Microblog Sentiment Classification Method Based on Dual Attention Mechanism and Bidirectional LSTM

Abstract

In the information age, the network technology continues to develop. As an emerging social media, Sina Weibo has a huge user base. Every day, hundreds of millions of users express their opinions on hot events, or share the joys and worries in life on the Weibo platform. Therefore, the analysis of the user’s emotion has broad application prospects, which could also be used in the fields of public opinion monitoring, opinion guidance, and advertisement placement. This paper proposes a microblog sentiment classification method based on dual attention mechanism and bidirectional LSTM. Firstly, the bidirectional LSTM model is used to semantically encode the microblog text, then the self-attention and sentiment word attention are introduced into the bidirectional LSTM model. Finally, the Softmax classifier is used to classify the sentiment of microblog. In order to verify the validity of the model, several groups of comparative experiments are carried out, which use NLPCC2013 and NLPCC2014 evaluation task datasets as experimental data sets. The results show that the proposed microblog sentiment classification model based on dual attention mechanism and bidirectional LSTM is superior.