Computing Soft Constrains by Hierarchical Constraint logic Programming

概要

論文の詳細を見る
We have already formalized soft constraints based on interpretation ordering which is a generalization of circumscription. However, this formalization is written in a second-order formula and therefore is not computable in general. To make it computable, we have to introduce some restriction. In this paper, we propose a semantic restriction. By semantic restriction, we mean that we fix the considered domain so that interpretations of domain-dependent relations are fixed, and soft and hard constraints contain only domain-dependent relations. If we accept this restriction, the soft constraints can be expressed in a first-order formula. Moreover, there is already a proposed mechanism suitable for computing such restricted soft constraints in the literature, that is, hierarchical constraint logic _programming languages (HCLP languages). Firstly, we identify a solution to constraint hierarchy defined by HCLP languages with the most preferable solution for semantically-restricted soft constraints. Then, we provide an algorithm for calculating all the most preferable solutions for soft constraints without any redundant calls a constraint solver. Then, we show examples of computing soft constraints by using our HCLP language named CHAL (Contraintes Hierarchiques avec Logique).
一般社団法人情報処理学会の論文
1993-07-15