Consensus reaching process has received increasing attention in recent years,as the demand for decision results with mutual agreement has greatly grown.In order to improve the efficiency of CRP,different consensus models have been proposed.Specific patterns of behavior presented by DMs,such as non-cooperative behaviors and minority opinions,are strictly supervised in these models.However,not every behavior is specifically defined and given directed treatment,this include non-cooperative behavior presented by highly-weighted clusters and non-cooperative behavior of milder levels,which may seriously bias group consensus if not tackled carefully.If consensus reaching process can be equipped with the characteristic to recognize the behavior patterns behind every decision-making behavior,the convergence of group consensus can be greatly enhanced.In this paper,we formulated a group decision making problem under multi-granular linguistic information environment and abstracted it to a mathematical model.A general framework of how to handle such problem is also depicted.Second,a novel behavior classification model to recognize the pattern of every modification behavior is constructed.The essence lies in the calculation of a cooperative index and a non-cooperative index,which can be utilized to classify three kinds of modification behaviors.Then,according to the former classification results,every behavior is given a directed treatment,either reward or penalty,under an innovative application of uninorm aggregation operator.Furthermore,a floating neutral element is introduced into the uninorm aggregation operator to lay stricter supervision upon highly-weighted clusters to prevent their possible non-cooperative from sabotaging group consensus.Finally,an illustrative example and a numerical simulation together prove that this model is of high efficiency and feasibility.The merit of the proposed model lies in its ability to make rational quantification and detailed classification of every behavior involved in the model.Directed treatments upon decision makers can thus improving the convergence of group consensus.The model can provide favorable intelligence support to real-world group decision making problems. |