Research On Concept Drifting Detection Based On Two Kinds Of Rough Set Models

The Rough Set theory is an effective mathematical tool which can deal with the imprecise,incomplete and imperfect information and knowledge without the subjective experience.But it can’t deal with the successional data very well.The Fuzzy Set can deal with it better with the subjective experience.The Fuzzy-rough Set holds their advantages and is a good tool which can deal with the successional fuzzy data.But it fails to deal with the massive incremental data stream.Based on the traditional Fuzzy Rough Set theory,the F-Fuzzy Rough Set theory and its reduction provide an effective method of attribute reduction in the incremental and dynamic decision tables.As the part of the Rough Set theory,the Decision Rough Set theory consists of the Three-way Decision and Concept Fault Tolerance provided by the Decision Rough Set,which can simulate effectively the decision-making process.It extends the traditional binary decision-making semantic,consisting of the positive region and the negative one,to three decision-making semantics,consisting of the positive region,the negative one and the boundary one,and regarding the decision based on the boundary region as a feasible kind of decision,which accords with the methods with which people deal with the problem of Decision-making.The concept drifting is the phenomenon that the data in data flow change as time goes by.The classic algorithm of concept drifting detection is concerned with the discrete data and the data flow which is different from the one that is massive incremental ones produced by the real life.As for this problem,the F-Fuzzy Rough Set theory can solve it.The classic researches on the decision-making focus on the Binary Semantic,i.e.,the positive region and the negative one,which can’t describe accurately the imprecise of the real concept and the fault tolerance of the real classification.The three decision-making rough set is the one of the fundamental principles of Decision Rough Set.It added the decision-making based on the boundary region to the classic Binary semantic,namely delayed decision which is based on the decision-making of the boundary region,i.e.,a feasible kind of decision-making.The delay decision accords with the habit of our dealing with the everyday problem of decision-making and is extensively representative and significant.At first,the article combines the advantage of F-Fuzzy rough set with concept drifting detection in order to detect the fuzzy concept drifting in fuzzy data flow.Then the article adds the fault tolerance to the concept drifting detection by combining with the advantage of the three decision-making semantics.The researches in this paper conclude:1.The concept drifting detection based on the F-Fuzzy Rough Set.The classic rough set model can’t deal with the successional data and detect the concept drifting in the data flow directly.Based on the F-Fuzzy rough set,this paper presents a method which can deal with the massive successional data directly.It is efficient to deal with the incremental and fuzzy data flow which is common in our life.It provides the new way of detecting fuzzy concept drifting.2.The concept drifting detection based on three-way decision-making.The classic binary semantics,consisting of the positive region and the negative one,can’t describe accurately the decision-making model of dealing with the problems in our life.As for it,this paper combines the advantage of three decision-making semantics with the concept drifting detection,which extends the classic binary semantic to the Three decision-making semantics,regarding the decision-making based on the boundary region as a feasible kind of decision.It can describe accurately the decision-making model in our everyday life.Therefore,the proposed works are extensively representative and significant.The contributions of this paper are as follows:1.This paper applies the F-Fuzzy rough set theory to the researches on the classic concept drifting detection and extends the data field of concept drifting detection algorithm to the massive incremental data.2.This paper applies the three-way decision-making to the researches on the classic concept drifting detection,which accords with the decision-making model in our everyday life.The works are extensively representative and significant.