Formalizing case based inference using fuzzy rules
Résumé
Similarity based fuzzy rules are described as a basic tool for modeling and formalizing the inference part of the case based reasoning methodology within the framework of approximate reasoning. We discuss different types of rules for encoding the heuristic reasoning principle underlying case based problem solving, which leads to different approaches to case based inference. The use of modifiers in fuzzy rules is proposed for adapting basic similarity relations, and hence for expressing the case based reasoning hypothesis to be used in the inference process. Moreover, the idea of rating cases based on the quality of information they provide is touched upon.