On some fuzzy extensions of association rules

Abstract

This paper discusses the semantics of two fuzzy extensions of the classical concept of an association rule. Both extensions are based on the aggregation of sufficiently close data into fuzzy sets, by means of user-defined fuzzy partitions of the domains, thus leading to fuzzy generalized association rules. The first approach is based on fuzzy cardinalities whereas the second one relies on gradual rules. The issue related to the evaluation of the validity of such rules is discussed and the principles of two discovery algorithms are outlined.