Sentiment analysis of COVID-19 related social distancing using twitter data based on deep learning

Social distancing is an important non-pharmaceutical intervention tool (NPIs) to prevent the spread of COVID-19. However, it also created negative impacts of economic activities. Understanding the emotions and public opinions about social distancing are important for the future policy making of COVID-19 mitigation and the assessment of public health impacts. This study collected 77,627 number of Twitter messages (tweets) between February 1, 2020 and April 30, 2020 from five English-speaking countries (United States, the United Kingdom, India, Canada, and Australia) using the social distancing keywords. We adopted a multi-module hybrid convolutional neural network model sentiment analysis on social distancing related tweets with 85.95% accuracy. This paper conducts a sentiment analysis of tweets from the public on social distancing measures in five countries. Our findings show similar sentiments in tweets from these five countries, which is more positives than negatives about social distancing measures in the public. Additionally, when the daily number of new cases changes, public sentiment fluctuates with it. We believe that social distancing is effective in preventing the spread of coronavirus.