Clustering Association Rules with Fuzzy Concepts

Abstract

Association rules constitute a widely accepted technique to identify frequent patterns inside huge volumes of data. Practitioners prefer the straightforward interpretability of rules, however, depending on the nature of the underlying data the number of induced rules can be intractable large. Even reasonably sized result sets may contain a large amount of rules that are uninteresting to the user because they are too general, are already known or do not match other user-related intuitive criteria. We allow the user to model his conception of interestingness by means of linguistic expressions on rule evaluation measures and compound propositions of higher order (i.e., temporal changes of rule properties). Multiple such linguistic concepts can be considered a set of fuzzy patterns (Fuzzy Sets and Systems 28(3):313–331, 1988) and allow for the partition of the initial rule set into fuzzy fragments that contain rules of similar membership to a user’s concept (Höppner et al., Fuzzy Clustering, Wiley, Chichester, 1999; Computational Statistics and Data Analysis 51(1):192–214, 2006; Advances in Fuzzy Clustering and Its Applications, chap. 1, pp. 3–30, Wiley, New York, 2007). With appropriate visualization methods that extent previous rule set visualizations (Foundations of Fuzzy Logic and Soft Computing, Lecture Notes in Computer Science, vol. 4529, pp. 295–303, Springer, Berlin, 2007) we allow the user to instantly assess the matching of his concepts against the rule set.