Research And The Application Of Granular Computing

Facing a large amount of information, people hope to find useful knowledge from them. The important task of data mining is to find knowledge, model and rule from a large amount of information, so data mining especially association rule mining becomes more and more important. When facing and dealing with the problem with a large amount of information, because of the limitation of human being's cognizance ability, people used to divide the information into some simple blocks, or granules in order to do analysis and settling according to their feature and function. The granulation of information makes the use of granule to be necessary when facing and dealing with the problem, so the application of Granular Computing in data mining, especially association rule mining is of important and practical sense. The aim of data mining is to find knowledge, while reasoning when using knowledge is another important application. Because of the imprecise,uncertain,incomplete and fuzzy information of real world, the application of uncertain reasoning becomes more important. Granular Computing, which is according with the impersonal law of human's problem solving and covers the theory of Rough Set and Fuzzy Set, has its own advantage and potential ability when dealing with the uncertain knowledge and the uncertain reasoning.Based on the analysis of the development and the research status quo of Granular Computing, this paper introduces the essential knowledge of Rough Set and Granular Computing. The features and main task of this paper: (1)From the view of information granule, this paper newly recognize the item,item sets and the support degree of item sets of association rule mining based on he a sequence pair of concept of Granular Computing, and also proposes an algorithm called GLIG of association rule mining based on granule and binary operation. This algorithm scans the data base only once and the speed of binary operation is faster than the speed of comparison of item sets and transactions.(2)Combined the theory of Rough Set and credibility, this paper proposes a method of reasoning under the condition of non-noise based on Granular Computing; when the information of decision table is incomplete or disturbed , a approximate reasoning method is proposed under the condition of noise; at last a description of an algorithm of uncertain reasoning based on Granular Computing is...