A novel probabilistic linguistic decision-making model based on discrete evidence fusion and attribute weight optimization

Probabilistic linguistic term set (PLTS) is an effective tool to describe qualitative evaluation, where decision-makers’ hesitation is expressed by multiple linguistic terms and decision-makers’ preferences for possible linguistic evaluations are obtained through probabilistic information. Recently, many multi-attribute decision models based on PLTSs have been developed greatly. However, the existing models fail to consider the interactions among multiple attributes under the framework of intuitionistic fuzzy sets for a decision-making problem given in the form of PLTSs. To solve this problem, a novel probabilistic linguistic decision-making model of discrete evidence is proposed in this paper. Specifically, an optimization model of evidence entropy is established to obtain non-additive attribute weights. Then by using these weights as fuzzy measures, a Choquet intuitionistic fuzzy logit (CIFL) model is defined to acquire decision results. The proposed CIFL model fully takes interactive attributes into account and extends the intuitionistic fuzzy logit (IFL) model. Finally, in a case study of an emergency plan selection, the optimal alternative is chosen correctly with the highest probability by the proposed model. A sensitivity analysis under two entropy functions and four scores, a comparative analysis with seven existing methods and a theoretical analysis are utilized to prove the rationality and validity of the developed model.