Specialization models for a general fuzzy set covering framework

Set covering is a well-known rule induction paradigm for learning crisp classification rules. In this paper we present a generalization of the crisp set covering approach to fuzzy sets. We propose a general fuzzy set covering framework for finding good rules by a general-to-specific search as a further generalization of the set covering approach to learning. The paper illustrates four different specialization model algorithms as instantiations within this framework, and concludes with an empirical study to substantiate the usefulness of this new paradigm for the induction of fuzzy classification rules.