相对比较信息协调性与排序结果不一致性比较

On Comparison Consistency and Ranking Result Inconsistency

讨论了从相对比较矩阵获取排序结果的两种不同方法（特征根法和几何平均法）。采用特征根方法时，同一问题因描述方式不同得出不同排序结果（排序结果不一致性）的原因在于原始信息的不协调。采用几何平均方法虽然不存在特征根方法存在的问题，但其排序结果的可信性依然取决于原始信息是否协调。Ｓａａｔｙ 提出的协调性指标不存在不一致现象，只要达到一定的协调指标值，无论采用何种方法得到的排序结果都可认为是可信及可接受的。

This paper deals with the ranking derivation problem using two methods (Eigen vector Method and Geometric Mean Method). It is found that when using E M to deal with the problem described in different ways, the ranking result inconsistency is caused by the inconsistency of the comparison information. No ranking result inconsistency exists when using G M, but the confidence of ranking results still depends on the consistency level of the comparison information. Saaty's Consistency Ratio can be applied to both methods in order to test the consistency of the comparison information and guarantee the confidence of the ranking result.