摘要： 在提出基于优势关系粗糙集方法下的类区分矩阵概念后，为解决不一致数据问题，新的优势区分矩阵定义及其求核方法被提出，但是该方法计算代价高。为了能够快速求出存在不一致数据的信息系统的核，该文给出改进的优势区分矩阵定义和求核算法，其空间和时间复杂度都优于现有的算法。实验证明，该算法适用于处理大数据集。

关键词: 粗糙集, 属性约简, 优势区分矩阵, 核

Abstract: The attributes core of a decision table is the start point to many existing algorithms of attributes reduction. In order to solve the inconsistent data problem, A new discernibility matrix definition based on Dominance based Rough Set Approach(DRSA) and the computation of core is proposed, but it is too complex. An improved discernibility matrix definition with a method for computing the core is introduced, which can get the right core with low space and time complexity. Experimental results prove the effectiveness of the algorithm, and it suits the large data.

Key words: rough set, attributes reduction, dominance discernibility matrix, core

中图分类号: 

• TP311