Topic modeling as a tool for analyzing tweets: A case study of the Russia-Ukraine war on Arabic social media

Abstract

The transformation of information dissemination and public discourse was heavily influenced by the rapid rise of social media platforms, especially Twitter, and this was particularly evident in times of conflict and war. The huge explosion of social media usage has, in turn, created a large amount of data that can be analyzed through different methods such as text mining and natural language processing. In this paper, we employ topic modeling to extract and analyze the topics of discussion around the Russian-Ukraine conflict in the Middle East. The analysis is facilitated by collecting dialectical Arabic tweets specifically containing terms relating to the Ukraine-Russia conflict. Investigation and comprehension of the dominant themes and views carried by the discussion are held through comparative research of two important topic modeling tools: BERTopic and LDA. From our findings, the influence of social media in forming public opinion, the spread of information, and the creation of a discourse regarding the Russian-Ukrainian war in the Middle East becomes visible. The topic modeling in our study presents the broad spectrum of views and themes emerging through social media discourse. This comprehensive perspective assists in the understanding of the intricate complexities surrounding this geopolitical conflict and offers a deep dive into the multifaceted nature of the matter.