Abstract: Classical Rough set theory which analyzes complete information systems mainly using the indiscernibility relation.After the extension of Rough set model for existing，it is able to process complete information systems.There is a series of extension for the indiscernibility relation now，such as tolerance relation，non-symmetric similarity relation，limited non-similarity relation and limited tolerance relation based on object complete degree.Through the analysis of its advantage and disadvantage，it puts forward a new extension of Rough set which is limited non-symmetric similarity relation model based on object difference degree.The numerical example shows that the new model can acquire more sufficient information from incomplete information system，which is according with the actual situation.

Key words: Rough set, incomplete information system, object difference degree, limited non-symmetric similarity relation

摘要： 经典Rough集理论主要是利用了不可分辨关系对完备信息系统进行分析的。对现有粗糙集模型进行扩充后，才可以应用于不完备信息系统。容差关系、非对称相似关系、基于对象间完备度的限制容差关系、限制非对称相似关系等是现有的扩充模型。通过分析其优点和不足之处，提出一种新的模型——对象间差异度的限制非对称相似关系模型，在该模型中，给出的知识粒度更精确，更符合实际。实例结果也证明新模型可以从不完备信息系统获取更加精确的知识粒度。

关键词: Rough集, 不完备信息系统, 对象间差异度, 限制非对称相似关系