A fuzzy-based approach to numeric constraint networks

Abstract

Current numeric constraint propagation systems accepting exact numeric values or intervals have a limitation on representing the preference among values. To represent and handle the preference on the values, the notion of fuzzy numeric constraint networks is introduced. After defining the notion of fuzzy consistency, we show that fuzzy consistency can be represented by an extension of interval consistency. Using this relation, we propose a propagation algorithm for solving fuzzy numeric constraint networks.<>