The construction of characteristic matrixes of dynamic coverings using an incremental approach (1210.5706v4)

Abstract: The covering approximation space evolves in time due to the explosion of the information, and the characteristic matrixes of coverings viewed as an effective approach to approximating the concept should update with time for knowledge discovery. This paper further investigates the construction of characteristic matrixes without running the matrix acquisition algorithm repeatedly. First, we present two approaches to computing the characteristic matrixes of the covering with lower time complexity. Then, we investigate the construction of the characteristic matrixes of the dynamic covering using the incremental approach. We mainly address the characteristic matrix updating from three aspects: the variations of elements in the covering, the immigration and emigration of objects and the changes of attribute values. Afterwards, several illustrative examples are employed to show that the proposed approach can effectively compute the characteristic matrixes of the dynamic covering for approximations of concepts.