不一致决策表中规则提取的矩阵算法

由于数据采集能力不足等原因,决策表通常都不是一致的。如果将不一致的对象完全删除,则丢失了大量隐含在这些对象中的信息。针对不一致决策表,在分配约简、分布约简和最大分布约简的定义基础上,通过定义相应的决策矩阵并比较它们与条件属性矩阵的关系,得到提取信息系统的所有分配规则、分布规则和最大分布规则的矩阵方法。该方法的优点是直观有效,能获得所有规则,并同时得到相应的约简。

Matrix computation for rule extraction in inconsistent decision tables

Rule extraction from decision tables is one of the most important tasks in rough set theory. Usually, it is difficult to keep an information system consistent. Therefore, how to extract rules from inconsistent decision tables is valuable. The decision matrices based on distribution reduction, maximum distribution reduction and assignment reduction are defined first. Then rule extraction from inconsistent decision tables is realized by comparing the decision matrix and the matrix determined by condition attribute subsets. Finally, an experiment proves the validity of this method. The method shows the merits in extracting all rules from inconsistent decision tables and achieving expected reduction.