中文摘要：

基于容差关系和先验概率容差关系的粗集模型是粗糙集理论的重要扩充，然而其却均有自己的局限性。结合上述两种模型的优点．提出了一种新约束容差关系，新约束容差关系的粗集模型是建立在对不完备信息系统属性值统计数据基础上的．可有效提高分类精度和同类对象问的相似程度。实例分析表明，新的模型更具有实用性和操作性。

英文摘要：

The prior probability tolerance relation and tolerance relation are the significant expansion of the rough set. However, they do have their own limitations. In this paper,we combine the advantages of these two models, and propose a new constraint tolerance relations.The new relation is based on the incomplete information system property value based on statistical data. The new relation can effectively improve the classification accuracy and the same degree of similarity between objects. The case shows that the new model is more practical and operational.