Multi-agent Architecture-based Intelligent Decision Support System

To solve the decision problem of complicated ill structure or non-structure, \ve need knowledge and intelligent Intelligent Decision Support Systems (EDSS) is a combination result of Decision Support Systems and Artificial Intelligent (AI). We apply the Multiple Agents Model of AI to Decision Support Systems, and each agent has respective task and function By use of the interaction among agents, Systems can share not only data, but also resource such as models, rules, schemes and methods, and enlarge the scope of decision knowledge source. Moreover, because of the agent autonomy, the fussy, mechanical task can be finished by correlative agent. So we can enhance the systems ability and function for handling decision problem. The paper studies IDSS based on multiple agents from three aspects:Firstly Bayesian Network Theory is our basic model on theory. It builds a circular, non-structure reasoning system. With Bayesian Network Theory as basic model of agents, we design distributed Intelligent Decision Support Systems. Decision Support Systems based on Bayesian Network Agent have characteristics as follows:(1) Bayesian Network Agent deduces many possibilities by using quantitative probability. So it can balance all factors affecting decision accurately.(2) Bayesian Network methods can avoid subjective prejudice brought by apriority information, sightless search when we lack sample information, and the infection of interferential factors. Therefore, it is the same with Data Mining with probability statistic character and knowledge discovery problems, especially with die problems that obtain sample information or need high cost.(3) Systems can learn knowledge and perform reasoning under the uncertainty environment automatically.Secondly making full use of JESS flexible reasoning mechanism and its character that combine Expert System and JAVA language, we get the way of designing Intelligent Agent System based on Bayesian Network.Thirdly, on the application, the decision problems of greening management are the system background Analyzing die characteristic of greening management decision system, we design the Intelligent Decision Support System based on Bayesian Network for greening management.