Abstract:

Along with the rapid development of Internet technology and the rise of new social media， the pattern of group decision making has been changed profoundly， and it often occurs in a linguistic context. The multi-granular linguistic representations and semantics complexity have been the focuses in computing with words and linguistic decision-making. The personalized individual semantics （PISs） model with continuous learning and behavior modeling is proposed to support group consensus decision-making under multi-granular linguistic environment. Firstly， it is considered that the PISs of decision makers will be affected by the decision-making environment and decision-makers' preferences， so it will be changing dynamically during the consensus reaching process. Thus， a PISs model of behavior modeling and continuous learning based on prospect theory has been proposed. Then， the consensus measurement and feedback recommendation based on PISs under multi-granular linguistic context are developed to support the consensus process. Finally， numerical and simulation analysis are presented to illustrate the use of the proposed model， and to show the influences of the decision-making behaviors （e.g.， dynamic PISs， loss aversion） on the evolution of group consensus. The research results in this paper show that the PISs and consensus decision-making model can be improved by integrating continuous learning and behavior modeling.

Key words: multi-granular linguistic decision making, continuous learning, prospect theory, personalized individual semantics, consensus