An Relationship Of Degree Based Heuristic Algorithm For Attribute Reduction And Its Application In Group Reasoning

Rough sets theory, as a new kind mathematics theory which dealing with uncertain and incomplete knowledge, becomes an important branch of data analysis tools with fuzzy theory, neutral network and evidence theory. The feature of this theory is that it can give the most possible solution when the information is incomplete. The research on rough sets in our country starts from 1994,in recent years, this theory are becoming a promising research area and has been widely used in many fields.Attribute reduction is an important concept in rough sets data analysis. To find a high performance reduction algorithm is a focus point in the research on rough sets, but there doesn't exist an efficient method so far. It has been proved the computation of minimal reduction and full reduction both is NP-hard problem, in artificial intelligence the common way is to employ heuristic knowledge to reduce. In this thesis, we first analyze several classical reduction algorithms, and then propose a new heuristic algorithm based on attribute relationship of degree, compare it with other algorithm from several aspects. An example is given to show its applicant.Group reasoning based on rough sets-combining uncertain information deployment method with group decision theory and method, is a new kind research area. The main idea is to use rough sets theory to reduce the rule, which generated by experts, and then integrate these group rules to improve the ability of group reasoning. In order to illustrate the reasoning process, the thesis gives an example at last.