Automatic text classification based on knowledge tree

Automatic text classification is one of important fields in intelligent information process. Most researchers focus on statistic method (Rocchio, SVM, KNN etc.) which is based on vector space model (VSM) representing text. On the basis of analyzing their disadvantages, a new method -automatic text classification based on background knowledge is proposed in this paper. This method is to simulate the classification process of human being. And it includes background knowledge and classification algorithm in order to make computer cognitive ability. It combines text semantic structure and background knowledge to activate relative branches of knowledge tree and decide which classification it belongs to by reasoning. The experiment indicates that the model has higher classification precision and recall.