Lexicon generation for detecting fake news

With the digitization of media, an immense amount of news data has been generated by online sources, including mainstream media outlets as well as social networks. However, the ease of production and distribution resulted in circulation of fake news as well as credible, authentic news. The pervasive dissemination of fake news has extreme negative impacts on individuals and society. Therefore, fake news detection has recently become an emerging topic as an interdisciplinary research field that is attracting significant attention from many research disciplines, including social sciences and linguistics. In this study, we propose a method primarily based on lexicons including a scoring system to facilitate the detection of the fake news in Turkish. We contribute to the literature by collecting a novel, large scale, and credible dataset of Turkish news, and by constructing the first fake news detection lexicon for Turkish.