RMIT at PAN-CLEF 2020: Profiling Fake News Spreaders on Twitter


Automatic detection of fake news in social media has become a prominent research topic due to its widespread, adverse effect on not only the society and public health but also on economy and democracy. The computational approaches towards automatic detection of fake news span from analyzing the source credibility, user credibility, as well as social network structure and the news content. However, the studies on user credibility in this context have largely focused on the frequency and times of engaging in a fake news propagation rather than profiling users based on the content of their tweets. In this paper, we approach this challenge through extracting linguistic and sentiment features from users’ tweet feed as well as retrieving the presence of emojis, hashtags and political bias in their tweets. These features are then used to classify users into spreaders or non-spreaders of fake news. Our proposed approach achieves 72%accuracy, being among the top-4 results obtained by systems for the task in theEnglish language.

CLEF 2020 Labs and Workshops, Notebook Papers