The Research On Attribute Reduction Algorithms Based On Rough Set Theory And Its Application Of TCM Syndrome Diagnosis

The attribute reduction based on rough set usually reflects the essence of information table and it is the core content of rough set theory. Normally, the reduction of the information system is not only one and people would like to get a minimum. Therefore, the main research direction of attribute reduction focus on the more effective attribute reduction algorithm,the optimal feature subset,reduce the time complexity algorithm and improve the accuracy of the optimal. Based on the research and analysis of existing algorithms, a new attribute reduction method which applied to complete information system and incomplete information system has been proposed. At the same time, the application of attribute reduction in the TCM syndrome has been studied. This paper work includes the following aspects:Firstly, the basic knowledge of rough set theory has been introduced in this paper. It included information system, the lower approximation, the attribute reduction and nuclear, attribute dependence and important degree and similar relation in the incomplete information system.Secondly, the attribute reduction algorithm has been analyzed and studied. To find all the advantages and disadvantages of existing algorithms, attribute reduction algorithm of complete information system and incomplete information system has been mainly summarized.Thirdly, with the analysis of advantage and deficiency of the existing attribute reduction algorithm, a new attribute reduction method based on attribute action set difference degree has been proposed. This method is also applicable to complete information system and incomplete information system. In the algorithm, the similar relation and equivalence relation has been combined and the attribute action set difference degree has been used as heuristic factor to find the reduction set. And also, the approximation classified quality has been applied to evaluate the quality of reduction set. Finally, test and experiment results show that the proposed method is feasible especially the method can find out reduction subsets rapidly and accurately on data set with a core. In addition, the attribute reduction in the application was also studied. The TCM diagnosis of data pretreatment process has been expounded in this paper and with the algorithm proposed in the paper, the attribute symptom has been found which are consistent with expert's diagnosis. And also the research significance of application of attribute reduction in the TCM diagnosis has been found and this provides theory basis for syndromes research.Finally, the paper research was summarized, and the future work prospected.