基于偏序关系的粗糙集规则提取方法

A method to extraction rules based on the partial relation of rough set

规则提取算法中通常先约简属性再约简属性值,但该算法当属性数量增多时,会增加约简的复杂性,从而影响规则提取的速度.针对此问题,本文提出了一种基于偏序关系的粗糙集规则提取方法.首先,在不同粒度的知识空间上建立偏序关系;然后,利用各知识空间中隐含的属性冗余度作为启发式信息,对冗余属性进行逐层约简;最后,在约简后的属性集上提取决策规则.实例表明,该方法降低了属性约简的复杂性,提高了规则提取的速度.

The traditional methods for rule extraction algorithm are attribute reduction and attribute value reduction. But the increasing of attributes number increased the complexity of reduction, and finally affected the speed of rule extraction. In this study, an extraction method based on partial relation of rough set is proposed. Firstly, a partial order relation is constructed in different knowledge spaces; and then, the redundant attributes are reduced according to redundancy which are implied in different knowledge spaces; finally, the decision rule is extracted from the attribute set of the reduction. Examples showed that the method can reduce the complexity of the attribute reduction and improve the speed of rule extractiofi.