【摘要】 基于议题偏好修改的Agent劝说能更好提高Agent在动态环境下的谈判效率,保证谈判效果。针对该模式,结合形式逻辑理论,提出了相应的劝说机制,包括相应的表述模型和支持该模型的算法,并提出了劝说影响因子的概念,最后基于Jade平台和Java语言,通过系统实现验证了该机制。结果表明,该机制能更进一步发挥Agent在劝说中的人工智能特性。更多还原

【Abstract】 With the development of global economic integration,electronic commerce has become a significant business field because of its effectiveness and low-cost. negotiation is is an indispensable communication method for e-commerce transactions. Therefore, how to guarantee a more efficient negotiation in e-commerce is not only an important criterion to measure the core competitiveness of modern enterprises,but also a research trend in e-commerce related negotiation. However,the current negotiation system needs much human interference,which has limited influence on the improvement of effectiveness and cost reduction.Some scholars introduce the concept of agent into the design of a negotiation system. Agent-based negotiation,with less human involvement and the ability to improve the efficiency and reduce the cost of negotiation,has drawing increasingly attention in the automated negotiation research field. In general,agent-based negotiation has three approaches,including game-theoretic approach,heuristic-based approach,and persuasion-based negotiation. Comparatively,persuasion-based negotiation,which matches human negotiation thoughts and makes the negotiation more rational,has attached increasing attention in the field of automated negotiation research.In the current persuasion-based negotiation,most studies are about the generation and evaluation,but multi-issue negotiation is seldom considered. In contrary to single issue negotiation,multi-issue negotiation can reduce the possibility of negotiation deadlock to a certain degree.This paper propose an improved expression model by combining preference modification features in persuasion,and summarizing the existing research on persuasion strategies and multi-issue algorithm. The persuasion influence factor has also been considered to verify the effectiveness of the model. Moreover,an evaluation algorithm is provided along with the model and concept. Furthermore,a correspondent java-based persuasion negotiation system is developed to prove the effectiveness of the model and algorithm. At last,a comparison with similar researches is conducted to explain the contribution of the research.There will be more research on agent persuasion and multi-issue negotiation. This persuasion negotiation mode is going to have increasing influence on global economic development. Comparatively speaking,our research has only made a trivial contribution in the field,and much in-depth research is needed in the near future.更多还原