区间值信息系统上决策规则的获取与优化

Decision Rule'S Acquisition and Optimization in Interval-Valued Information System

以区间值信息系统上的变精度相容关系所确定的极大变精度相容类作为的基本知识,在相似水平不变的情形下,提出了极大变精度相容类的属性描述、相对约简、决策规则及相对最优决策规则等概念.最后,针对极大变精度相容类,定义了一种基于区分矩阵的区分函数,并通过计算区分函数的析取范式得到获取区间值信息系统相对最优决策规则的具体操作方法.

This paper takes maximal variable precision tolerance classes based on variable precision tolerance relation as basic knowledge in interval-valued information system, and proposes the concepts of descriptions of attribute, relative reduction, decision rule and rel- ative optimal decision rule when similarity level is invariable. At last, we define a kind of discernibility function based on the discernibility matrix for the maximal variable precision tolerance classes, and by computing the disjunction normal form of discernibility function, get approach to obtain relative optimal decision rule of interval-valued information system.