A Multi-Agent E-Commerce System Based on Evolutionary Process

Nowadays, multi-agent technology is a pop research domain. Agent has the attributes of autonomy, consecutive execution, individuation, abundant semantic expression, learning and adaptability etc, which make it especially suitable for applications that have lots of heterogeneous information and processes. The agent-mediated electronic commerce is one of important applications of multi-agent systems. This paper makes the agent-mediated electronic commerce as background. The research works include the theories and methods of system structure, negotiation, protocol and communication in E-commerce. In the system, how to make the negotiation agents gain satisfying result farthest and negotiate efficiently is a key issue. As for this problem, a genetic algorithm based on metropolis rule is present. In the experiments, the two methods are used to compare. The first is simple genetic algorithm, the second is the genetic algorithm based on metropolis rule. The difference is the later algorithm adds the metropolis rule. The experiments show that the e-commerce system based on evolutionary process present in this paper can help agents to negotiation more efficiently.