A Chinese Text Classification Method Based on BERT and Convolutional Neural Network

Abstract

Text classification has always been an important task in natural language processing. In recent years, text classification has been widely used in emotion analysis, intention recognition, intelligent question answering and other fields. In this paper, the word vector is generated based on the Bert model, and the text features extracted by Convolutional Neural Network (CNN) are fused to get more effective features, so as to complete the Chinese text classification. Experiments are conducted on the public data set. Compared with the text classification model in recent years, it is proved that the Bert+CNN model can accurately classify Chinese text, effectively prevent over fitting, and has good generalization.