For the unsure attribute weights problem of multiple attribute decision, the paper proposes the superiority index
relationship between the decision-making object and the ideal object has the relation of equivalence with the probability
measure of attribute value. According to minimal and maximal rules, this paper propose the measurement of attribute value
based on superiority index relationship. Then following the superiority index relationship between each decision making
object and overall object, the information weights are ranked and picked over. Finally, the probabilistic multi-attribute
decision making problem of using attribute value as interval number is analyzed. At the same time, the characters between
maximizing deviations and the algorithm proposed are compared, which shows the effectiveness of the algorithm.