知识过滤和它的依赖推理发现

知识具有颗粒，知识具有粒度；颗粒，粒度是知识的两个基本特征；知识的颗粒特征，知识的粒度特征与知识的属性保持着紧密联系，属性的变化引起知识颗粒特征，知识粒度特征的变化．本文将物理学中物质具有颗粒的概念和过渡的概念引入到粗集（rough set）理论与应用的研究中，提出了知识过滤的概念，给出知识过滤的数值度量，知识筛子原理．利用知识过滤的概念，给出知识过滤在依赖推理中的推理结构，依赖推理的数值关系和在知识发现中的应用．知识过滤与依赖推理相互交叉，互补共享，是知识发现研究的一个新方向．

Knowledge Filter and Its Dependent Reasoning Discovery

Knowledge has granulation and granular degree; both granulation and granular degree are the basic characteristic of knowledge;they keep close relation with the attributes of knowledge. The variation of attributes leads to the variation of knowledge tranular degree. This paper introduces the conception of granular and filter into the research of the theoty and applications of rough sets,gives numerical degree of knowledge filter and the principle of knowledge screen. By the concept of knowledge filter,it gives reasoning structure of knowledge in dependent reasoning, numerical relation of dependent reasoning and its application in knowledge discovery. It is a novel approach for knowledge discovery research to combine knowledge filter and dependent reasoning.