An efficient method of solving constraint satisfaction problems in IR spectrum interpreting

This paper presents a simple constraint-based reasoning method used in an expert system which we have developed for infra-red (IR) spectrum interpretation where getting an initial solution is much more difficult than refining it. The most difficult constraints concerning that different variables cannot take on the same value (called constraint α) are considered after the other constraints (called constraint β) have been considered. So an efficient pattern-driven algorithm can be used to generate an initial solution which satisfies constraint β. Then an overlap-reduce heuristic is applied to minimize the redundancy in the initial solution until all conflicts among the variables are eliminated. We have developed an expert system for IR spectrum interpreting by means of this reasoning method. This paper also briefly describes the expert system