Exploring the Explanatory Capabilities of Intelligent System Technologies
Shirley Diane Gregor, Xinghuo Yu · Studies in fuzziness and soft computing · 2002
Explanatory capabilities in expert systems can explain to their human users both the knowledge they contain and the reasoning processes they go through. Explanations in such systems, when suitably designed, have been shown to improve performance and learning and result in more positive user perceptions of a system. Justification-type explanations have been found to be particularly effective in leading to positive outcomes. In this paper we explore the use of explanatory capabilities in new-paradigm systems including neural nets, evolutionary computing and fuzzy logic. Explanations in neural net and evolutionary systems are relatively low-level at present, relying mainly on rule explication. There has been surprisingly little research on explanations from fuzzy systems. These systems, being similar to expert systems apart from the extended capacity to represent vague expert knowledge, can in principle provide explanations and justifications in a manner similar to conventional expert systems. Suggestions for providing explanations from the new-paradigm systems are given. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.