Approximate reasoning in a system combining prototypical knowledge with case-based reasoning

In this paper we describe a system which combines prototypical knowledge with case-based reasoning. In particular, prototypical knowledge is used to solve typical cases while case-based reasoning is mainly used for dealing with cases which correspond to 'exceptions' with respect to prototypical knowledge. The presence of a case-based component allows us also to deal with problems at the boundary of the knowledge represented in the prototypical part. The need for approximate reasoning mechanisms is apparent in both the prototypical and the case-based components. In fact, some forms of fuzzy matching are needed both for retrieving casesand for discovering the similarities between them. Particular attention has been paid to the design of a uniform approach for dealing with uncertain knowledge in the two components of the system.