The Research On The Cooperation Behaviors Based On MAS

With the globalization of economy and the dramatic development of science and technology, the survival environment and the modes of the enterprise have changed vehemently. In this case, in order to create a profit zoom, many enterprises adjust their strategies by means of exploring cooperative interests, creating the most possibility of the cooperative interests and seeking for a good reputation of the cooperation. Above all, their ultimate pursuit and goal is to reap much more profit. However, at present, the model for the study of the enterprise is established according to the traditional economic theories and analytical tools, which fail to provide a valid and scientific explanation to the enterprise's rationality and cooperative practice.Nevertheless, it is supposed that all the difficulties and defects can be overcome and compensated for by the application of Multi-Agent System in a model-establishing analysis of the enterprise's conduct. Multi-Agent System is both a newly-born field in the computer science and, meanwhile, a highly interdisciplinary field which bears relation with a variety of fields such as economics, philosophy, logic, ecology and sociology and so forth. Multi-Agent System is successfully free of all the defects existing in the traditional approaches of economy analysis by means of simulating the flexibility of society. It acts as a very significant and effective tool for simulating society and turns out the dominant approach in studying the artificial society.This thesis gives the present study based on the game theory and the theories of economy and takes the Multi-Agent System as the scaffolding to support the analysis. By taking the present game theory, economic theory and the Multi-Agent System approach into a total consideration, the thesis puts up with a framework of studying the enterprise's cooperative conduct; analyzes the profitable elements in the virtual system affecting each representative of the enterprises and the adjustment of Agent itself through learning in the process of interaction; discusses each of the enterprise's practical conduct of cooperating as well as competing in the mimic models according to the Multi-Agent System theory;and presents an interpretation by using the game theory and the economic theory. With what have been presented above as the aim to achieve, this thesis lays its emphasis on such aspects as follows:(1) The thesis gives an analysis of the current situation of the cooperative conduct; establishes an Agent model for a typical cooperative conduct in an enlightening way; analyzes by the emulate experiment, the influence of such elements as equality, altruism, credit, fraud, venture and the like on the cooperation in the economic world; especially emphasizes the influence of the two typical elements-credit and fraud-on the cooperation.(2) The key issue of the cooperative study based on the Multi-Agent System is how to promote cooperation, while credit is a most significant factor which is well worth considering in the Multi-Agent cooperation. The analysis and definition of credit we will study later should be achieved by a psychological analysis of the people in the process of their making decisions. This thesis brings in emulate experiments by means of which it proves that the credit mechanism is helpful for the cooperation among the Agents.(3) Fraud is one of the main factors which prevent the multi-agent cooperation from being carried out smoothly. By elaborating the problem of fraud, we find out that fraud is a normal phenomenon in the multi-agent field. We give the various levels of fraud and classify it; we also give several ways to deal with it. According to a vague control, construct a fraud model through which we obtain the countermeasures and approaches and testy them by mimic experiment. According to the experiment, it can intensify and consolidate the cooperation to recognize the conduct of fraud.The last part of the thesis summarizes the whole study and gives some suggestions for the future research in the field of multi-agent cooperation.