Second-order uncertainty calculations by using the imprecise Dirichlet model

Abstract

Natural extension is a powerful tool for combining the expert judgments in the framework of imprecise probability theory. However, it assumes that every judgment is “true” and this fact leads to some difficulties in many applications. Therefore, a second-order uncertainty model is considered in the paper where probabilities on the second-order level are taken by using the imprecise Dirichlet model. The approach proposed in the paper is illustrated by application and auxiliary examples.