Identifying Partisan Slant in News Articles and Twitter during the Ukraine-Russia Crisis

• In Event: Methodological Challenges of Hateful and Problematic Content Detection Online

Abstract

This talk reflects on the methodology of the study on the flow of partisan news frames between mainstream and social media during mass crises. We specify how the language used in those frames may allow to infer political slant and partisanship of a source. We study the lingual choices for political agenda setting by analyzing a dataset of over 40M tweets and over 4M news articles from the Euromaidan protests in Ukraine during 2013-2014 and the post-Euromaidan conflict in eastern Ukraine and Crimea. Using word embedding and machine learning approaches, we design a natural language processing algorithm to analyze at scale the linguistic markers which point to a particular political leaning or partisan slant in online media. While we reach a high level of accuracy, in this talk we focus on our research design, the challenges we faced and methodological decisions that have allowed us to build a successful classifier.

Authors

• Dmytro Karamshuk, Skyscanner / King's College London

• Tetyana Lokot, Dublin City University

• Olexandr Pryymak, Facebook

• Nishanth Sastry, King's College London