序区间值决策系统中最优可信规则获取与约简

Acquisition of all credible rules and reductions in ordered interval-valued decision systems

基于描述子的规则获取可导出序值决策系统中的所有可信规则,但对包含区间值序决策系统却不能有效支持。因此,首先根据每个属性值域的范围,提出了一个区间段值的概念,用以将序区间值决策系统转换为序区间段值决策系统;然后,在序区间段值决策系统中提出了基于区间段值的优势和弱势描述子概念,用以导出序区间值决策系统中的所有可信规则;最后,研究了两种新的描述子的约简以及相对约简问题,给出了相应的判定定理与区分函数。以上为从序区间值决策系统中获取有效的最优可信决策规则提供了一种新理论基础与操作手段。

In the ordered decision system,it can induce all of the credible rules based on the descriptors,but which cannot direct support for interval value.To solve this problem,this paper proposed a new concept of segmented interval value based on the range of domain of each condition attributes,then the ordered interval-valued decision system could be changed to an ordered segmented interval-valued decision system.Secondly,based on the segmented interval value,it proposed the concepts of the increasing and decreasing descriptors.Finally,it investigated the reductions and relative reductions in terms of the decision classes of these two new descriptors,and gave the judgment theorems and discernibility functions with respect to these reductions.These methods provide a theory and practical approach,which can generate all compact and effective optimal credible decision rules from the old ordered decision system including interval data.