提案誘導型知識獲得支援システム

概要

論文の詳細を見る
This paper describes a proposal guided knowledge acquisition support system which generates and proposes candidates of knowledge by induction. This system is an enhanced version of the interactive knowledge acquisition support system EPSILON/One. This system elicits basic information from an expert, makes hypothesis, and modifies target knowledge with the expert. Some of the major problems of using induction is that it requires a large amount of computing power and creates numerous hypotheses. In order to make hypotheses by induction more efficient, the forms of hypotheses must be limited. EPSILON/One has a knowledge representation, Expert Model, which consists of operations. The operations are limited to seven types which are derived from analyzing real diagnostic knowledge bases. The system realizes efficient inductive hypothesis generation by using this limited numbers of operation representations. This papers discusses inductive operation presumption algorithms and interactions between the expert and the system in this architecture.
社団法人人工知能学会の論文
1991-09-01