|
Professor |
|
 |
School of Computing,
Phone: (312) 362-5174
|
|
|
|
|
Professor |
|
|
|
School of Computing,
Phone: (312) 362-5174
|
|
|
|
- Course Web Sites
- ECT 584 - Web Data Mining for Business Intelligence
- DS 575/IS 575 - Intelligent Information Retrieval
- CSC 480 / 380 - Foundations of Artificial Intelligence
- CSC 426 - Values in Computer Technology
- Research Interests
- Data mining, Web mining
- Web personalization and Recommender Systems
- Intelligent agents for the World Wide Web
- Multi-agents systems, and agent-based virtual markets
- Logic programming and non-monotonic reasoning
See my research profile on: [Google Scholar] [CiteSeer] [DBLP] [DBLife].
- Publications
A chronological list of selected publications that are available online.
- Recent and Noteworthy Events
Invited Talk at Recommenders'06 - Present and Future of Recommender Systems in Bilbao, Spain, on securing recommender systems against malicious attacks. You can download presentation or watch the video. Recommenders 2007 (Sponsored by ACM) will be held in Minneapolis, October 19-21, 2007.
Book Announcement: Advances in Web Mining and Web Usage Analysis, by B. Mobasher, O. Nasraoui, B. Liu, B.Masand (eds.), Lecture Notes in Artificial Intelligence (LNAI 3932), Springer, 2006. This book constitutes the thoroughly refereed and extended articles from the 6th International Workshop on Mining Web Data, WEBKDD 2004, held at the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2004. The 11 chapters were carefully selected and reviewed for inclusion in the book. Topic areas include: Web usage analysis and user modeling, Web personalization and recommender systems, search personalization, and semantic Web mining.
Book Announcement: Intelligent Techniques for Web Personalization, by Bamshad Mobasher and Sarabjot Singh Anand (eds.), Lecture Notes in Artificial Intelligence (LNAI 3169), Springer, 2005. The book consists of 17 thoroughly edited contributed chapters providing full coverage of issues related to Web Personalization, including: user modeling, recommender systems, enabling technologies, personalized information access, and systems and applications. [Read the introductory chapter] Recent TV Interview on Data Mining, Privacy, and Security - in: "Politics in Perspective with Kevin McDermott", aired in the Chicago area. View the program Online (only in Windows Media)
Presentation at Microsoft Research on the NSF-funded project "Secure Personalization" (Dec. 2005). View the lecture online at the Research Channel: Secure Personalization: Towards Trustworthy Recommender Systems. - Publications in Specific Areas
- Web Personalization, Recommender Systems
- Secure Recommender Systems
- Semantically Enhanced Web Mining and Personalization
- Web Usage Mining
- Data Preparation for Web Usage Mining
- Web Content Mining, Text Mining
- Intelligent Agents
- Multi-Agent Systems, Virtual Markets, Agent-Based Contracting
- Computational Logic, Logic Programming, Non-Monotonic Reasoning
- Related Workshops, Tutorials and Events
- 6th Workshop on Intelligent Techniques for Web Personalization & Recommender Systems,
at AAAI 2008. July 13, 2008, Chicago.- WebKDD 2008: Web Mining and Web Usage Analysis - 10 year anniversary workshop,
at SIGKDD 2008, August 24, 2008.- 5th Workshop on Intelligent Techniques for Web Personalization,
at AAAI 2007. July 23, 2007, Vancouver, British Columbia, Canada.- Joint WebKDD and SNA-KDD Workshop: Web Mining and Social Network Analysis.
Held at ACM SIGKKD 2007, August 12-15, 2007, San Jose , California- Recommenders'06 - Present and Future of Recommender Systems.
Organized by MyStrands. Bilbao, Spain, September. 2006.- ECAI Workshop on Recommender Systems, Riva del Garda, Italy, August 2006
- 4th Workshop on Intelligent Techniques for Web Personalization,
at AAAI 2006, July 16-20, Boston, 2006.- Tutorial on Web Usage Mining for E-business Applications,
at ECML/PKDD 2002, [Tutorial Slides in PDF].- KDD for Personalization Tutorial, at ECML/PKDD 2001, [Tutorial Slides in PDF].
- Workshop on Semantic Web Mining, at ECML/PKDD 2002, August 2002, Helsinki, Finland.
Demos and Projects
Secure Personalization: Building Trustworthy Recommender Systems [Funded by NSF Cyber Trust program]
The purpose of this project is to explore the vulnerabilities of recommendation and personalization systems in the face of malicious attacks, explore techniques for enhancing their robustness, and examine methods by which attacks can be recognized and possibly defeated. Most research in computer security focuses on protecting assets inside an organization's security perimeter from unauthorized access and modification. This project examines the problem of security for open systems that are designed to be accessed and modified by the general public. How do we protect such a system from the legal but biased inputs of an attacker trying to subvert its functionality? The goal is to advance our understanding of the trustworthiness of recommender systems, now a crucial component in many areas from e-commerce and e-learning to content management systems. We have explored the spectrum of possible attacks against recommendation systems, and developed models characterizing these attacks and their impacts. We are examining a range of recommendation algorithms including user-based, item-based and model-based collaborative recommenders, and also explore hybrid recommendation by combining collaborative recommendation techniques with content-based and knowledge-based ones. Informed by these results, we are considering how recommender systems can be secured, through improved algorithms but also by detecting attacks and responding appropriately. See relevant results and publications related to this project.Web Personalizer - Automatic Personalization Based on Web Mining
This project involves automatic and real-time personalization of users' navigational experience within a site based on aggregate user access patterns discovered through Web usage mining. The specific mining techniques include the identification of user transactions, association rule discovery, and the derivation of usage profiles through clustering transactions and/or pageviews. The online component of the system combines the discovered knowledge with an active user session to customize pages based on user's actions. A general overview of the system is available in HTML. For a more detailed overview of issues and techniques related to Web usage mining for personalization, please read the paper: Web Usage Mining and Personalization (PDF), a chapter in Practical Handbook of Internet Computing, Munindar P. Singh (ed.), CRC Press, 2005. Other related publications and results are available online. We have also studied methods, techniques, and architectures for integrating semantic knowledge such as that available from domain ontologies or through automatic content analysis with Web usage mining and user profiling to create more effective adaptive and personalized systems. For more information on this work please see related publications on Semantically Enhanced Web Mining and Personalization.
ARCH - An Adaptive Agent for Retrieval Based on Concept Hierarchies
ARCH combines domain knowledge inherent in Web-based classification hierarchies such as Yahoo together with automatically learned long-term user profiles and user's short term interests (expressed as a query or implicitly through navigation) to determine the user's information access context. In the current implementation, this context takes the form of a re-formulated query which is used for Web search. A more detailed discussion of ARCH and a list of relevant publications is available here. A demonstration of this version of ARCH using a portion of the Yahoo concept hierarchy can be found at http://arch.ahusieg.com/. The current directions in this project is to extend the methods and the architecture for Search Personalization. A user's context is represented through an "ontological user profile", an instance of the underlying domain classification hierarchy annotated by node weights representing user's evolving interests. The interest weights on concepts in the hierarchy are updated automatically as a result of user interactions with Web documents or search results. If the user performs a search (on a standard search engine), the resulting documents are first classified into the concept hierarchy, and then re-ranked according to the interest weights in the user's ontological profile.
