基于优势关系粗糙集的属性约简算法

Attribution Reduction Algorithm Based on Dominance-Based Rough Set

针对目前对不同信息系统下优势粗集的处理几乎都是将其划为单值完备信息系统来处理这种现象,以及基于区别矩阵和基于正域的属性约简算法是不等价的这个问题,提出了一种基于单值完备信息系统的快速属性约简算法。首先,提出了对象集约简;其次,使用基数排序思想计算;再次,结合了区别矩阵和正域约简2种算法,在大量提高算法时间复杂度的同时提高了算法的可信度;最后,采用实验验证了约简算法的实用性。

On the fact that all the process of dominance-based rough set on different information systems is transforming them to monodrome and complete information system, and on condition of the problem that the attribution reduction algo- rithm based on discernibility matrix and the algorithm based on positive region were equivalent, in this paper, a new rapid attribution reduction algorithm which was used on dominance-based rough set was proposed. The algorithm firstly pro- posed a thought to do reduction on object set, and secondly calculated IND（C） by using the method of radix sorting to. Then a attribution reduction algorithm was proposed combining the algorithm based on discernibility matrix and the algorithm based on positive region. And this algorithm reduced time complexity and added result＇ s persuasion. At last, the relevance of this attribution reduction algorithm was turned out.