On The Attrbute Reduction Approaches Of Covering Decision System

Rough set theory, proposed by Polish mathematician Z.Pawlak in 1982, is a new mathematical tool for dealing with vagueness and uncertainty. At present, rough set theory has been successfully applied into many fields, such as machine learning, pattern recognition, decision analysis, process control and data mining. Pawlak rough set theory was established on the basis of equivalence relation. To promote the application of rough set theory, people put forward many kinds of generalized rough set models. Covering rough set model has become an important generalization form.In this paper, we take covering rough set model as a tool. Then we research attribute reduction theory, methods and reduction algorithms in covering decision system. This paper contains mainly two aspects as follows.Firstly, we discuss the relevant properties of consistent and inconsistent covering decision systems. And we design two reduction algorithms of consistent covering system based on information entropy with the aid of covering approximate operators in [14], that compute all the reduction of the system, and verify the validity of given algorithms by an example.Secondly, we construct an attribute discernibility matrix based on generalized academic discerption idea in complete information system. And we present a method of attribute reduction of inconsistent covering decision system, and construct a corresponding reduction algorithm with the aid of generalized discerption idea and attribute discernibility matrix. Also, we compare it with other reduction algorithms at the base of discernibility matrixes to account for the superiority of the algorithm proposed in this paper.