An expert system for automobile car diagnosis

The authors outline the development of an AI-based expert system for trouble-shooting a car. The system is built on the basis of a hierarchical classification of concepts. Each node in the hierarchical structure represents a malfunction hypothesis. The knowledge is represented by forward-chaining production rules. The organization of rules emulates an expert's approach in problem solving. A control strategy termed 'propose-refine' is used that proposes a set of possible malfunctions and refines them depending on the prevailing situation. Incorporation of judgemental knowledge and uncertainty management are discussed. The expertise embedded in the system is based on the judgemental knowledge obtained from the experts in the field of automobile trouble-shooting.<
>