Information aggregating networks based on extended Sugenos fuzzy integral

Abstract

Sugenos fuzzy integral is a functional to aggregate partial evaluations for an object in consideration of importance degrees of evaluation items. This paper presents the issues related to Sugenos fuzzy integral for information aggregation. For the identification of importance degrees of evaluation items with the properties of fuzzy measures, we suggest to use a genetic algorithm based method. To improve the behavior of the fuzzy integral by avoiding excessive emphasis of pessimistic aspects, we introduce compensatory operators into the fuzzy integral. On the other hand, to tune the parameters for the used compensatory operators and to perform the fuzzy integral in parallel computation, we propose a network model.