Mapping Extremist Discourse Communities on Telegram: The Case of the Russian Imperial Movement and Its Affiliates

Measuring extremist behavior and discourse online presents a significant methodological challenge largely due to the volume of content. This paper assesses the extent to which discourse communities created by extremists identified through qualitative analysis in prior research are present in a novel Telegram dataset and can be identified through Latent Dirichlet Allocation topic modeling and network analysis. It then compares the results of Louvian and Girvan-Newman network community detection methods and the topics assigned to each community to see if the underlying structure of association between topics is robust to the use of different community detection methods. The results indicate that it may be possible to map discourse communities through topic modeling and network analysis. However, the comparison of the algorithms is inconclusive. This work contributes to our understanding of how computational social science methods can be used to measure and analyze extremist use of the Internet at scale.