Year of Graduation
Cross-country Analysis of Political Discourse on Social media: Comparative Analysis of Topic Modeling Algorithms
In this paper we conduct a cross-country political discourse analysis using multilingual topic modeling algorithms and comparing their results. Our methodology allows us to process multilingual cross-country corpora in order to identify current trending politics concerned topics for US, France, and Russia. We are able to do so by applying extensions of LDA for multilingual data. The results of the models are validated with parallel (TEDx transcripts) and non-parallel (politics related tweets) data. As a result of the study, we propose a methodology for conducting comparative cross-lingual discourse analysis and rank chosen multilingual model by the quality of the topics that they have produced.