Research On Multi-agent Group Decision Support System Of The Cooperation Based On Bayesian Network

The large data from the network environment is a hot research topics fordecision support information during information time. How to get accurate andefficient decision support information of methods are the core of the researchwork. Bayesian Network (BN), graph theory and statistics are a combinationwhich It can be used to express the advantages causal information of the method,be good at mining data relationship and uncertainty of random variables,especially in the decision support system research has an important researchvalue。Based on the theory of BN tools, aiming at the complex environmentdecision support problem, based on agent group decision support system, basedon researching on BN of learning decision methods, from the bayesian networkbased on learning and multi-agent group decision support system based on BN.inthe following two aspects of research work is as follow:(1) Find a kind of learning algorithm improved Bayesian Network structure.According to studying the modeling process and learning Bayesian Networkstructure from data from, discuss MCMC algorithm and analysis K2, focus on thetwo algorithm’s advantage, then introduced the idea of the average model that thenew algorithm, and concluded that the simulation algorithm from without priorknowledge and the constraints of the initial structure, fast convergence rate andthe structure of the right to get stability, show the dependence of right betweenthe nodes. Thus the algorithm in scientific decision-making has a highapplication value, can be applied to large data fields related decision analysis(2) Based on the proposed agent group decision support system of BN,theBN agent decision support system of the consultation process of interaction. According to the discussion and analysis in BN agent the advantage of thedecision support system, based on the study of the design of the single agent BN,master the theories model, model structure and the learning process; Researchlearning to the decision task agents organization process flow and build touchagent group decision support system of organization structure; Combined withthe given much consultation agent model, obtained the negotiation interactiveprocess, and according to the example to illustrate.In conclusion, multi-agent group decision support system based on the BNcan effectively cooperation, and can make effective and scientific decision. Andaccording to the example discover methods for science decision.Moreover,it istested that these methods are feasible.