中文摘要：

针对粗糙集理论在知识约简中的实际需要，提出了建立在一般二元关系基础上的广义粗糙集知识约简方法。首先证明了广义粗糙集是经典粗糙集的一般性推广，而经典粗糙集是广义粗糙集的特例；然后以一般二元关系为分类基础，给出一般关系决策系统中的知识约简判定定理和辨识矩阵；最后根据实例提取最小的属性集，验证了该方法的实用性。该方法摆脱了二元等价关系对经典粗糙集的困扰，既保证了粗糙集理论在知识发现研究中的理论优势，又拓展了粗糙集理论在实际应用中的适用范围，具有较强的实用性。

英文摘要：

To meet the practical demand of the rough set theory in knowledge reduction, the paper establishes a method of knowledge reduction based on generalized rough sets. Firstly, the paper proves that an important value of generalized rough sets is based on arbitrary binary relations on a universal set, which may extend applications of the classical rough set theory, and then presents the decision theorem of knowledge reduction and discernible matrix based on some general binary relations. Finally, the validity of the method is verified by the application of a practical knowledge system, which can accurately abstract a minimal attribute set. The major contributions of this paper are the method of knowledge reduction based on generalized rough sets may overcome the shortage of the classical rough set theory, and extend many practical applications in various areas.