Optimizing Expert Systems

Abstract

Until now, expert systems only have been little successful in planning and construction domains. Especially, when optimal solutions are required they fail. On the other side, algorithmic methods proposed in the field of Operations Research only work on single problems, or if designed for a wider range of problems, are highly inefficient. In this paper we show that some of this problems can be avoided by combining the techniques of expert systems and Operations Research. The system proposed in this paper generates an optimal solution to a construction problem in two phases. In the first stage the given concrete problem will be mapped to a mathematical problem; in the second stage an algorithm is applied to this problem and the result is transformed back by the expert system and presented to the user.