Invited speaker : Marzena KRYSZKIEWICZ

Talk : Concise representations of rules

Rules are one of basic types of knowledge, which express dependencies between sets of attributes (features, items). There is a number of statistical measures that express significance of rules. The most typical are confidence and support. A rule is considered strong if their support and confidence exceed required threshold values. The number of strong rules is often huge, which limits their usefulness. Applying concise rule representations can lessen the problem. Typically, a rule representation consists of a set of rules and an inference mechanism, which enables derivation of other rules. Ideally, a rule representation should be lossless (should enable derivation of all strong rules), sound (should forbid derivation of rules that are not strong) and informative (should allow correct determination of support and confidence). A number of representations have been offered in the area of formal concept analysis and in the area of data mining. Rules that belong to such representations are usually built from specific sets of attributes such as closed sets, pseudo-closed sets and generators. s axioms, s method for determining s confidence, cover operator, and closure operator are used as inference mechanisms. During the talk, a number of representations of strong rules such as representative rules, Duquenne-Guigues basis, proper basis, Luxenburger basis, minimal non-redundant rules, generic basis, and informative basis will be surveyed. For each representation, we examine whether it is sound and informative. For the representations that are not sound, we discuss ways of turning them into sound ones.

Short biography

Marzena Kryszkiewicz is currently an assistant professor of Computer Science at Warsaw University of Technology. She received her M.Sc., Ph.D., and D.Sc. degrees in computer science from Warsaw University of Technology in 1988, 1995, and 2003, respectively. Her research interests include: data mining and knowledge discovery, concise knowledge representations, information systems, rough sets, approximate classification and concept learning, incompleteness, uncertainty. She has published around 60 papers. She was awarded for scientific achievements from Foundation for Polish Science in 1995, from the Rector of Warsaw University of Technology in 1996, 1999, and 2001, and from the Minister of MENiS in 2004. She has worked as a reviewer for IEEE TKDE, Int'l Journal of Intelligent Systems, Int'l Journal of Information Sciences, KAIS Journal, European Journal of Operational Research, and as a member of program committees of the ISMIS and RSCTC conferences and selected ICDM's and AM's workshops, as well as Data Mining track of SAC'05.