Abstract: The study of knowledge acquisition for the real value decision system is one of the main directions of the granular computing research. For giving an efficient knowledge acquisition method, based on the neighborhood of rough set theory and according to the characteristics of the real value, the reduction definition based on discriminate degree is proposed in the paper. Moreover, the heuristic algorithm of attribute reduction based on discriminate degree is given for compatible real decision system, and the properties of algorithm are discussed. The effectiveness of the algorithm are showed by an example.

Key words: rough set theory, real system, attribute reduction

摘要： 研究实值决策系统中的知识获取是粒计算研究的主要方向之一。为给出一种高效的知识获取方法，基于邻域粗糙集的原理，针对实值特点，在相容的实值决策系统中，给出判别度描述下的约简定义。进一步提出一种基于判别度的属性约简的启发式算法，同时对算法的性质进行了分析，通过实例验证了算法的有效性。

关键词: 粗糙集理论, 实值系统, 属性约简