Towards Mathematical Detection of Fake News using Graph Theory

Vyacheslav Tykhonov, Juliette Vion-Dury, Andrea Scharnhorst, Yves Rozenholc

Research output: Contribution to conferencePosterScientific

Abstract

A multidisciplinary, international team in data science, computer science and comparative literature has been working for four years on an open science, reproducible science and FAIR data project: the Covid-19 Museum, now Now.Museum, which aims at providing access to digital artefacts of the present in the future.The Now.Museum has built up a back-up, organized using Dataverse, of the public sections of European newspapers in their original languages since 2020, into a dynamic, viewable chronological corpus with origin, title, date, etc. which it is possible to query in natural language.From this unique database, we aim at testing the hypothesis: rhetoric or fake news could be detectable using graph theory and statistical testing. Our hypothesis is that a change in shape in the dynamic geometric representation, and the associated change in measurements in the graph, matrix and eigenvalues are detectable, categorizable and interpretable.

Conference

ConferenceDARIAH Annual Event 2024
Abbreviated titleDARIAH AE 2024
Country/TerritoryPortugal
CityLisbon
Period18/06/202421/06/2024
Internet address

Cite this