Abstract: The computation of a core of a decision table is one of important parts researched in rough set theory, which mostly is also a key step of many existing algorithms of attributes reduction. At present, many algorithms were proposed for the computation of a core. However, very little work has been done in case of updating.Thus,Ref[9] introduces an incremental updating algorithm of the computation of a core, but only considers the case of inserting. Therefore, in this paper, we introduce an updating algorithm of the computation of a core based on discernibility matrix in the case of updating, which only inserts, deletes or adjusts corresponding rows and columns when updating the discernibility matrix, so the updating efficiency of a core is remarkably improved. Experimental results show that the algorithm of this paper is efficient and effective.

Key words: rough set, discernibility matrix, core, updating

摘要： 核的求解是粗糙集理论的重要研究内容之一,也是很多属性约简的关键步骤.目前已有一些关于核的求解算法,但有关核的更新算法却报道不多.文献[9]给出了对象增加情况下核的更新算法,但没有考虑对象修改情况下的核更新问题.为此,本文提出一种基于差别矩阵的核更新算法,主要考虑对象值被修改情况下的核更新问题,将核的更新转化为差别矩阵的更新.该算法在更新差别矩阵时只对涉及的少部分行,列进行删除,插入或调整,因而可有效提高核的更新效率.实验结果表明本文提出的算法是有效可行的.

关键词: 粗糙集, 差别矩阵, 核, 修改