Abstract: The bipolar-valued information systems based on dominance relation and its application on multi-attribute decision-making have been investigated. A probability-rough model based on [α]-dominance relation is defined by given the probability of [xi] superior to [xj]. Taking into account granular computing theory and attribute types, it extends single granulation to two-granulation rough set and a new dominance degree formula is defined. Then, calculating the whole dominance degree with aggregation thought and ranking them, the best choice can be made.

Key words: bipolar-value information systems, probability-rough model, two-granulation, whole dominance degree

摘要： 讨论了优势关系下的双极值信息系统以及在多属性决策上应用。通过定义属性对象[xi]优于[xj]的概率，建立[α-]优势关系的概率粗糙模型，并依据粒计算理论和属性类别，将单粒度推广到二粒度下的粗糙集，定义了二粒度下优势度公式，再通过集成得到综合优势度，通过比较大小值确定最优决策。

关键词: 双极值信息系统, 概率粗糙模型, 二粒度, 综合优势度