Research On Swarm Inteligence Adaptive Strategy Selection Based On Argumentation

This paper researches that how agents achieve the purpose of adaptive strategy selection by adaptive strategies by their coordination. If agents can do that, they would do well in adaptation, so better results and efficiency of negotiations can be achieved. With considering argumentation as the mode of negotiation, building the flexible mechanism of negotiation triggers and importing the third party into negotiation system, this paper finishes the framework of swarm intelligence adaptive strategy based-on argumentation. This paper also builds the models of adaptive strategy selection in the scene of single product and multiple products, and finally develops the prototype system in the JADE platform. This prototype shows the effectiveness of adaptive strategy selection. This paper mainly contains several aspects as follows:In the area of argumentation, negotiation mechanism base-on argumentation is created. In this mechanism, according to intervention time, argumentation is divided into three classes: explain, argumentation base-on expectation, argumentation base-on impasse. This mechanism is beneficial for agent to change their strategy flexibly according to environment and achieve adaptation.In the area of Third Party negotiation service support platform and swarm intelligence, Third Party negotiation service support platform which provides swarm management is considered the environment. This paper depicts that function structure which this kind of Third Party should have, and studies how buyer agents achieve their adaptation by swarm intelligence. Third Party negotiation service support platform can help users overcoming the difficulties in implementation of multi-agent system, and also provides a new way for negotiation.In the area of adaptive strategy selection, this paper studies that agents improve their adaptation by coordination between them. Swarm interaction protocol, negotiation process protocol and negotiation executive protocol are defined in this paper by referencing the Ant Colony Algorithm, Particle Swarm Optimization and other swarm intelligence algorithm. This paper depicts swarm generation algorithm and swarm organization and schedule algorithm. Finally, this paper develops the prototype system in the JADE platform, and shows the feasibility and effectiveness of theories.