Encyclopedia of Data Warehousing and Mining

Transcription

1 Encyclopedia of Data Warehousing and Mining Second Edition John Wang Montclair State University, USA Volume III K-Pri Information Science reference Hershey New York

2 Director of Editorial Content: Director of Production: Managing Editor: Assistant Managing Editor: Typesetter: Cover Design: Printed at: Kristin Klinger Jennifer Neidig Jamie Snavely Carole Coulson Amanda Appicello, Jeff Ash, Mike Brehem, Carole Coulson, Elizabeth Duke, Jen Henderson, Chris Hrobak, Jennifer Neidig, Jamie Snavely, Sean Woznicki Lisa Tosheff Yurchak Printing Inc. Published in the United States of America by Information Science Reference (an imprint of IGI Global) 701 E. Chocolate Avenue, Suite 200 Hershey PA Tel: Fax: Web site: and in the United Kingdom by Information Science Reference (an imprint of IGI Global) 3 Henrietta Street Covent Garden London WC2E 8LU Tel: Fax: Web site: Copyright 2009 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher. Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark. Library of Congress Cataloging-in-Publication Data Encyclopedia of data warehousing and mining / John Wang, editor. -- 2nd ed. p. cm. Includes bibliographical references and index. Summary: "This set offers thorough examination of the issues of importance in the rapidly changing field of data warehousing and mining"--provided by publisher. ISBN (hardcover) -- ISBN (ebook) 1. Data mining. 2. Data warehousing. I. Wang, John, QA76.9.D37E dc British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library. All work contributed to this encyclopedia set is new, previously-unpublished material. The views expressed in this encyclopedia set are those of the authors, but not necessarily of the publisher. If a library purchased a print copy of this publication, please go to for information on activating the library's complimentary electronic access to this publication.

3 1346 Section: Multi-Agent Systems A Multi-Agent System for Handling Adaptive E-Services Pasquale De Meo Università degli Studi Mediterranea di Reggio Calabria, Italy Giovanni Quattrone Università degli Studi Mediterranea di Reggio Calabria, Italy Giorgio Terracina Università degli Studi Della Calabria, Italy Domenico Ursino Università degli Studi Mediterranea di Reggio Calabria, Italy INTRODUCTION An Electronic-Service (E-Service) can be defined as a collection of network-resident software programs that collaborate for supporting users in both accessing and selecting data and services of their interest present in a provider site. Examples of e-services are e-commerce, e-learning, e-government, e-recruitment and e-health applications. E-Services are undoubtely one of the engines presently supporting the Internet Revolution. Indeed, nowadays, a large number and a great variety of providers offer their services also or exclusively via the Internet. BACKGROUND In spite of their spectacular development and present relevance, E-Services are far to be considered a stable technology and various improvements could be thought for them. Many of the present suggestions for improving them are based on the concept of adaptivity, i.e., on the capability to make them more flexible in such a way as to adapt their offers and behaviour to the environment they are operating in. In this context, systems capable of constructing, maintaining and exploiting suitable profiles for users accessing E-Services appear capable of playing a key role in the future (Kobsa, 2007). Both in the past and in the present, various E-Service providers exploit (usually rough) user profiles for proposing personalized offers. However, in most cases, the profile construction methodology adopted by them presents some problems. In fact, it often requires a user to spend a certain amount of time for constructing and updating his profile; in addition, the profile of a user stores only information about the proposals which he claims to be interested in, without considering other ones, somehow related to those just provided, possibly interesting him in the future and that he disregarded to take into account in the past. In spite of present user profile handlers, generally, when accessing an E-Service, a user must personally search the proposals of his interest through it. We argue that, for improving the effectiveness of E-Services, it is necessary to increase the interaction between the provider and the user, on one hand, and to construct a rich profile of the user, taking his interests, needs and past behaviour into account, on the other hand. In addition, a further important factor must be taken into account. Nowadays, electronic and telecommunications technology is rapidly evolving in such a way as to allow cell phones, palmtops and wireless PDAs to navigate on the Web. These mobile devices do not have the same display or bandwidth capabilities as their desktop counterparts; nonetheless, present E-Service providers deliver the same contents to all device typologies (Communications of the ACM, 2002; ; Smith, Cotter & Oman, 2007). In the past, various approaches have been proposed for handling E-Service activities; some of them are agent-based. As an example: In (Anand, Kearney & Shapcott, 2007) an approach to helping users looking for relevant items Copyright 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.

