中文摘要：

传统的决策表规则提取需先进行属性约简再进行值约简,过程中存在大量冗余计算,并且当数据包含一定不确定性时效果不佳。为此,提出一种最简规则获取方法,将属性约简与值约简过程合二为一,使用变精度粗糙集模型,从属性多粒度的角度分析,按粒度的大小将决策表转换成不同的知识空间,并利用矩阵简单直观的特点,在不同的知识空间内定义粒矩阵、粒关系矩阵等概念,通过充分挖掘隐含在β粒关系矩阵中的启发式信息Sω,确定属性约简顺序,实现对不同粒度知识空间下最简规则的快速获取;设置覆盖率α为终止条件,以概率方法加快算法收敛速度。最后,从实例分析以及与现有算法进行UCI测试对比两方面对算法进行了验证,实验结果证明了所提算法的正确性与有效性。

英文摘要：

Traditional algorithm extraction rules by attribute reduction and attribute values reduction.There is a lot of redundant computation in the process and the result is not good when the data contains noise.Thus,the variable precision rough set model is used to acquire the rules of decision and attribute reduction and value reduction process are combined in this paper.The decision table is granulated into different granular spaces from fine to coarse in the perspective of attribute multi-granulation.By defining granular matrix,βgranular relation matrix,as well as mining the heuristic information Sωhidden in theβmatrices to determine the order of attribute reduction,the rules in different granular space are acquired.By defining the concept of coverageα,the convergence speed of the algorithm is accelerated by the method of probability.Finally,the proposed algorithm is illustrated by an example and verified by UCI test set,proving the validity and effectiveness of the proposed algorithm.