A Text Mining Approach for Evaluating Event Credibility on Twitter

Users on social media platforms tend to readily believe the contents of posts related to the events. However, some of these events might be fake or incredible. The spread of such events takes the form of rumors that have the potential for affecting negatively the individuals and society. To this end, in this paper we develop a text mining approach for automatic evaluation of events on social networks. We consider Twitter as a case study. Given a set of popular Twitter events along with different credibility ratings assigned manually by human annotators (i.e., crowdsourcing), we study the problem of automatically assessing the credibility of such events. The conducted experiment in this paper using events extracted from the CREDBANK dataset; a corpus of tweets annotated with human credibility judgements shows that our approach is promising. It achieves an automatic credibility assessment of events of 82.86% with Decision Tree (DT) classifier.