4 is described. In order to generate its recommendations, this approach integrates user ratings with an ontology describing involved items. This approach is particularly suited for the e-commerce domain. In (Mahmood & Ricci, 2007) a travel recommender system, based on the Intelligent Agent technology, is presented. This system builds user profiles by exploiting Reinforcement Learning techniques, and models the recommendation process as a Markov Decision Process. In (Medjahed & Bouguettaya, 2005) the Authors propose WebSenior, a system using ontologies to automatically generate Web Services customized to senior citizen needs and government laws. WebSenior is able to manage both simple and composite services. In order to generate these last, it defines a graph, called dependency diagram, in which each vertex is associated with a simple service and each edge denotes a dependency relationship between a pair of services. The problem of computing composite services is, then, regarded as the problem of computing paths in the dependency diagram; this last problem is solved by applying the Floyd-Warshall dynamic programming algorithm. In (Ahn, Brusilovsky, Grady, He & Syn, 2007) YourNews, a system capable of helping users in accessing news located on the Web, is proposed. YourNews relies on a user profile built by unobtrusively monitoring user behaviour; this profile is open, in the sense that the corresponding user can interactively provide feedbacks that will be exploited by YourNews to enhance its accuracy. In (De Meo, Quattrone, Terracina & Ursino, 2007) a XML-based multi-agent recommender system for supporting online recruitment services is presented. This system handles user profiles for supporting a personalized job search over the Internet. In order to perform its recommendations, it exploits some advanced techniques (e.g., least square error and Pavlovian learning). In (Fang & Sheng, 2005) ServiceFinder, a system conceived for supporting citizens in their selection of relevant e-government services, is proposed. ServiceFinder uses Web Mining techniques to discover the N services best matching user needs and modifies the home page of an institutional e-government portal by adding to it N hyperlinks pointing to these services. In (Srivihok & Sukonmanee, 2005) a system capable of supporting e-tourism activities is proposed. This system analyzes past user behaviours and applies the Q-Learning algorithm to build a user profile. After this, it applies a reinforcement algorithm on both user and trip profiles in such a way as to associate a score with each trip proposal. These last are, then, ranked on the basis of their scores and only the top five are presented to the user. All these systems construct, maintain and use rich data structures regarding both user needs and behaviours; therefore, we can consider them adaptive w.r.t. the user; however, none of them is adaptive w.r.t. the device. On the other side, in many areas of computer science research, a large variety of approaches that adapt their behaviour on the basis of the device the user is exploiting, has been proposed. As an example: In (Samaras & Panayiotou, 2004) the system mpersona, aiming to support users equipped with wireless devices to access information sources located on the Web, is proposed. mper- SONA relies on a mobile multi-agent architecture; it associates a user profile (consisting of a set of keywords) with each user and represents the contents of an information source as a hierarchy (called metadata tree). Each time a user submits a query to an information source, mpersona isolates the portion of hierarchy (and the corresponding information objects) best fitting his requests; after this, it considers the features of the device he is currently exploiting and adapts the selected contents to them. In (Lee, Kang, Choi & Yang, 2006) an approach to Web content adaptation for mobile users is presented. This approach stores user preferences in a suitable profile. When a user is involved in Web browsing activities, it partitions a Web page into blocks, filters out those judged unnecessary and sorts the other ones on the basis of their relevance to the user. Finally, it presents sorted blocks to the user. M 1347

