Multi-Agent Negotiation Based Virtual Bundle Coalition Market System

In multi-Agent system (MAS), capsulated software Agents are able to take flexible, autonomous actions for achieving individual design objective in runtime. coalition is one of the most important methods for cooperation in MAS. For a task, resolution to optimal coalition is a comlex combinatorial optimization problem.Agent can improve ability of problem resolution and gain more utility by coalition. Negotiation is key technology for multi-Agent cooperation in problem resolution, so is coalition.As the development of intelligent Agent technology, more and more fields are involved and e-market is one of the most attractive fields. Given buyer set and seller set, it is a NP-complete problem that how we should market for gaining maximal customer surplus. How to achieve an optimal approximation in time is an urgent task. It brings about applications of intelligent Agent in e-market.Agent can be used to simulate buyer (seller) in a distributed environment. Agents can form coalition, and participants can realize free group marketing in e-market through Agent coalition.Investigate the feasibility and efficiency of Agent coalition formation. A multi-Agent negotiation based virtual bundle coalition marketing system is realized, it provides a platform for multi-Agent negotiation and coalition formation, and proves our bundle search algorithm, coalition candidation filter algorithm and extended contract net protocol. The resulting coalitions are stable in the core in terms of coalition rationality. The simulation results show that this mechanism is very efficient. The resulting cost to buyers is close to the optimal cost.