A Novel Representation of Concept Hierarchy Based on Quotient Space Model

Concept hierarchies are important in many generalized data mining applications, such as multiple-level fuzzy association rule mining. Usually concept hierarchies are given by domain experts. However, it is extremely difficult and time-consuming for human experts to discover concepts and construct concept hierarchies from the domain. In literature, several representations of concept hierarchy are possible, for example tree, lattice, table, linked list, arbitrary graph etc. In this paper, we apply quotient space model to representing concept hierarchies. In contrast to others, the representation model is much more extensible and compatible. The results indicate that this technique can improve the efficiency of performing the generalization and specialization operation in concept hierarchies.