At present, our society is becoming more and more informational and intelligent. Because of the data’s explosive growth and the phenomenon that the data itself will lose information, in incomplete information systems, data mining has become an important research area. And attribute reduction in incomplete information systems is one of the core content. It combines many of the theories and techniques, such as Rough Sets, Intuitionistic Fuzzy Sets, Database and so on.In the paper, we mainly research attribute reduction in incomplete information system, and discuss with two cases:1) Incomplete information system which value of attribute is exact. We expand the rough set by introducing tolerance relation, defining the discernibility matrix under the tolerance relation, reducing the attribute set by discernibility method. At last, we select data sets from UCI(Data mining Database in University of California Irvine) as the testing data sets, using Java to implement the attribute reduction algorithm, and the validity of the algorithm is verified.2) Incomplete information system which value of attribute is intuitionistic fuzzy sets. We combine limited tolerance relation with dominance relation through defining dominance relation of intuitionistic fuzzy sets, proposed the concept of generalized dominance relation. And we define the discernibility matrix based on generalized dominance relation of the incomplete intuitionistic fuzzy information system. Finally, we obtain the attribute reduction and core attributes by example. |