Characterizing twitter user topics and communication network dynamics of the “Liberate” movement during COVID-19 using unsupervised machine learning and social network analysis

Abstract

This paper analyzes online user conversation topics and discourse on Twitter related to the “Liberate” Protest movement in reaction to social distancing guidelines at the early stages of the COVID-19 pandemic. Interdisciplinary approaches in big data, machine learning, content analysis, and social network analysis (SNA) were used to characterize the communicative behavior, conversation themes, and network structures of Liberate protest supporters and non-supporters. Tweets were content coded and grouped within topic clusters produced from an unsupervised machine learning algorithm using natural language processing. An analysis of topic clusters found that tweets that support the protests are highly concentrated and have higher volumes of replicated tweets. Protest Supporters were also more likely to retweet other users while Non-Supporters were more likely to include a URL from an outside media source and produce a unique tweet. SNA was also used to assess the characteristics of retweet networks and found that the Protester Supporter network had a more centralized structure and was strongly influenced by a political organization, in contrast to the Non-Supporter network that had a larger number of smaller and more evenly-sized nodes and more driven by media personalities and commentators. Collectively, these characteristics indicate that protest supporters had more centralized, consistent and disseminated discourse protesting COVID-19 social distancing requirements compared to non-supporters who were more diverse in their criticism of the Liberate movement and generally more fragmented in their support of public health measures. Results from this study provide important insights into pandemic communication dynamics of opposing twitter communities, including in the context of those who oppose and support public health measures in a highly politicized social and online environment. Results are important in the context of assessing the messages, communication propagation and overall activities of social media communities in response to basic public health measures needed to contain this post-digital era global pandemic.