Analyzing a fake news authorship network

Abstract

This paper analyzes 246 fake news websites previously identified in three research projects. From this dataset, we extract a set of authors who have written for these sites in 2016, which we make publicly available. Applying a novel shared authorship construct, we analyze a network of fake news sites. This analysis shows a tight cluster of sites, with a trend of article reposting, wherein sites copy content from each other but preserve author bylines. We also show the most central authors, while associated with different sites, share common affiliations with a single site: Infowars.