Attribute Reduction Based on Rough Approximation Set in Algebra and Information Views
Qinghua Zhang, Jingjing Yang, Yao Longyang · IEEE Access · 2016
Rough set proposed by Pawlak in 1982 is an important tool to process uncertain information. As an extended model of rough set, an approximation set model of rough set was proposed and proved to be feasible to establish an approximation target set with existing knowledge base. However, there still is a lack of effective methods for knowledge acquisition based on the approximation set model. In this paper, related methods of attribute reduction based on approximation set model of rough set are discussed in algebraic view and information view, respectively. First, a distribution reduction method on the basic of discernibility matrix according to approximation set is proposed and discussed in algebraic view. Furthermore, an algorithm of attribute reduction based on conditional information entropy of approximation set model is presented in information view. Finally, many experimental results show that the proposed algorithm could acquire more effective knowledge from uncertain information system compared with other algorithms based on classical rough set theory.