5 In (Muntean & McManis, 2006) an approach to supporting mobile users in their browsing activities is proposed. This approach considers both user preferences (encoded in a user profile) and device features to suitably modify the content of the Web pages visited by the user; for instance, it can compress or eliminate images, if their sizes are excessive. In (Wei, Bhandarkar & Li, 2007) an approach to delivering multimedia data, taking both user preferences and device constraints into account, is proposed. This approach estimates how much a multimedia information object fits both the exigencies of a user and the constraints of the device he is currently exploiting. This information is used to define a knapsack-like problem whose solution allows the selection of the objects to deliver. These approaches are particularly general and interesting; however, none of them has been conceived for handling E-Services. MAIN THRUST OF THE CHAPTER Challenges to Face In order to overcome the problems outlined above, some challenges must be tackled. Firstly, a user can access many E-Services, operating in the same or in different application contexts; a faithful and complete profile of him can be constructed only by taking his behaviour on accessing all these sites into account, and by storing the corresponding profile in a unique data structure on his side. Secondly, for a given user and E-Service provider, it should be possible to compare the profile of the user with the offers of the provider for extracting those proposals that probably will interest him. Existing techniques for satisfying such a requirement are mainly based on either log files or cookies. In the former case they cannot match user preferences and E-Service proposals; in the latter one they need to know and exploit some personal information that a user might consider private. Thirdly, the typical one-size-fits-all philosophy of present E-Service providers should be overcome by developing systems capable of adapting their behaviour to both the profile of the user and the characteristics of the device he is exploiting for accessing them (Communications of the ACM, 2002). Fourthly, relationships and interactions among users should be taken into account; in fact, they can be exploited to guarantee a certain proactivity degree to the E-Service provider, on one hand, and to create communities of citizens showing similar needs and desires, on the other hand. System Description The system we present in this chapter (called E-Service Adaptive Manager, ESA-Manager) aims to tackle all challenges mentioned above. It is a multi-agent system for handling user accesses to E-Services, capable of adapting its behaviour to both user and device profiles. In ESA-Manager a Service Provider Agent is present for each E-Service provider, handling the proposals stored therein, as well as the interaction with users. In addition, an agent is associated with each user, adapting its behaviour to the profiles of both the user and the device he is currently exploiting. Actually, since a user can access E-Service providers by means of different devices, his profile cannot be stored in only one of them; indeed, it must be handled and stored in a support different from the devices generally exploited by him for accessing E-Service providers. As a consequence, it appears compulsory the exploitation of a Profile Agent, storing the profiles of both involved users and devices, and a User-Device Agent, associated with a specific user operating by means of a specific device, supporting him in his activities. As a consequence of the previous reasoning, for each user, a unique profile is mined and maintained, storing information about his behaviour in accessing all E-Service providers. In this way, ESA-Manager tackles the first challenge mentioned above. Whenever a user accesses an E-Service by means of a certain device, the corresponding Service Provider Agent sends information about its proposals to the User- Device Agent associated with him and the device he is exploiting. The User-Device Agent determines similarities among the proposals presented by the provider and the interests of the user. For each of these similarities, the Service Provider Agent and the User-Device Agent cooperate to present to the user a group of Web 1348

