Leveraging Twitter data to understand public sentiment for the COVID‐19 outbreak in Singapore

Abstract

Microblogging has become one of the most useful tools for sharing everyday life events and news and for expressing opinions about those events. As Twitter posts are short and constantly being generated, they are a great source for providing public sentiment towards events that occurred throughout the COVID-19 period in Singapore. In this project, we perform sentiment analysis and topic modeling on the tweets about COVID-19 in Singapore, from 1 February 2020 to 31 August 2020. We accomplished this by collecting tweets discussing about COVID-19 and geolocated as ‘Singapore’, using the Python library ‘SNSCRAPE’. We used the sentiments returned from the VADER lexicon-based classifier and emotions from pre-trained recurrent neural networks to find correlations between real-life events and sentiment changes throughout the whole period. From our analysis, we discovered an increase in tweets about COVID-19 during key periods such as the circuit breaker and found that the overall sentiment polarity was dominantly positive. However, emotion analysis revealed that there were changes in the prevalence of fear and joy emotions over time, due to real-life COVID-19 developments in Singapore. Additionally, sentiment polarity was found to differ from topic to topic.