Multi-Value Attribute Concept Lattice Reduction Based on Granular Computing

Abstract

Concept lattice essentially describes the relationship between objects and attributes. The reduction of multi-value attribute concept lattice is a hot topic in the fields of information retrieval, knowledge discovery and data mining etc., while the granular computing emphasizes observing and analyzing the same problem from different granular worlds. It makes the complex problems around us be mapped to an easy to handle and more simple theory of calculation. The paper gives definitions of the concept granule and the compatible concept granular set by application of information granular and the granular of layered theory, and provides an algorithm to compute concept granular set through calculation of the compatible relationship. The paper further constructs the concept granule lattice, and then deletes the attribute of smaller contribution to concept granule. Through the comparison of the concept granule lattices, the multi-value attribute reduction could be achieved and the core attribute set in the formal context could be obtained. Instances could demonstrate the high efficiency and accuracy of this algorithm that is easier to realize through programming. Through the resolution of attribute, the calculation complexity could be reduced and the efficiency of calculation could be improved.