Abstract: Considering the fuzzy information system, based on incompact approximation operator defined by R. Jensen, a strict approximation operator is constructed that can ensure monotone increasing of lower approximation with the attributes number of information system. The conceptions of relative reduction about fuzzy information system are given, and a heuristic knowledge reduction algorithm is proposed by dependency degree. The effectiveness of the proposed algorithm is illustrated by applying it to the knowledge reduction of an information system for threat level assessment.

Key words: fuzzy information system, rough set, fuzzy rough set, relative reduction

摘要： 针对模糊信息系统，通过分析R.Jensen所定义近似算子的松散性，构造了一种严格状态下的近似算子，该算子可以保证下近似随信息系统中属性个数的增加而单调递增；在此基础上，对模糊信息系统的相对约简概念进行了定义，应用所定义依赖度提出了一种模糊信息系统的启发式知识约简算法；将该方法应用于目标威胁等级评估信息系统的知识约简，计算结果验证了该方法的有效性。

关键词: 模糊信息系统, 粗糙集, 模糊粗糙集, 相对约简