针对属性权重未知的多属性决策问题,指出决策对象与理想对象之间的优势度与其属性值的概率测度及属性值大小之间有等价关系;借鉴博弈决策中的极大极小算法,提出用属性值的优势关系确定属性权重,并利用优势关系对决策对象进行排序并择优;最后,对属性值为区间数的多属性决策实例进行了分析,同时将该算法与离差最大化算法的异同点进行了对比,表明了算法的有效性. Aiming at the problem of multi-attribute decision making with unknown attribute weights, it is pointed out that there is an equivalent relationship between the dominance degree of the decision object and the ideal object and the probability measure of the attribute value and the size of the attribute value. Based on the minima algorithm in game decision, The weights of the attributes are determined by the dominant relation of the attribute values, and the decision objects are sorted and optimized according to the dominant relations. Finally, the multi-attribute decision instances whose attribute values ​​are interval numbers are analyzed, and at the same time, Similarities and differences were compared, indicating the effectiveness of the algorithm.