News: * Submission site will open the first week of May. * New deadlines adjusted to other KDD workshops. * Notification is delayed due to delays in reviews * Tentative Program posted

Important Dates June 6, 2005: Abstract Submission June 6, 2005: Paper Submission June 27, 2005: Notification July 18, 2005: Camera-ready Submission August 21, 2005: Workshop

WebKDD 2005 Workshop on Knowledge Discovery in the Web August 21st, 2005, Chicago, Illinois, USA http://db.cs.ualberta.ca/webkdd05/ webkdd05 @ cs.ualberta.ca Held in conjunction with The Eleventh ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Keynote Speaker: Dr. Charu Aggarwal, T.J. Watson Lab, IBM
On Change Diagnosis and Monitoring in Data Streams

Workshop Description

The Web presents a key driving force for a large spectrum of applications in which a user interacts with a company, a governmental authority, a non-governmental organization or other non-profit institution or other users. User preferences and expectations, together with usage patterns, form the basis for personalized, user-friendly and optimal services. Key metrics enabled by proper data capture and processing are essential to run an effective business or service. Enabling technologies include data mining, scalable warehousing and preprocessing, sequence discovery, real time processing, document classification, user modeling and quality evaluation models for them. Recipient technologies that demand for user profiling and usage patterns include recommendation systems, Web analytics applications, application servers coupled with content management systems and fraud detectors. Most previous emphasis seem to have eluded the fact that Web access patterns tend to be very dynamic in nature, due not only to the dynamics of Web site content and structure, but also to changes in the user's interests, and thus their navigation patterns. The access patterns can be observed to change depending on the time of day, day of week, and according to seasonal patterns or other external events in the world that may alter all the different facets of the Web users experience, such as usage, content, structure, and semantics. In addition to the evolving nature of usage patterns, current Websites can generate massive data throughput. Can an intelligent Web usage mining system continuously learn in the presence of such massive and evolving data streams without any ungraceful stoppages or reconfigurations? The sheer mass of Web clickstreams can easily force Web usage mining to process each new data item no more than once Thus Web clickstreams, as the name implies, can be considered as an instance of data streams, where the ruling constraint is Once you see it, you can't look back!. For all these reasons, it is worthwhile to consider Web usage data as an evolving data stream, thus calling for a synergy between stream data mining and web usage mining. Furthermore, the inherent and increasing heterogeneity of the Web has required Web-based applications to more effectively integrate a variety of types of data across multiple channels and from different sources such as content, structure, and more recently, semantics. A focus on techniques and architectures for more effective exploitation and mining of such multi-faceted data is likely to lead to the next generation of more useful and more intelligent applications. WebKDD2005 is interested in techniques that enhance Web usage mining through the use of other knowledge channels and sources, and not just information integration in the traditional sense. These considerations can help answer questions such as Can a web usage mining system reason about the discovered (usage) patterns or user models? And Can recommender systems explain their recommendation to users? Finally, another imminent issue that threatens to become more important in the future is the issue of vulnerability in recommender systems. Can an intelligent recommender system be designed to resist various malicious manipulations, such as schilling attacks that try to alter user ratings to influence the recommendations?. This motivates the need to study and design robust recommender systems. The WebKDD'2005 workshop aims to bring together practitioners and researchers with a specific focus on the emerging trends and industry needs, as well as an emphasis on mining massive, evolving, and multifaceted Web data, as well as emerging problems and applications.

Download a call for papers in PDF format.

Important Dates June 6, 2005: Electronic submission of titles and abstracts June 6, 2005: Electronic submission of full papers June 27, 2005: Author notification July 11, 2005: Submission of Camera-ready papers (hard deadline) August 21, 2005: Workshop in Chicago, Illinois. Contact: For inquiries send e-mail to webkdd05 @ cs.ualberta.ca Co-Chairs Olfa Nasraoui, University of Louisville, USA Osmar Zaïane, University of Alberta, Canada Myra Spiliopoulou, Otto-von-Guericke-University Magdeburg, Germany Bamshad Mobasher, DePaul University, USA Philip Yu, IBM T. J. Watson, USA Brij Masand, Data Miners, Inc., USA