A multi-objective decision-making approach to syllogistic reasoning with consistency maintenance

We discuss a syllogistic reasoning problem of the type Q/sub 1/(A/spl rarr/B), Q/sub 2/(B/spl rarr/C)/spl rArr/Q/sub 3/(A/spl rarr/C) where Q/sub i/, i=1, 2, 3, are imprecise quantifiers such as "most" and "usually", and A, B, and C are fuzzy sets such as "big cars". We propose a new approach that integrates reasoning with evaluation of knowledge based on the framework of multi-objective decision-making. The reasoning process is formulated as a maximization of the compatibility degree of knowledge with observed data. >