Empowering Public Trust in Vaccines for Effective Outbreak Response

Abstract

Social media sites like Twitter act as vibrant centres of worldwide communication. highly influential due to the large number of users and the constant stream of information shared. This can be both positive and negative. On the negative side, social media can be a breeding ground for misinformation and can exacerbate public anxieties, especially during crisis situations like pandemics. This paper highlights Understanding how important it is to measure public opinion on social media sites is critical to surviving in today's digital environments, especially regarding concerns lingering after a pandemic. By analyzing these concerns, policymakers can make better decisions to address public anxieties and improve public health policies in the future. Preprocessing involves tokenization, stop word removal, stemming/lemmatization, and normalization. Each preprocessed token is then embedded using an LSTM layer to capture context and sequence. An attention mechanism focuses on crucial aspects of the text. Sentiment analysis with polarity and objectivity ranking is performed alongside context extraction. Finally, an LSTM classifier leverages both sentiment and context features to categorize the text. This approach offers a comprehensive method for text classification considering sentiment, context, and inherent textual features. proposed approach achieved higher accuracy (92.8%) compared to Random Forest and LSTM models used with these embeddings.