A novel granular computing model based on three-way decision

Abstract

Granular computing and three-way decision are two very important methods in the field of knowledge discovery and data mining. In this paper, based on the idea of three-way decision, all attributes in the information table first are divided into three disjoint parts named indispensable attributes, rejected attributes and neutral attributes, respectively. According to the three parts of attributes, many basic and important information granules and granular structures can be induced from the information table. Then a novel granular computing model is proposed by the description operator. On the one hand, many mathematical properties related to the model proposed in this paper are systematically discussed. On the other hand, we make a preliminary and meaningful attempt to deal with network security by using this model. In addition, in order to apply the model more conveniently, two algorithms for computing description set, description degree, attribute reduction and reduction degree are developed. Finally, through numerical experiments, the validity of the algorithms and the related factors that affect the effectiveness of the algorithms are discussed in detail.