Empirical Plausible Reasoning by Multiple-Valued Logic

A method to approximate the heuristic reasoning of an expert in judging the behaviour of a system is proposed. The method is based on the use of label functions, mapping the observed value of one attribute into a local judgment, and of multiple-valued logic trees, mapping a set of local judgements into a global one. The method is introduced within the fuzzy set approach, so that the involved approximation can be discussed.