Customer Selection For Multi-Agent Negotiation Based On Prediction And Integrity

With flourish development of e-commerce, it is a new hotspot of research thatAgent technology is applied in the field electronic commerce into consultationscustomer choice in recent years. As for a customer selection of the e-commerce,existing the consultation customers choose method there is a certain lack, mainly thestudy of the negotiation history information is not sufficient, the accuracy ofprediction is not high, and did not consider the impact of the integrity of the wholeproduce. So this paper proposes a selection method based on prediction and integrityof the Multi-Agent negotiation customers. Firstly, according to the interactive featuresof Agent negotiation, introduce the GE-LS-SVR, from the Agent's negotiation historyinformation learning samples to train the GE-LS-SVR, to predict the outcome of theconsultations of trading clients in the negotiation of trading that may occur inconsultation and the corresponding income in consultation; Secondly, in view of thissituation that the transaction customers agent does keep its promise, and the case ofbad faith the way trading in based on multi-agent e-commerce transaction executionand evaluation phases, multi-agent-based e-commerce negotiation integrity model isproposed to establish a model of integrity in the transaction between the agents, and togiven the calculation of the integrity and credibility of assessment methods; Thirdly,according to agent self-interest features, to choose the final income value ofconsultations clients to combine the predicting the negotiation proceeds with integrity.Finally, effectiveness and superiority of the method presented in this paper areverified through simulation experiments.