Multi-label text classification method based on ReBERTa-TextCNN

Multi Label Text Classification (MLTC) is an important research problem widely used in the field of natural language processing (NLP) text classification. Aiming at the problem of small sample data sets and unbalanced labels, this paper proposes a multi label text classification method based on prompt learning and RoBERTA Te People's Republic of ChinaNN model. First, use the RoBERta model to obtain the label word vector in the text, and obtain the global features containing the text context information. The dynamic mask can improve the efficiency of the work. Then use the TePeople's Republic of China NN model to extract the local features of the text, and use the feature fusion method to fuse the global features and local features, so as to improve the classification accuracy of the model. Among them, prompt learning is used to process small sample data, and loss function is used to solve the problem of label imbalance and longtail distribution. The experiment on the actual dataset shows that the accuracy rate of multi label text classification reaches 96%, which reflects the advantage of this model in effectively improving the effect of multi label text classification.