Three WeVerify teams @EUvsVirus Hackathon

By on May 3rd, 2020 in Events, News

From 24 to 26 April 2020, Europe joined forces to develop innovative solutions for coronavirus-related challenges in the EUvsVirus hackathon, the official and worldwide programming, coding and developments event to fight COVID-19, organised by the EU Commission.

The challenges were divided into several topical areas. The participating WeVerify teams focused on the area Social & Political Cohesion – Mitigating fake news spreading, where our experience and knowledge about disinformation helped us understand many of the underlying problems, and where we felt most suited to suggest a variety of solutions.

We formed three teams, with the overarching aim to enhance WeVerify tools and technology and leverage their synergies to help tackle the COVID-19 disinfodemic.

Our teams worked on three complementary challenges:

  • Team 1 (led by Ontotext): Database of known fakes and verification assistant;
  • Team 2 (led by Sheffield University): Disinformation annotation and machine learning-based categorisation
  • Team 3 (led by EUDL and AFP): COVID-19 Disinformation detection through cross-social network search and analysis

Team 1 worked on enhancing the WeVerify Database of Known Fakes (DBKF) with COVID-19 debunks published by IFCN-certified fact-checkers, and then providing a browser-based verification assistant with which users (e.g. fact-checkers, citizens) can easily provide the URL or content of a post and find any matching debunks. A video of the results is available here.

Example of automatic recommendation of relevant debunks
from the WeVerify database

Team 2 worked on automatic classification of COVID-19-related posts into new, distinct categories (e.g. origin, social distancing, government lockdown policies). The ability to do so automatically and reliably is paramount, as fact-checkers can then easily navigate prior debunks related only to the relevant topic (e.g. origin) in order to speed-up their work. Further details and a demo video are available in this dedicated blog post.

Examples of COVID-19 disinformation classification

Team 3 worked on enhancing the incipient Twitter social network analysis feature to be released in the InVID-WeVerify verification plugin (currently used by more than 25k journalists and fact-checkers worldwide). During the hackathon, the team enhanced it with new functionalities to reveal accounts and websites spreading COVID-19 related disinformation, as well as highlighting possible disinformation cross-postings around networks. Further details and a demo video are available here.

New Verification Plugin Functionality

All in all, a very successful event with lots of new insights and promising results. Our project work was enhanced significantly by this initiative, and we will follow up on many of the activities. Stay tuned!

Author: Kalina Bontcheva (University of Sheffield)
Editor: Jochen Spangenberg (Deutsche Welle)