Sentiment Analysis Method of News Comment Based on Bidirectional Attention Mechanism
In the previous task of news comments sentiment analysis,the analysis method mainly focuses on the feature extrac-tion and optimization of comments,ignoring the complementary role of news on comments understanding.Therefore,a sentiment analysis model based on bidirectional attention mechanism is proposed to establish the connection between news and comments,en-rich the semantic background of comments,and improve the effect of sentiment analysis.The contextual features of news and com-ments are extracted through the bidirectional gated recurrent neural network,and the bidirectional attention of news to comments and comments to news is generated by calculating the correlation matrix of news and comments in the bidirectional attention layer,the results are generated in the output layer.In the comparison experiment with the bidirectional gated recurrent neural network,the model with bidirectional attention layer increases by 4.7%in the sentiment multi-classification,which also improves to a certain ex-tent in the multiple classification effect compared to other mainstream models,which proves the effectiveness of this method in the sentiment analysis of news commentary.
万方数据
维普
