Sentiment Analysis of Weibo Comments Based on Bi_directional Long Short-Term Memory Network

In the problem of sentiment analysis in Weibo comments, contextual information has a significant impact on sentiment analysis. If contextual information cannot be considered, the accuracy of sentiment analysis will be low. The paper uses a Bi_directional Long Short-Term Memory Veural network(BiLSTM) model for sentiment analysis, considering the influence of contextual information to improve the accuracy of sentiment analysis in Weibo comments. To fully consider the impact of facial expression information on emotional orientation in Weibo comments, in the data preprocessing process, facial expression information is extracted from Weibo to retain it for later emotional judgment; Secondly, using CBOW model of Word2Vec to convert word vectors, Weibo comments can be represented in vector form; Finally, sentiment analysis was conducted using the BiLSTM network model to determine the emotional tendencies of Weibo comments. Compared with RNN and LSTM network models, this method has improved accuracy, recall, and F1 value. The experimental results indicate that the method used to implement sentiment analysis has significant practical value in practical applications.