A GAHP approach for probability elicitation for influence diagrams

During the process of building influence diagrams for real-world decision analysis, the most challenging task is probability elicitation from domain experts. Many human experts are used to describing probabilities by verbal or other inexactness expressions. It is hard for them to directly provide numerical probabilities. In this paper, we deal with this issue by using the Group Analytic Hierarchy Process(GAHP), which allows a group of experts to assess probabilities by making pairwise comparisons among mutually exclusive and collectively exhaustive events. The adoption of GAHP not only offer a convenience way to make judgement, but also improve the accuracy of the probabilities by combining multiple experts' opinions. We give the detailed steps of the elicitation process and finally use this approach to construct the influence diagram for decision making for college student recruitment.