Research On The Strategy Of Multi-agent Argumentation Negotiation Based On The Prospect Theory

The purpose of negotiation is to resolve the conflict of interests, and promote the negotiation effective. Research of automatic negotiation based on Agent makes full use of the intelligence and interactivity of Agent, can finish cumbersome process preferably, who has become the frontier in the field of negotiation. Effective negotiation strategy is the basis of negotiation based on Multi-Agent, which can not only help negotiators make decisions, but also can improve the negotiation efficiency,at last to promote e-commerce transaction consistency. As the new brand perspective of negotiation, argumentation is not confined to the fixed form, as a result, can improve the flexibility and adaptability of the negotiation. However most of the current argumentation research are focused on the formal level, and do not make use of the advantages of the argumentation, which are short of the research on the producing, estimating and deciding of argumentation.In view of this, the article applied the Prospect Theory to the argumentation based on Multi-Agent, and gave full play to the ability in dealing with the uncertainty decision-making. Besides, it changed the process of traditional utility maximization decision-making and took the Agent’s preference or the decision-making behavior into consideration. We set the Multi-Agent argumentation reasoning model based on the Prospect Theory, which had set up the basic application elements and the interaction rules. So we believed that this model can control the negotiation process effectively. We proposed the argumentation strategy based on the Prospect Theory, which stipulated the both parties’ attitudes toward the relative losses and gains. Based on which, we set the proposed reference point. This point could help us to learn and predict the opponent preference and make the decision more accord with person behavior. At last this research could promote the argumentation consistency.Besides, by considering the advantages of machine learning algorithms in Multi-Agent negotiation, we introduced a new intelligent optimization dynamic learning algorithm-PBIL(Population-based Incremental Learning).it could guide to select the optimal solution based on the knowledge guided through learning competitively. Besides, this algorithm could optimize the argumentation strategy based on Prospect Theory, and adjust the learning rate by setting the market evaluation. At last we formed a new adaptive capability algorithm---APBIL, whichcould optimize the negotiation process.We believe that it’s a new exploration in the field of negotiation in this article,First of all, we take the people’ subjective factors into negotiation strategy and operational mechanism research and make the negotiation process more reasonable.We regard this negotiation strategy a new train of thought to deal with the current electronic commerce negotiation dynamics, flexible and uncertainty of demand.