The Study And Application Of Core And Attribute Reduction Algorithm On The Basis Of Rough Set

The rough set theory is a tool of statistical analysis on inaccurate, inconsistent and incomplete information by directly analyzing and deducing the data without any prior information except the data set to find underlying knowledge and potential rules. It's been widely used in the fields of machine learning, knowledge discovery in database, decision analysis, pattern recognition, medical diagnosis and expert system, etc.The attribute reduction and core are two important concepts of the rough set theory. The former deletes the redundant attributes with classification capability remaining unchanged. Core is the intersection of all reductions. The attribute core of a decision table is often the starting point and key to the reduction process of decision information. The fast algorithm of core and attribute reduction is one of the most important contents of the rough set theory research.This dissertation studies the algorithm of core and attribute reduction. The main work includes:First of all, it reviews typical algorithm for counting core of decision table, analyzes the complexity of time & space and puts forward a new algorithm of higher efficiency through exemplification to verify its validity.Secondly, the attribute reduction algorithm based on discernibility matrix and its limitations are explored to present a new attribute reduction algorithm based on dependence degree, which does not need core calculating, saving time & space and simplifying the sloving process as well.Finally, the data characteristics of Chinese traditional medicine are analyzed, introducing methods for pretreating chinese traditional medicine data based on rough set theory are and an application of Chinese traditional medical diagnosis is given.