The Reserach On Some Problems Of Ordered Decision Imformation Systems

Rule acquistion is one of the key issues in rough set theory. Currently, some methods of rule extraction do reduction and then extract rules. These methods will lead to the loss of information and cause decision-making errors in ordered decision information systems. The dissertation discusses how to extract rules before reduction and do reasonable classification without rule extraction, so that the information can be fully used and the accuracy of classification is improved. The main contents of the dissertation are as follows.Firstly, a kind of method of inducting all minimum rules is proposed on order information system, which takes the matrix for the tools. The proposed method can effectively improve the efficiency of fetching rule, because of avoiding the generation of no-minimum rules.Secondly, the concept of negative domain matrix is proposed base on the characteristics of extracting rules. Through the negative domain matrix, the rules dominated by objects are found in lower approximation. meanwhile, judge whether the rules are also covered by other objects. Thus the process of extracting rules can be avoided, and then objects covered by rules can be directly found.Finally, in order to obtain the reduction of upper approximation, matrix is used as a tool. And then uncertainty decision making rules are obtained. The time complexity of the method is O(|A|3|U|2+|A|2|U|3).