Constraint satisfaction using soft quantifiers

Ronald R. Yager Machine Intelligence Institute, Iona College New Rochelle, NY 10801 [email protected] (212)249-2047 ABSTRACT Fuzzy sets and other methods have been used to model a softening of constraints in CP problems. Here we suggest an approach to the softening of the CP problem at the meta level, the process used to aggregate the satisfactions to the individual constraints. We suggest the use of soft quantifiers such as "most" to guide the process of aggregating the satisfactions to the individual constraints. Use is made of the ability to represent these soft quantifiers by fuzzy sets and the ability to implement their authorized aggregation by the OWA operator.