THE CONCEPT-REASONING NETWORK AND ITS APPLICATION IN TEXT CLASSIFICATION

Following the analysis of the current text classification methods, a new classification model is proposed in this paper. The key of the model is to simulate the process of human classification. Machine learning and data mining techniques are applied to acquire knowledge and build a concept reasoning network based on semantic dictionary and large training set. Given a document, this model can activate several networks and then reason to decide which classification it belongs to. The experiment indicates that the model has high classification precision and recall.