Sorting And Granularity Ruduction Based On Set-Valued Information Syetem

The classical theory of rough set is processing incomplete information system, can analyze data and process without any prior information. Since there is error when data acquisition, make the information is incomplete. So the information system is incomplete. Set-valued information system is a method that could deal with incomplete information system. The method is to put the missing attribute value by the rest object value of set. Form the following aspects to research in this paper.Firstly, new rough set model of set-valued information system is presented, called majority dominance relation of rough set model of set-valued information system. It not only enhances the ability of data processing, also overcomes the lack of dominance class too strict in dominance relation set-valued information system. It also is the development of incomplete information system of rough set. Properties of a majority dominance relation of set-valued information system of rough set are investigated.Secondly, based on the disadvantages of current sorting method, new formula for sorting method is put forward, its properties are investigated and its superiority is proved. Then for the attribute reduction, discernibility matrix can be obtaineddistinguish function. Reduction of condition attribute set is the distinguish function of all conjunction in the minimal disjunctive normal form type.Thirdly, new set-valued information system is proposed on the multi-granularity rough set. The dominance relation of existing model is too strict. So majority dominance relation is relaxed the number of attributes in dominance relation and its properties are discuss.Fourthly, granularity reduction method ofset-valued information system of multi-granulation rough set based on majority dominance relation is presented. In multi-granulation space, because of the large amount of data, and there are redundant granularity, so granularity reduction is complicated and is an important research content. Then the heuristic algorithm of granularity significance is introduced in this paper. The feasibility of this method through examples.