Stance Detection in the Context of Fake News

Online Social Networks (OSNs) are overwhelmed by the daily volume of news reported by humans from all around the world. The diffusion of information can start from any OSNs user and spread rapidly. The process of news fact-checking is very labor and resource intensive. Researchers are looking for machine-learning approaches to automate the detection of fake news. Toward such goal, this paper focused on stance detection of content producers, whether they are in favor or against the content subject. In this study, our goal is to develop and evaluate advanced text-mining models that are enhanced with the cosine similarity between the headline and body of news articles to predict users' stance from the articles' content. We specifically aim to explore if the cosine distance feature will enhance the models' prediction performance.