Deep Sentiment Analysis System with Attention Mechanism for the COVID-19 Vaccine

Abstract

Sentiment analysis has attracted huge interest, which has been a trend topic in last years. It has significant applications in several areas, such as marketing based on opinion recognition and mining, movie reviews, product reviews, and healthcare-based sentiment understanding. In this paper, COVID-19 vaccine has been considered as an experimental design and performs sentiment analysis to understand the opinions of the public toward getting vaccinated. The topic of vaccination has been associated with a great deal of hesitancy and different points of view from people who may trust or distrust taking the vaccine. The proposed system aims to understanding data from chats related to the COVID-19 vaccine on the Twitter platform. A deep learning framework has been built based on a bidirectional long-short-term memory (Bi-LSTM) network and use an attention mechanism to obtain precise results. Three categories are used to classify the obtained results as positive, negative, neutral. The overall accuracy of the proposed method is found to be 94%, in addition accuracy of our case study results show for the three opinion mining classes of negative, neutral, and positive on the training set was 0.96%, 0.89%, and 0.95%, respectively. On the test data, the accuracy was 0.96% for negative sentiment, 0.88% for neutral sentiment, and 0.95% for positive sentiment.