6 pages, adapted to the exploited device, illustrating the proposal. We argue that this behaviour provides ESA- Manager with the capability of supporting the user in the search of proposals of his interest delivered by the provider. In addition, the algorithms underlying ESA- Manager allow it to identify not only the proposals probably of interest to him in the present but also other ones possibly of interest to him in the future and that he disregarded to take into account in the past. In our opinion, this is a particularly interesting feature for a new approach devoted to deal with E-Services. Last, but not the least, it is worth observing that, since user profile management is carried out at the user side, no information about the user profile is sent to E-Service providers. In this way, ESA-Manager solves privacy problems left open by cookies. All reasonings presented above show that ESA-Manager is capable of tackling also the second challenge mentioned previously. In ESA-Manager device profile plays a central role. Indeed, the proposals of a provider shown to a user, as well as their presentation formats, depend on the characteristics of the device the user is presently exploiting. However, the ESA-Manager capability of adapting its behaviour to the device the user is exploiting is not restricted to the presentation format of the proposals; in fact, exploited device can influence also the computation of the interest degree associated with the proposals presented by each provider. Specifically, one of the parameters which the interest degree associated with a proposal is based on, is the time the user spends in visiting the corresponding Web pages. In ESA-Manager, this time is not considered as an absolute measure, but it is normalized w.r.t. both the characteristics of the exploited device and the navigation costs. This reasoning allows us to argue that ESA-Manager tackles also the third challenge mentioned above. In addition, ESA-Manager uses a Social Network to partition citizens accessing the system into homogeneous clusters called communities; these communities are set up on the basis of shared interests/needs of their members rather than on demographic data. Each time a citizen submits a query, ESA-Manager forwards the corresponding answers (i.e., the corresponding service proposals) also to the members of his community. All these activities are carried out by a Social Network Agent which catches user queries and exploits them to both organize communities and handle communications with them. This confers a high level of proactivity to our system because it can recognize and recommend potentially relevant services to a citizen even though he is not aware of their existence. In addition, if, for a citizen, interests/needs have been changed, the Social Network Agent automatically assigns him to another community and, at the same time, suggests him the latest services recommended to the members of his new community. In this way, our system tackles also the fourth challenge mentioned above. A final important characteristic of our system is that it encourages a socially inspired mechanism for service management. In fact, users can freely discuss among them to propose new services; as a consequence, the top-down approach for service delivery (in which providers push their services to users and impose them their formalism, terminology, access policy, features, etc.) is coupled and, hopefully, substituted by a bottomup one (in which users join up to form communities who raise their requirements about desired services to providers). This provides our system with notable social capabilities in that it can suitably channelize user demands. In fact, in this scenario, users are encouraged to debate in such a way as to propose the activation of new services of interest to them. FUTURE TRENDS The spectacular growth of the Internet during the last decade has strongly conditioned the E-Service landscape. Such a growth is particularly surprising in some application domains, such as financial services or e-government. As an example, the Internet technology has enabled the expansion of financial services by integrating the already existing, quite variegate, financial data and services and by providing new channels for information delivery. However, E-Services are not a leading paradigm only in business contexts; for instance, they are vigorously applied by governmental units at national, regional and local levels around the world. A further interesting research trend consists of investigating the possibility to integrate an E-Service access system with a Decision Support one; in this way, user behaviour could be analyzed (for instance, by means of a Data Mining tool) for determining the key features of the most appreciated services. This information could be particularly useful for provider managers when they need to decide the new services to propose. This last feature would allow the realization of a hybrid E- M 1349

