决策系统基于特定决策类的上近似约简

The upper approximation reduction of decision system based on specific decision classes

决策系统针对特定概念的约简相对于整体约简可以获得更简洁的决策规则.借助粗糙上近似算子,提出决策系统针对特定决策类的上近似约简概念,针对这种约简标准给出了需要区分对象的条件,借助区分函数给出了约简计算方法;刻画了保持所有决策类上近似不变的分配约简、基于特定对象的分配约简以及针对特定决策类的上近似约简之间的相互关系.可以通过分配约简构造基于特定对象的分配约简以及针对特定决策类的上近似约简,也可以通过针对特定决策类的上近似约简构造基于特定对象的分配约简.

In decision system,with respect to global attribute reduction,the reduction based on specific decision class can reveal more compact decision rules.Based on upper approximation operators,the notion of upper approximation attribute reduction of specific decision class is proposed.The discernibility conditions between objects with respect to this reduction criteria are presented and the reduction approach is proposed by using discernibility functions.The relationships among assignment reduction,upper approximation attribute reduction of specific decision class and upper approximation attribute reduction of specific object are surveyed.It is possible to derive an upper approximation attribute reduction of specific decision class and an upper approximation attribute reduction of specific object from a assignment reduction,as well as to derive an upper approximation attribute reduction of specific object from an upper approximation attribute reduction of specific decision class.