Learning Models Based on Formal Concept

摘要

From the classic view in which a concept consists of the set of extents and the set of intents, a concept learning system extended from a formal context is introduced and two concepts such as an under concept and an over concept are defined. Any pair of subsets from extents and intents in this concept learning system can be changed to an under or an over concept. Further it can be changed to a concept by learning from the set of extents or from the set of intents. It is proved that the concept learned in this framework is an optimal concept. This process of learning a concept describes the recognizing ability from unclear to clear.