Knowledge Reduction Based On Extended Model In Set-valued Decision Information System

Rough set is a mathematic tool to describe the imprecise and uncertain knowledge, it is proposed by Pawlak in 1982 initially. This theory and method is an effective tool to deal with the inconsistent, inaccurate, incomplete data and so on. Knowledge reduction is one of the core problems in rough set theory. Based on the different binary relation in the universe, the thesis aims to study the knowledge reduction of extended model in set-valued decision information system.。In the set-valued decision information system based on neighborhood relation, a new dis-cernibility matrix is defined by using positive domain. By means of which, the text summarizes the detailed operation methods for knowledge reduction and the text illustrates the validity of this method through the concrete example finally.· In the set-valued decision information system, the variable precision compatibility relation is also introduced. The thesis defines variable precision compatibility matrix and the distance of variable precision compatibility matrix. A knowledge reduction algorithm for set-valued decision information system is formulated by using the distance of variable precision compatibility matrix as heuristic information.· The concepts of implication relationship and fuzzy conditional entropy for set-valued decision information system which based on fuzzy tolerance relation are introduced and the specific operation methods of knowledge reduction are proposed. Moreover, the examples are given to validate the effectiveness of the proposed method.