Multi-agent based IDSS architecture and negotiation mechanism

We propose a general multiagent architecture for intelligent decision support system (MAIDSS). In this architecture, there are three types of agents: interface agents, task agents and information agents. In order to solve complex problem, multiple agents are divided into several teams, each team is made of some agents and takes charge of one subproblem. Each agent in a team embodies its own algorithm for creating a solution, and one agent may modify the solutions of the other agents by means of negotiating. Humans may also participate as agents in the process of decision-making.