Principles for dynamic multi-agent organizations

Many models of organizations for multi-agent systems have been proposed so far. However the complexity implied by the design of social organizations in a given multi-agent system is often not mentioned. Too little has been said about rules that must be applied to build the architecture of acquaintances between agents. Moreover, tools for managing the dynamic evolution of organizations are seldom provided in current framework propositions. In this paper we discuss self-adaptation of organizations in multi-agent systems according to the dynamic of interactions between agents. Starting from a default organization, the architecture of acquaintances evolves autonomously depending on messages flow in order to improve the global behaviour of the system. We propose three principles that can be applied to adapt the organization: "have a good address book", "share knowledge", "recruit new able collaborators". These principles have been applied in our multi-agent platform called MAGIQUE.