7 Service system, embodying the functionalities of both seeker-oriented and company-oriented systems. CONCLUSION In this paper we have proposed ESA-Manager, an adaptive multi-agent system for supporting a user, accessing an E-Service provider, in the search of proposals appearing to be appealing according to his interests and behaviour. We have shown that ESA-Manager is adaptive w.r.t. the profile of both the user and the device he is currently exploiting. Finally, we have seen that it is proactive in that it suggests potentially interesting services to a user even if he did not explicitly require them. As for future work, we would like to investigate whether a hierarchical clustering algorithm can be fruitfully exploited to improve the quality of user community management. The output of this algorithm would be a tree-like data structure whose nodes represent potential communities. This data structure would allow specialization/generalization relationships among communities to be handled; in fact, nodes placed at the top of the hierarchy would represent wide communities of loosely linked users (e.g., the community of students), whereas nodes at the bottom would be associated with narrow communities of tightly linked users (e.g., the community of working students). This hierarchical community organization could be extremely useful in the proactive suggestion of services, carried out by our system. In fact, in order to find the citizen communities potentially interested to a service, our system could run across the hierarchy, starting from its root, in such a way as to find the widest community potentially interested to it. REFERENCES Ahn, J., Brusilovsky, P., Grady, J., He, D. & Syn, S.Y. (2007). Open user profiles for adaptive news systems: help or harm?. Proc. of the International Conference on World Wide Web, pages 11-20, Banff, Alberta, Canada. ACM Press. Anand, S.S., Kearney, P. & Shapcott, M. (2007). Generating semantically enriched user profiles for Web personalization. ACM Transactions on Internet Technologies 7(4), Article N. 22. Communications of the ACM (2002). Adaptive Web. Volume 45(5). ACM Press. De Meo, P., Quattrone, G., Terracina, G. & Ursino, D. (2007). An XML-based Multi-Agent System for Supporting Online Recruitment Services. IEEE Transactions on Systems, Man and Cybernetics, 37(4), Dolog, P., Simon, B., Klobucar, T. & Nejdl, W. (2008). Personalizing Access to Learning Networks. ACM Transactions on Internet Technologies. 8(2), Forthcoming Fang, X. & Sheng, O.R.L. (2005). Designing a better Web portal for digital government: a Web-mining based approach. Proc. of the National Conference on Digital Government Research, pages , Atlanta, Georgia, USA. Digital Government Research Center. Kobsa A.(2007). Generic User Modeling Systems. In The Adaptive Web, Methods and Strategies of Web Personalization. P. Brusilovsky, A. Kobsa and W. Nejdl (Eds.), pages , Springer. Lee, E., Kang, J., Choi, J. & Yang, J. (2006). Topic- Specific Web Content Adaptation to Mobile Devices. Proc. of the IEEE/WIC/ACM International Conference on Web Intelligence, pages , Hong Kong. IEEE Computer Society Press. Mahmood, T. & Ricci, F. (2007). Learning and adaptivity in interactive recommender systems. Proc. of the International Conference on Electronic Commerce, pages 75-84, Minneapolis, Minnesota, USA. ACM Press. Medjahed, B. & Bouguettaya, A. (2005). Customized delivery of E-Government Web Services. IEEE Intelligent Systems, 20(6), Muntean, C.H. & McManis, J. (2006). Fine grained content-based adaptation mechanism for providing high end-user quality of experience with adaptive hypermedia systems. Proc. of the International Conference on World Wide Web, pages 53-62, Edinburgh, Scotland, UK. ACM Press. Samaras, G. & Panayiotou, C. (2004). mpersona: Personalized Portals for the Wireless User: An Agent Approach. Mobile Networks and Applications, 9(6),

8 Smyth, B., Cotter, P. & Oman S. (2007). Enabling Intelligent Content Discovery on the Mobile Internet. Proc. of the AAAI Conference on Artificial Intelligence. pages , Vancouver, British Columbia, Canada, AAAI Press. Srivihok, A. & Sukonmanee, P. (2005). E-commerce intelligent agent: personalization travel support agent using Q Learning. Proc. of the IEEE International Conference on Electronic Commerce, pages , Xian, China. ACM Press. Wei, Y., Bhandarkar, S.M. & Li, K. (2007). Video personalization in resource-constrained multimedia environments. Proc. of the International Conference on Multimedia, pages , Augsburg, Germany. ACM Press. KEy TERMS Agent: A computational entity capable of both perceiving dynamic changes in the environment it is operating in and autonomously performing user delegated tasks, possibly by communicating and cooperating with other similar entities. Agent Ontology: A description (like a formal specification of a program) of the concepts and relationships that can exist for an agent or a community of agents. Adaptive System: A system adapting its behaviour on the basis of the environment it is operating in. Device Profile: A model of a device storing information about both its costs and capabilities. E-Service: A collection of network-resident software programs that collaborate for supporting users in both accessing and selecting data and services of their interest handled by a provider site. Multi-Agent System (MAS): A loosely coupled network of software agents that interact to solve problems that are beyond the individual capacities or knowledge of each of them. User Modeling: The process of gathering information specific to each user either explicitly or implicitly. This information is exploited in order to customize the content and the structure of a service to the user s specific and individual needs. User Profile: A model of a user representing both his preferences and his behaviour. M 1351