Fake news detection from Bulgarian Facebook pages

Abstract

In recent years, we've been observers to the increasingly free news sharing both from reliable sources and from obviously false ones. When talking about social networks, the belief in free speech is applied in full force. Platforms such as Facebook, Twitter and Instagram give everyone the opportunity to present themselves as a "journalist"and nowadays you no longer have to have the necessary education in order to cover the news and grow an auditory, which will accept and spread them. This paper aims to propose a method for the detection of fake news coming from Bulgarian Facebook Pages that are presenting themselves as reliable and trustworthy sources. The models trained in this study are based on three machine learning algorithms - KNN, SVM and Logistic Regression. All of the features that are being taken into consideration are derived from the title of the news post as this the first thing the readers see and trust.