The traditional classification method of Chinese sensitive information mainly relies on the frequency of the co-occurrence of sensitive words and keywords. However, it is difficult to detect the meaning and context relationship of some complex statements. In this paper, a new model is proposed to classify Chinese network sensitive information. The model, which is based on CNN (Convolutional Neural Network) and latest pre-trained BERT (Bidirectional Encoder Representation from Transformers), is called the BERT-CNN deep learning model. Firstly, n...
