Alliance-based evidential reasoning approach with unknown evidence weights

The importance of evidence weighting is indicated.A new equivalent equation for the pignistic probability distance is presented.Pieces of evidence are classified into different alliances.Optimization models for intra- and inter-alliances are built to provide evidence weights.Rationality analysis of the ERBA approach is given in experiments. In the evidential reasoning approach of decision theory, different evidence weights can generate different combined results. Consequently, evidence weights can significantly influence solutions. In terms of the psychology of economic man, decision-makers may tend to seek similar pieces of evidence to support their own evidence and thereby form alliances. In this paper, we extend the concept of evidential reasoning (ER) to evidential reasoning based on alliances (ERBA) to obtain the weights of evidence. In the main concept of ERBA, pieces of evidence that are easy for decision-makers to negotiate are classified in the same group or alliance. On the other hand, if the pieces of evidence are not easy to negotiate, they are classified in different alliances. In this study, two negotiation optimization models were developed to provide relative importance weights based on intra- andinter-alliance evidence features. The proposed models enable weighted evidence to be combined using the ER rule. Experimental results showed that the proposed approach is rational and effective.