaries created by humans is important in extractive multi-document summarization. In this paper, we formulate extractive summarization as a two step learning problem building a generative model for pattern discovery and a regression model for inference. We calculate scores for sentences in document clusters based on their latent characteristics using a hierarchical topic model. Then, using these scores, we train a regression model based on the lexical and structural characteristics of the sentences, and use the model to score sentences of new documents to form a summary. Our system advances current state-of-the-art improving ROUGE scores by ∼7%. Generated summaries are less redundant and more coherent based upon manual quality evaluations.

Removing objects that are noise is an important goal of data cleaning as noise hinders most types of data analysis. Most existing data cleaning methods focus on removing noise that is the result of low-level data errors that result from an imperfect data collection process, but data objects that are irrelevant or only weakly relevant can also significantly hinder data analysis. Thus, if the goal is to enhance the data analysis as much as possible, these objects should also be considered as noise, at least with respect to the underlying analysis. Consequently, there is a need for data cleaning techniques that remove both types of noise. Because data sets can contain large amounts of noise, these techniques also need to be able to discard a potentially large fraction of the data. This paper explores four techniques intended for noise removal to enhance data analysis in the presence of high noise levels.

The explosive growth of Internet has given rise to many websites which maintain large amount of user information. To utilize this information, identifying usage pattern of users is very important. Web usage mining is one of the processes of finding out this usage pattern and has many practical applications. Our paper discusses how association rules can be used to discover patterns in web usage mining. Our discussion starts with preprocessing of the given weblog, followed by clustering them and finding association rules. These rules provide knowledge that helps to improve website design, in advertising, web personalization etc.

Pattern-based Java bytecode compression techniques rely on the identification of identical instruction sequences that occur more than once. Each occurrence of such a sequence is substituted by a single instruction. The sequence defines a pattern that is used for extending the standard bytecode instruction set with the instruction that substitutes the pattern occurrences in the original bytecode. Alternatively, the pattern may be stored in a dictionary that serves for the bytecode decompression. In this case, the instruction that substitutes the pattern in the original bytecode serves as an index to the dictionary. In this paper, we investigate a bytecode compression technique that considers a more general case of patterns. Specifically, we employ the use of an advanced pattern discovery technique that allows locating patterns of an arbitrary length, which may contain a variable number of wildcards in place of certain instruction opcodes or operands. We evaluate the benefits and the limitations of this technique in various scenarios that aim at compressing the reference implementation of MIDP, a standard Java environment for the development of applications for mobile devices.

Inductive databases (IDBs) represent a database view on data mining and knowledge discovery. IDBs contain not only data, but also generalizations (patterns and models) valid in the data. In an IDB, ordinary queries can be used to access and manipulate data, while inductive queries can be used to generate (mine), manipulate, and apply patterns. In the IDB framework, patterns become "first-class citizens" and KDD becomes an extended querying process in which both the data and the patterns/models that hold in the data are queried. The IDB framework is appealing as a general framework for data mining, because it employs declarative queries instead of ad-hoc procedural constructs. As declarative queries are often formulated using constraints, inductive querying is closely related to constraint-based data mining. The IDB framework is also appealing for data mining applications, as it supports the entire KDD process, i.e., nontrivial multi-step KDD scenarios, rather than just individual data mining operations. The goal of the workshop was to bring together database and data mining researchers interested in the areas of inductive databases, inductive queries, constraint-based data mining, and data mining query languages.

Scientific measurements are often depicted as line graphs. Stateof-the-art high throughput systems in life sciences, telemetry and electronics measurement rapidly generate hundreds to thousands of such graphs. Despite the increasing volume and ubiquity of such data, few software systems provide efficient interactive management, navigation and exploratory analysis of large line graph collections. To address these issues, we have developed Line Graph Explorer (LGE). LGE is a novel and visually scalable line graph management system that supports facile navigation and interactive visual analysis. LGE provides a compact overview of the entire collection by encoding the y-dimension of individual line graphs with color instead of space, thus enabling the analyst to see major common features and alignments of the data. Using Focus+Context techniques, LGE provides interactions for viewing selected compressed graphs in detail as standard line graphs without losing a sense of the general pattern and major features of the collection. To further enhance visualization and pattern discovery, LGE provides sorting and clustering of line graphs based on similarity of selected graph features. Sequential sorting by associated line graph metadata is also supported. We illustrate the features and use of LGE with examples from meteorology and biology.

