What is Real or Fake?-Machine Learning Approaches for Rumor Verification using Stance CIassification

In a recent survey, over half (54%) of a global sample agree or strongly agree that they are concerned about what is real and fake when thinking about online news. Rumors are spreading all the time and affect people's perceptions and behavior. In this paper, we apply several machine learning approaches, from simple supervised algorithms to deep learning models, for the stance classification and rumor verification tasks; and evaluate the impact of the stance information in the performance of rumor veracity evaluation. According to the results, the traditional machine learning algorithms presented better performance than deep learning models, in both tasks, and the information of stance (deny or query) do not improve the results of the rumor verification task.