A novel belief function reasoning approach to MCDM under uncertainty

This paper presents a novel belief function reasoning approach to the multiple criteria decision-making problem under uncertainty. In contrast to exist approaches, which make decisions based on the expected utility values derived directly from the combined belief function distributions, we introduce an alternative two-level reasoning transferable belief model approach to the aggregation and decision-making phases. Within this framework, the analyst can combine the beliefs regarding various sub-criteria at the credal level, and calculate the expected utility values for decision making at the pignistic level based on real probability distributions. We also propose a measure of uncertainty to capture the degrees of total uncertainty involved in different belief assessments. This measure can assist the decision maker in making rational decisions based on incomplete information.