Research On Similarity Rough Set Theory Based On Pansystems

Rough Sets theory is a kind of theory to study the representation, learning and induction of incomplete, undetermined knowledge and data. Recently good result and application have been made in theory model and in algorithm and also in engineering. Up to now Rough Sets theory has been one of the most active research aspects in the field of information science. At the same time Rough Sets theory has been applied to medical, chemical, material, geographical, managing, financial and any other fields.In this paper we using conformation method, extend classical Rough Sets theory, and combine similarity rough sets theory with Pansystems theory. On the basis of searching for the new similarity rough sets model, we expend the theory model and improve the algorithm.At first, we give some basic concepts of Rough Sets theory and give some explanations of these concepts. From chapter 3 to chapter 5, these chapters are research of this thesis. Chapter 3:1. At first, we suggest attribute similarity information system based on similarity degree and similarity degree measure function, and consider attribute similarity information system is similarity system of some certain Knowledge representation system. Then basing on Pansystems theory, we suggest attribute similarity information system that similar with given Knowledge representation system constructing method based on pansystems body-shadow relation and similarity degree measure function.2. Basing on attribute similarity information system, we firstly suggest domination function, and then construct positive similarity, negative similarity, and pure similarity rough sets. Further we present the decomposing of attribute similarity information system based on domination function and decomposing method based on pansystems body-shadow relation.Chapter 4:1. We firstly suggest pansystems binary relation and pansystems quasitransitivity, then discuss and extend their property. We discuss the connection between transitivity and quasitransitivity, and further discuss the connection between transitivity in equivalence relation and quasitransitivity in similarity relation.2. Further based on quasitransitivity and b-s relation employ b-s core to take place similarity class as neighborhood operator, and consequently the upper and lower approximate operators are constructed. Compared with the traditional similarity rough set model, the approximate operators of this paper have finer accuracy of description to unknown knowledge.Chapter 5:At first, we suggest confirmation property abort similarity information table, then discuss the estimation method for suggest confirmation property abort similarity information table. We extend attribute reduction algorithm to similarity rough sets model based on pansystems binary relation.