概率粗糙集模型

Probabilistic Rough Set Models

1.引言 原始的RS模型(常称为Pawlak RS模型)[1,2]是建立在二元等价关系的基础上的,但由于实际问题的需要,PawlakRS模型的应用受到了限制,因此人们将二元等价关系推广成一般的二元关系,得到了一般关系下的RS模型[3],Yao还在文[4]中讨论了基于邻域算子的RS模型.另一方面Pawlak粗糙集模型是基于可利用信息的完全性的,因而忽视了可利用信息的不完全性和可能存在的统计和随机信息,这类模型对于不协调的决策表的规则提取往往显得无能为力.本文我们从概率论的观点出发来研究粗糙集理论,为研究不确定信息系统提供了新的粗糙集模型.

In this paper .rough set models based on a probabilistic measure are studied. The notions of lower and upper approximations based on a probabilistic approximation space are derived and properties of the approximation operators are discussed. The relationship between the probabilistic rough set model and Pawlak's rough set model is compared. At last,the notions of approximation accuracy,attribute dependence,and reduction of attribute related to this model are defined.