Computational method for agent-based E-commerce negotiations with adaptive negotiation behaviors

This paper presents a computational method to organize agent-based E-commerce negotiations with adaptive negotiation behaviors aiming at enhancing the negotiation power and flexibility of software agents to alleviate human involvements in Ecommerce negotiations. Firstly, the computational expression of E-commerce negotiation, including negotiation issues and strategies, is specified to assist agents’ computing functions. Then, an adaptive negotiation behavior configuration mechanism is proposed to tackle the negotiation dynamics through computation. In this three-staged mechanism, agents’ negotiation behaviors are deployed by a case-based strategy assignment mechanism before the starting of negotiation; then along the on-going negotiation sequence, opponents’ negotiation behaviors are tracked through Back-Propagation Neural Network (BP_NN) learning model to make strategy adjustment to confront the opponent. After the negotiation, opponents’ concession functions are recorded and analysed using time series measure. Finally, the feasibility of the BP_NN learning model is verified through a set of tests. The computational negotiation method is exemplified using a two-issue buyer-seller negotiation case. The outcomes show that the adaptive negotiation behavior configuration mechanism can benefit an agent to win more in the E-commerce negotiation