Detection of Fake News on Social Media Using Network Science Approach

Abstract

This paper addresses the detection of fake news on social networks by combining complex networks and artificial intelligence techniques. Recent works have shown progress in solving the problem of detecting fake news using deep learning, which, in general, are penalized by the lack of interpretability and require large amounts of labeled data. In addition, to represent instances, solutions in the literature generally use textual characteristics, social relationships, and information related to engagement on social media. However, there are still gaps to be explored regarding the most relevant features of a fake post taking as a premise the interpretability of the solution. We propose modeling data through ego networks, extracting features from the underlying network, matching with textual features, and using traditional machine learning algorithms to detect and identify fake news on social networks. The experiments, carried out on Twitter data using the popular fake news dataset – FakeNewsNet, show the potential of the proposed approach from the perspectives of interpretability, precision and recall.