OPTIMUM PROBLEMS IN CONSTRAINT SATISFACTION

Abstract

With the advanced spread of Expert Systems (ES) more and more emphasis has been put on efficient knowledge representation. Constraints, a fast developing field in Artifi- cial Intelligence (AI), try to cope with the new requirements imposed on the knowledge representation tools in practical applications. Efficient algorithms operating on constraint networks have been developed [4, 7, 10], and recently some constraint systems have been constructed [3], which are capable of solving relatively complicated problems stated in a declarative way. However, these systems put emphasis mainly on finding a feasible solu- tion, they do not rank the different solutions. In some cases this will not suffice. When optimum criteria is meaningful to define, an optimal solution can be of importance. The present paper gives a brief overview of constraints, emphasizes the importance of searching for an optimal solution; and gives an application area where the optimal solution would highly improve the performance of the embedding system.