A novel systematic approach of constructing protests repertoires from social media: comparing the roles of organizational and non-organizational actors in social movement

Activism today is no longer bound to be organization-led, but has evolved to be mostly crowd-enabled connective actions via social media where much of the coordination and mobilization take place. This poses a challenge for social movement researchers to even keep track of the full set of action repertoires. In the social movement in Hong Kong in 2019, protesters have relied on Telegram, an encrypted messaging service, and other digital channels to mobilize thousands of collective actions of various scales and disseminate real-time updates on police’s anti-riot measures such as the use of tear gas. The months-long conflicts and the lack of official statistics render conventional manual data collection approaches difficult to implement. Using text-mining techniques, we extracted spatial–temporal information of the protesters’ call for actions and the police’s tear gas use in the social movement from over 12,000 messages collected from more than 100 Telegram channels operated by the activists and news media. The validation shows that the resulting datasets are more inclusive, especially small-scale actions, than manually compiled ones. Using the data, we identify a pattern of hybridized mobilization between organizational and non-organizational activists in the Anti-ELAB movement. This paper demonstrates how utilizing social media data can complement existing data collection methods and build a more-comprehensive record of collective actions with greater potential in supporting social movement research in the age of digitization.