Research On Algorithm For Detecting Cause-and-Effect Chains In Bayesian Network And Its Applications

Bayesian network in the 1980s is the uncertainty of reasoning, it is dependent on relationships and provides a causal link between the natural and effective way of expression. Bayesian network has probability reasoning ability, semantic clear, easily understood, and other technical characteristics, can be found on the potential relationships between the data and model, data mining shows that the unique superiority. It is for the starting point, Bayesian network inference as a core research in this article, by theoretical research and experimental analysis, provide an effective basis for the practical application of Bayesian network.In this paper, the work reflected in the following areas: (1) Bayesian network theory has a systematic exposition, Bayesian network model describes the composition and uses the example of a warning network model being semantic detail; (2) Concluding and sum up the Bayesian network of advantages and characteristics compared with other methods,the Bayesian network inference mechanism and mode, the Bayesian inference is NP hard; (3)Bayesian network causal chain search algorithms, and the forward and backward reasoning algorithms.In the applications of Bayesian network, we often need to search nodes in some causal link to determine what kind of causal relationship exists. In this paper, innovation reflected in: Applying directed acyclic network' simple path detecting algorithm to Bayesian network' cause-and-effect chains detecting. According to the nodes' conditional probability table of the cause-and-effect chains, we can calculate the nodes' status probability and complete the probability's spread and renew in each cause-and-effect chains. Applying the onward and backward reason method in this text to the cause-and-effect chains' sensitive analysis, it can provide the basis for the sciencal decision.