Research on Chemical News Text Classification Method Based on BERT-ATTCNN Model
In recent years,due to the increase in the number of chemical news texts,it has become in-creasingly difficult for chemical practitioners to obtain high-quality chemical news text information,and text classification technology can help chemical practitioners find the information they need more easi-ly.This paper proposes the BERT-ATTCNN(BERT-Attention-TextCNN)model,which combines the BERT and TextCNN models and combines the Attention mechanism for chemical news text classifica-tion.This model can learn the semantic representation and feature extraction of text information at the same time,so as to achieve more accuracy classification tasks.Through experimental verification,the BERT-ATTCNN model performed well in the chemical news classification task,achieving an accuracy rate of 97.65%,with higher classification accuracy and better generalization ability.Therefore,the BERT-ATTCNN model can provide an effective solution for natural language processing tasks such as chemical news classification.
text classificationnatural language processingBERTAttention mechanismTextCNN
NETL
NSTL
万方数据
