Abstract: In this paper，with respect to the ordinal variables，the definitions of attribute reduction of dominance-based rough set approach in ordered information systems and ordered decision tables are given.Then the approach of sorting objects comprehensively using rough set theory is brought in to obtain a series of ranks.Based on these ranks，nonparametric methods to analyze correlation between two attribute subsets are introduced，and Spearman rank correlation coefficient is used as a measure of attribute correlation.Based on this measure，a new method of attribute reduction of ordered information systems and ordered decision tables is presented without changing the ordinal information of the universe.Finally，the experiments show that the approach proposed is feasible and it provides a statistical evidence for rough set approach.

Key words: rough set, dominance relation, rank correlation, attribute reduction

摘要： 针对有序尺度变量，给出了有序信息系统与有序决策表在优势关系下的粗糙集约简定义；利用粗糙集方法将对象综合排序，进而得到一组秩；根据这些秩，运用非参数统计的思想研究了两个属性子集之间的相关性，并将Spearman秩相关系数作为属性相关性度量；在不改变总体序信息情况下，给出了基于此度量对有序信息系统与有序决策表进行约简的新方法。最后通过数值例子说明该方法是可行的，且为粗糙集方法提供了统计依据。

关键词: 粗糙集, 优势关系, 秩相关, 属性约简