This paper presents the design and implementation of an automatic classification system based on Bayes tree. Analyzed the characteristics of the tree, the system research potential issues that may affect the classification results, factors that should be considered when the trees classification modeling is proposed and then induced by these factors, and complete the characteristics of six main aspects of This article from the high abstract point of view, the most relevant content and information classification, extraction set trees classification model, and finally implement a Bias classification model of automatic classification system based on tree Bias. This paper designs and implements the self learning function parameters in tree automatic classification system, in order to obtain the best classification results of actual data through self-study and certain adjustment of parameters. The experimental results show that the proposed method can better solve tree automatic classification problem.