非一致决策表的一种新的水平分解方法

A New Horizontal Decomposition Approach of Inconsistent Decision Tables

在提出三异等价类和三异组这2个新概念的基础上,给出了基于粗糙集的一种新的非一致决策表水平分解方法。该方法也可以像现有分解方法那样将非一致决策表分解为1个一致决策表和1个完全不一致决策表,但分解得到的一致决策表可能比用现有方法分解得到的一致决策表更大一些,造成的信息损失更小一些。同时,新方法还可以用来分解一部分完全不一致的决策表,而现有方法无法分解这些完全不一致的决策表。

On the basis of the presented two new concepts of tri-singular equivalence class and tri-singular set a new approach based on the rough set for horizontal decomposition of inconsistent decision tables is proposed. The new approach, just as the existing approach does, can decompose an inconsistent decision table into two decision tables, i.e., a consistent decision table and a complete inconsistent decision table. But the consistent decision table decomposed may be greater than that obtained by using the existing approach, thus causing less information loss. It is noteworthy that the new approach may be used to decompose parts of complete inconsistent decision tables, whereas complete decision tables can not be done by u- sing the existing approach.