Russia-Ukraine Conflict: A Text Mining Approach through Twitter

Abstract

The focus of this study is to use social media to investigate the Russia-Ukraine conflict. With the assent of the Russian parliament, Russian President Vladimir Putin proclaimed that they will begin invading Ukraine on February 24, 2022. During the Russia-Ukraine conflict, social media, particularly Twitter, has been heavily used. For that reason, it becomes to strong tool for handling processes during the conflict such as political decision making, organizing humanitarian activities, and proving assistance for victims. As a result, social media becomes the most up-to-date, comprehensive, and large information source for current scenario analysis. A total of 65412 tweets are gathered as a dataset for analysis in the proposed study between February 24 and April 5. Then, for each tweet, a topic modeling method called Latent Dirichlet Allocation (LDA) is used to collect significant topics and their probabilities considering each tweets. Then, using the specified probabilities, Fuzzy c-means is utilized to generate clusters for the entire document. Finally, seven unique clusters have been gathered for processing. N-grams and network analysis are used to examine each resulting cluster for a better understanding. As a result of this study, worldwide public opinion, current situation of civilians, course of the conflict, humanitarian issues during the Russia-Ukraine conflict are extracted.