最大泛化规则生成

Generation of Maximally Generalized Rules

讨论了基于粗糙集理论的分类知识发现中最大泛化规则的生成。首先给出一个最大泛化规则的生成算法，提出采用基于信息论观点的Ｊ－ｍｅａｓｕｒｅ作为属性有效度量，用来在求最大泛化规则的过程中，启发式地选择试图删除的条件属性。最后通过实例说明了最大泛化规则生成算法的执行过程。

In this paper, the generation of maximally generalized rules in the course of classification knowledge discovery based on rough sets theory is discussed. Firstly, an algorithm is introduced. We propose that the information-based J-measure be used as another measure of attribute significance value. This measure is used for heuristically selecting the conditions to be removed in the process of extracting a set of maximally generalized rules. Finally, we present an example to illustrate the process of the algorithm.