Abstract: In order to deal with the general interval-valued information system, this paper provides the definition of similarity which is based on deviation degree, gives the dual-variables precision tolerance relation which is based on similarity and similar rate, discusses the attribute reduction and judgment of interval-valued information system in the condition of dual-variables precision tolerance relation, gives a new attribute reduction algorithm which is based on the binary discerni-bility matrix, and at the same time discusses the effect of similarity and similar rate to the attribute reduction. The example shows the specific operation method of attribute reduction and the effective feasibility of algorithm.

Key words: interval-valued information system, similarity, similar rate, dual-variables precision tolerance relation, attribute reduction

摘要： 为了处理一般的区间值信息系统，给出了基于相离度的相似度定义，提出了基于相似度和相似率的双精度容差关系，讨论了在双精度容差关系下区间值信息系统的属性约简与判定，并给出了一种新的基于二进制辨识矩阵的属性约简算法，同时还分别讨论了相似度和相似率对区间值信息系统的属性约简的影响。通过实例分析说明了属性约简的具体操作方法和算法的有效可行性。

关键词: 区间值信息系统, 相似度, 相似率, 双精度容差关系, 属性约简