Study On Data Mining Techniques Based On Rough Sets

Along with the rapidly developing of computer and network techniques, the amount of information in all fields is explosively increasing. Therefore the requirement of analyzing the data to retrieve useful patterns concealed in those data is more and more urgent. In this situation, a new technology, data mining has appeared and flourished in a short time. As an effective method of data mining, rough set theory has also become a main method for data mining due to its unique advantage. Applying rough theory in data mining field can improve the analyzing and learning ability for incomplete data of large database, which has extensive applied prospect and applied value.The integration situation of data mining is studied thoroughly in this paper, as well as rough set's theory frame, basic concept and knowledge reduction, the core of rough set theory. The paper discusses the relationship between knowledge reduction and dependence, knowledge expression system and decision table. The inner meanings of discernible matrix are analyzed, as well as the relationship between discernible matrix and reduction. These are basic theories for the latter chapters about reduction algorithms. The core of rough set is knowledge reduction and discovery, which is supported by a series of algorithms, such as equivalent relation, upper/lower approximation solution, attribute importance estimation, core computation and attribute reduction, etc. The success of the applications of rough sets is determined by it's rigorous theory. The important effective applications of methods or algorithms are all supported by the related theories. Based on the idea, we put emphases on the research of theory and technique, and further more, give out effective algorithms. The main research results are as follows:A construction method of discernibility matrix for both consistent information systems and inconsistent information systems is presented. By which the core of a information system can be computed out correctly. Besides, a new form of expectation quality is proposed which provides criteria for efficiently determining the important attribute in attribute reduction for information systems.Based on the expression of cardinal numbers the vague set form of a rough set is studied, which reveals relationship of rough sets and vague sets on the expression of knowledge. Furthermore, the concept of rough vague sets is introduced. The new concept can be helpful for dealing with the knowledge expressed in the form of roughness and vagueness.A new concept of fuzzy discernibility matrix is proposed, and related theorem is proposed. By using fuzzy discernibility matrix,attribute reductions of fuzzy information systems can be realized. The results provide a way to study discernibility matrix for real value information systems.