Sentiment Analysis of Public Opinion on Public Health Events Based on Deep Learning

In recent years, with the development of the information age, text information with subjective emotions has increased exponentially on network platforms. Analyzing emotional texts published around public health events can help relevant personnel understand and grasp people’s emotional trends, thereby making more accurate decisions in public opinion management and guidance. However, with the popularity of social media, Chinese social network texts present problems such as short length, fragmentation, semantic loss, and sparse features. This makes it difficult for sentiment analysis algorithms based on traditional deep learning to make effective emotional judgments and limits their scope of application. To address these issues, this paper proposes a deep learning model for sentiment analysis of public health event public opinion based on pre-training models, bidirectional long short-term memory networks, and attention mechanisms from multiple perspectives. The model details were optimized according to the social attributes and length of the text to improve the model’s ability to capture short text features. Compared with existing classic algorithms in real social network scenarios, this method has better effects in sentiment analysis of public health event public opinion.