One of the popular trends in computer science has been development of intelligent web-based systems. Demand for such systems forces designers to make use of knowledge discovery techniques on web server logs. Web usage mining has become a major area of knowledge discovery on World Wide Web. Frequent pattern discovery is one of the main issues in web usage mining. These frequent patterns constitute the basic information source for intelligent web-based systems. In this paper; frequent pattern mining algorithms for web log data and their performance comparisons are examined. Our study is mainly focused on finding suitable pattern mining algorithms for web server logs.

Most modern lossless data compression techniques used today, are based in dictionaries. If some string of data being compressed matches a portion previously seen, then such string is included in the dictionary and its reference is included every time it appears. A possible generalization of this scheme is to consider not only strings made of consecutive symbols, but more general patterns with gaps between its symbols. The main problems with this approach are the complexity of pattern discovery algorithms and the complexity for the selection of a good subset of patterns. In this paper we address the last of these problems. We demonstrate that such problem is NP-complete and we provide some preliminary results about heuristics that points to its solution.

Several pattern discovery methods proposed in the data mining literature have the drawbacks that they discover too many obvious or irrelevant patterns and that they do not leverage to a full extent valuable prior domain knowledge that decision makers have. In this paper we propose a new method of discovery that addresses these drawbacks. In particular we propose a new method of discovering unexpected patterns that takes into consideration prior background knowledge of decision makers. This prior knowledge constitutes a set of expectations or beliefs about the problem domain. Our proposed method of discovering unexpected patterns uses these beliefs to seed the search for patterns in data that contradict the beliefs. To evaluate the practicality of our approach, we applied our algorithm to consumer purchase data from a major market research company and to web logfile data tracked at an academic Web site and present our findings in the paper.

The study of metabolic pathways is becoming increasingly important to exploit an integrated, systems-level approach for optimizing a desired cellular property or phenotype. In this context, the integration of genomics data with genetic, metabolic and regulatory models is essential because the systematic design of artificial, biological systems requires the identification of robust building blocks like gene promoters, metabolic pathways or genetic circuits taken from natural organisms, and manipulated to develop ad hoc features. Computational tools allowing precise descriptions of natural pathways might thus allow improving the performance of artificial routes. In this review, we introduce the most recent bioinformatics tools enabling detailed characterizations of metabolic pathways in bacteria from different perspectives. Matteo Brilli is a postdoc at the University of Florence; he works on computational molecular evolution and metabolic pathways characterization with a specific interest in the evolution of regulatory mechanisms. Renato Fani is professor of Genetics at the University of Florence. In the last 20 years, he has studied the evolution of metabolic pathways both from the experimental and bioinformatics perspective.

To engage visitors to a Web site at a very early stage (i.e., before registration or authentication), personalization tools must rely primarily on clickstream data captured in Web server logs. The lack of explicit user ratings as well as the sparse nature and the large volume of data in such a setting poses serious challenges to standard collaborative filtering techniques in terms of scalability and performance. Web usage mining techniques such as clustering that rely on offline pattern discovery from user transactions can be used to improve the scalability of collaborative filtering, however, this is often at the cost of reduced recommendation accuracy. In this paper we propose effective and scalable techniques for Web personalization based on association rule discovery from usage data. Through detailed experimental evaluation on real usage data, we show that the proposed methodology can achieve better recommendation effectiveness, while maintaining a computational advantage over direct approaches to collaborative filtering such as the k-nearest-neighbor strategy.