PREFERRING DIAGNOSES BY ABDUCTION

Much research has been devoted to diagnosis, where two main approaches have been pointed out: the empirical-association-based diagnostic app roach and the model-based diagnostic one. Both approaches can be chara cterized by the kind of knowledge that has to be specified and the dia gnostic method that has to be used. However, it seems particularly dif ficult in real-world applications to obtain a complete description of the faulty (dually, correct) behavior of a system. This incompleteness of description is the reason why deductive reasoning alone is general ly insufficient to point out the actual diagnosis. Deduction only allo ws to generate some possible partial diagnoses. The latter must be sel ected and completed to get closer to the actual diagnosis. Both select ion and completion require hypothetical reasoning and can be character ized by some preference criteria. The authors' contribution is twofold . By a comparative study of the different diagnostic methods, light is first shed on the preference criteria they implicitly use. A new diag nostic method based on deduction and abduction is then proposed, which is sufficiently flexible to deal with multiple knowledge representati ons.