关键词: 智能体, 情感强度, 交易质量, 信任更新, 交互式多准则决策法, 决策模型

Abstract: Emotional persuasion based on an Agent's trust can better utilize a variety of artificial intelligence advantages of an Agent.Aiming at the currently insufficient research on decision models,based on an Agent's own character,emotion decay and external stimulation,the algorithm of emotional intensity was constructed by using the big five characters,the OCC(Ortony,Clore,Collins) emotion model and the first law of emotional intensity.Considering the effects of transaction quality on trust,fuzzification and defuzzification were carried out on the transaction quality of an Agent by using the language evaluation method,the triangular fuzzy number and the Converting Fuzzy data into Crisp Scores Converting  Fuzzy data into Crisp Scores(CFCS) algorithm,and then the algorithm of the transaction quality was correspondingly constructed.After the existing updating factor of trust is improved,the updating algorithm of an Agent's trust was developed.On the basis of the aforementioned algorithms,the emotional persuasion decision model based on an Agent's trust was established with TOmada de Decisao Interativa e Multicritévio(TODIM) method,and the attribute weights in the model were determined by the entropy weight method.To show the appropriateness and validity of the proposed model,a numerical example was designed for illustration,and the decision results obtained from different combinations of parameter values were analyzed and compared with those of the related research.Conclusions and managerial insights were drawn accordingly and then limitations and future research directions were further summarized.

Key words: Agent, emotional intensity, transaction quality, trust update, TODIM method, decision model