PROBLEM-SOLVING METHODS FOR DIAGNOSIS AND THEIR ROLE IN KNOWLEDGE ACQUISITION

Diagnosis is a popular topic in Artificial Intelligence. Recently, we presented a knowledge-level analysis of diagnosis covering a large scope of different approaches and systems. Such analysis gives us a better understanding of diagnosis in terms of the goals it comprises and the ways to achieve these goals. In the knowledge acquisition community, libraries with reusable knowledge components are becoming increasingly important. Such libraries facilitate the knowledge acquisition process in the sense that a problem-solving strategy can be constructed from ready made parts rather than to build it up from scratch. The result of the knowledge-level analysis of diagnosis can serve as such a library. The aim of this paper is, first, to present the collection of problem-solving methods and, second, to show some experiments that explore the usefulness of the library in knowledge acquisition. In order to do the experiments, it is necessary to associate problem-solving methods with suitability criteria that reflect the conditions under which problem-solving methods are applicable.