Study On Multi-agent Cooperation Mechanism And Cooperation Structure

Network-based multi-agent technology supports the cooperation of organization that includes the human society and machine system; at the same time the cooperation idea influences the development of multi-agent technology. How to support and promote the human cooperation and how to solve the complex problem and the optimizing question through multi-agent cooperation become a research focus in management science and artificial intelligence, and so on.Based on extensive and deep review of literature, multi-agent cooperation is studied deeply. The main contents are as follows:1,Agent and multi-agent technology, which is a theory base of the multi-agent cooperation, are introduced, and the definition of multi-agent cooperation is presented. 2,Multi-agent social convention forming mechanism is analyzed deeply . Based on the construction of multi-agent society evolving model, social convention and some related conceptions are defined; hidden social convention and decision rules are presented. During the introduction of multi-agent recognizing capability, a refined social convention forming mechanism is presented, moreover simulation experiments under more relax multi-agent game experiment environment are successfully done. 3,The trust problem is studied in-depth. First we analyzed the importance of the trust in network information society, the mental states during the human decision, and the relations between the trust and the control, then the trust is formally defined and a general trust model in electronic commerce is presented, moreover a simulation frame is introduced and under the frame we demonstrated that trust can promote the multi-agent cooperation. 4,Through the systematically analysis on deception problem,the level and the taxonomy of deception are defined, at the same time several deception punishment mechanism are introduced. 5,Multi-agent cooperation frame is presented, and under the frame we provided sixteen cooperation structures according to the three dimensions, whether cooperation is complete, whether the objective exchanging of both sides exists, and the number of both sides. Furthermore, the negotiation theory is introduced in detail. Finally the communication primitives is designed for multi-agent cooperation and algorithms of realizing the cooperation structures that presented in this paper, at one time an applying example is supplied.6,One hybrid cooperation-oriented agent structure is presented, and it is described in Z language, at the same time agent realization in Java class is introduced. In the end, a classic prototype based on the agent structure is provided.