Sentiment analysis of microblog public opinion based on deep learning

Public opinion sentiment analysis focuses on the public's emotional bias towards public events, and public opinion involving public health events will directly affect social stability, so it is essential for microblog sentiment analysis. In this paper, we use the RoBERTa-BiGRU-Attention model to analyze the sentiment of the text data of Weibo by taking the 2024 Paris Olympics as an example. Firstly, RoBERTa obtains the word embedding representation containing the text context information from the text data in the input layer. BiGRU obtains the character representation and sentence representation, and finally, the weight proportion of each character to the sentence in which it is located is calculated using the Attention mechanism, The text representation of the full text is obtained, and the softmax function carries out the sentiment analysis. To verify the effectiveness of the RoBERTa-BiGRU-Attention model, this paper takes Accuracy, Precision, Recall, and F1-measure as the evaluation indexes, and finds five models to compare with them. The results show that the RoBERTa-BiGRU-Attention model has good performance and advantages compared with the general model.