Research On Talents Aggregation And Organization Emergence Based On Multi-Agents System

Computational experiment methods are increasingly applied to various fields of social sciences and show surprised findings, so it is of innovative significance using computational experiment methods to research organizational management problems. There have been many traditional methodologies in organizational management research, such as transaction cost economics, behavior analysis, strategic theory, organization ecology, and evolutionary theory, which explaining the real world. Computational experiment methods pay more attention to the diversity and nonlinearity and adaptability of the reality. The comprehensive and integrated methodology based on the computational experiment is put forward to and the model framework of the artificial organization is developed. Then talents aggregation model and organization emergence model based on multi-agents system are designed.(1) The comprehensive and integrated methodology based on the computational experiment is proposed. From the point of view of the horizontal integration of the traditional methodology and computational experiment methodology, we not only admit and make full use of the advanced and correct scientific achievements of human beings, but also study the new potential methodology. From the longitudinal comprehensive angle, we need to combine "bottom-up" and "top down" methodology together. The computational experiment methodology based on complex adaptive system belongs to bottom-up methods, while most of the traditional methodologies are top down methods because the traditional organization theory considers more the integrity of the system. The bottom-up methodology is used right in the general nature or biological world. But it could not explain the reality of organizations well only using the bottom-up methodology because the man as the advanced animal has the wisdom and the ability of creation and often give orders. Therefore, we should combine "bottom-up" and "top down" methodology and integrate the computational experiment methodology and the traditional methodology to form the comprehensive and integrated methodology based on the computational experiment.(2) The artificial organization model framework is developed. The organization is a typical complex adaptive system and can be divided into human organization and artificial organization. The general model framework of the task oriented human organization is put forward to on the basis of the entity components of the organization by analyzing the limitations of individuals and the characteristics of traditional organizations and the complexity and uncertainty of organization environment. The various kinds of networks inside the organization and between organizations are analyzed.The artificial organization theory establishes new concepts and new theory and new knowledge of organization and develops analysis tools and programs for validating simulated organization model based on the multi-agents system and the application of mathematical method and computational method. Inversely, the result and conclusion of running the above computer model of the artificial organization is applied to instruct the practice of the man. According to the characteristics of complex adaptive system, the structure of the adaptive agent is designed and the knowledge acquisition and the learning model and the interactive mode and the behavior mechanism of the adaptive agent are discussed. The basic model framework of artificial organization is established and the background assumption and the design principles and the discipline base and the basic elements of the artificial organization are analyzed. The main study contents of the artificial organization are the organization emergence and the evolutions inside the organization and the organizational adaptation.(3) The concept and theory basis of talents aggregation are discussed and the static game model of talents flow and the evolutionary game model of talents aggregation are established to analyze the main existed obstacles and policy guidance. The talents aggregation model based on multi-agents system is developed and the goal, the environment, the agent and the design of the model are described. The conclusion on interactive impact and interaction mechanism between the regional economic society and individuals are discussed in the mode of information asymmetry environment, and regional strong alliance, and regional weak alliance, and complete information environment, and the closed system such as the island on the analysis and verification of experimental results and the plentiful experiment materials getting through running the talents aggregation model.(4) The traditional view of the organization origin theories and transaction cost economics are analyzed and the organization boundary definitions given by the new classical economics and the new institutional economics are discussed. We describe the basic concepts and features and analysis index of the social network analysis because of the social network of organization is the foundation environment that organization survives, and then we explore the construction of the team, especially the innovation team built from social capital concepts. The organization emergence model based on the multi-agents system is developed by analyzing the environment and agents and behavior rules. The running results of the computational model present the diversity of organization as a complex adaptive system. At the same time, it is proved and verified that the existence of the organization is not only because of the existence of transaction costs but also because of the limited ability of individual and the strong ability of organizations. This conclusion exactly matches the principle of system science "1+1> 2". Especially the diversity of the output of the computational model shows that the traditional theory can be verified directly or even new theory can be found by inputting the assumptions and rules of traditional theory as parameters and running the model and exploring the model output.