Words are the Window to the Soul: Language-based User Representations for Fake News Detection

Abstract

Cognitive and social traits of individuals are reflected in language use. Moreover, individuals who are prone to spread fake news online often share common traits. Building on these ideas, we introduce a model that creates representations of individuals on social media based only on the language they produce, and use them to detect fake news. We show that language-based user representations are beneficial for this task. We also present an extended analysis of the language of fake news spreaders, showing that its main features are mostly domain independent and consistent across two English datasets. Finally, we exploit the relation between language use and connections in the social graph to assess the presence of the Echo Chamber effect in our data.

Investigating Online Community Engagement through Stancetaking

“…Further work has shown that communities with more distinct words better retain users (Zhang et al, 2017), and tend to be smaller, denser, more active, and have more local engagement (Lucy and Bamman, 2021). Similar findings have been shown for communities with distinct word senses (Del Tredici and Fernández, 2017;Lucy and Bamman, 2021). We extend computational sociolinguistic work on community identity and engagement patterns by motivating and investigating variation of a different kind: variation in higher level properties of stancetaking.…”

Entity-Aware Dual Co-Attention Network for Fake News Detection

“…For example, fake news is proven empirically to influence the 2016 U.S. presidential election (Bovet and Makse, 2019;Grinberg et al, 2019;Budak, 2019). Given the impact of false information, previous studies paid a lot of effort to detect it from different aspects, including (1) news content only (Santos et al, 2020;Kim and Ko, 2021), (2) the combination of news articles and social media replies Lu and Li, 2020), and (3) additional publisher/user information (Long et al, 2017;Yuan et al, 2020;Del Tredici and Fernández, 2020). In this work, we focus on using both news contents and social media replies, and further add external knowledge to enhance the model's ability to capture critical entities.…”