New Paper: ‘Beyond Fact-Checking: Network Analysis Tools for Monitoring Disinformation in Social Media’

In the paper Beyond Fact-Checking: Network Analysis Tools for Monitoring Disinformation in Social Media’ (paper-Beyond-Fact-checking-.pdf), Stefano Guarino (Institute for Applied Computing, National Research Council & Data Lab, Luiss “Guido Carli” University), Noemi Trino (Data Lab, Luiss “Guido Carli” University), Alessandro Chessa (Data Lab, Luiss “Guido Carli” University & Linklab) and Gianni Riotta (Data Lab, Luiss “Guido Carli” University) presented an integrated toolbox for monitoring social disinformation, conceived as part of the H2020 Social Observatory for Disinformation and Social Media Analysis.

The DisInfoNet Toolbox builds on well-established techniques for text and graph mining to provide a wide spectrum of users instruments for quantifying the prevalence of disinformation and understanding its dynamics of diffusion on social media.

The authors presented a case study analysis focused on the 2016 Italian constitutional referendum, where in the natural bipolar political structure of the debate helps in reducing one of the most frequent problem in opinion detection on social media, related to the identification of all possible political orientations (associated to communities).

As the authors explain, they resorted to retweets in order to analyze accounts and their interactions according to their possible political orientation. The combined analysis of political communities and network clustering and centrality shows how the referendum caused a clear segregation by political alignment, configuring the existence of different echo-chambers.

From a thematic point of view, news stories related to conspiracy theories and distrust with political elite were especially popular and traveled deeper than any other category of disinformation. As it is pinpointed ‘we found evidence of a correlation between users’ polarization and participation to disinformation campaigns, and by highlighting the primary actors of disinformation production and propagation we could manually tell apart public figures, activists and potential bots.’

The DisInfoNet Toolbox will soon be available online and extended in the next future.

Last but not least, as Stefano Guarino, Noemi Trino, Alessandro Chessa and Gianni Riotta mention ‘we believe that the state-of-the-art techniques for classification and network analysis embedded in the Toolbox will pave the way for future research in the area, crucial to the preservation of our public conversation and the future of our democracies.’

 You can download the paper Beyond Fact checking