Learning fuzzy concept hierarchy and measurement with node labeling

A concept hierarchy is a kind of general form of knowledge representation. Most of the previous researches on describing the concept hierarchy use tree-like crisp taxonomy. However, concept description is generally vague for human knowledge; crisp concept description usually cannot represent human knowledge actually and effectively. In this paper, the fuzzy characteristics of human knowledge are studied and employed to represent concepts and hierarchical relationships among the concepts. An agglomerative clustering scheme is proposed to learn hierarchical fuzzy concepts from databases. Further, a novel measurement approach is developed for evaluating the effectiveness of the generated fuzzy concept hierarchy. The experimental results show that the proposed method demonstrates the capability of accurate conceptualization in comparison with previous researches.