Evolution of COVID-19 Tweets about Southeast Asian Countries: Topic Modelling and Sentiment Analyses (Preprint)
BACKGROUND Emergence and evolution of online discourse on Twitter concerning the COVID-19 pandemic are crucial for public health and public relation officials to understand the prevailing perception of the disease and its impact. Despite the global scale of this pandemic, comparison and contrast of topics, sentiment and emotions of tweets among countries are limited. Further, most previous studies covered a short timeframe due to the recency of the event and the large volume of tweets. OBJECTIVE The purposes of this research were to 1) identify the multiplicity of public discourse during the COVID-19 pandemic and how they evolved, (2) compare and contrast sentiment levels and (3) compare emotions about countries over time. METHODS The research scope covered ten countries in Southeast Asia (SEA). We analysed 115,553 tweets that mentioned these countries from 22 January 2020 to 31 July 2021 using a topic modelling algorithm (latent Dirichlet allocation — LDA), VADER and NRC sentiment analyses. RESULTS We identified the emergence and evolution of twelve topics that were further grouped into early, parallel and late discourse. Each country had distinct topics and trends depending on differing internal and external factors. Sentiment improved with positive tweets of hope and fell when negative narratives predominated as a result of new waves of COVID-19. Four emotions that are prevalent during the pandemic were trust, fear, anticipation and sadness CONCLUSIONS This research detected the many divergent phases and faces of the pandemic on Twitter, the intertwined discourse of a double disaster and the role played by an infodemic on place brand association. Our results will assist public health and public relations officials to better understand how public discourse emerged and evolved with the COVID-19 pandemic.