For a Human-Centered AI

Hatemeter, a counter-narrative to fight cyberhate

July 26, 2019

An artificial intelligence platform to help NGOs monitor and curb the phenomenon of online Islamophobia

To systematize, increase and share knowledge on hate speech on the web and increase the efficiency and effectiveness of the work of non-governmental organizations and civil society organizations in preventing episodes of Islamophobia in Europe.

This is the goal of Hatemeter – Hate speech tool for monitoring, analyzing and tackling Anti-Muslim hatred, the project coordinated by the University of Trento, with the participation of Fondazione Bruno Kessler, Amnesty International Italia, Centre National de la Recherche Scientifique and Collectif Contre the Islamophobie en France – Association de Défense des Droits de l’Homme (France), Teesside University and Stop Hate UK (England), which has developed a platform capable on the one hand to automatically monitor and analyze online data and social media, on the other hand to produce possible answers and useful suggestions for the creation of online counter-narratives and specific awareness campaigns.

The project aims to provide technological support to NGOs in monitoring Islamophobia online,” Sara Tonelli, researcher at Fondazione Bruno Kessler explains. Operators working for these organizations actively monitor what is happening on the web and in some cases, according to their guidelines, chime in when conversations are turning violent. As you can guess, it’s a very demanding job and what we want to develop with Hatemeter is a tool that speeds up these processes“.

A platform was therefore created – tested and validated in Italy by Amnesty – that can monitor the web in real time and at the same time provide indications to NGOs on how and when to intervene in online threads.

We started from a massive analysis and research work. In particular, Twitter and Instagram were monitored using around twenty specific hashtags that allowed us to identify and analyze the contents conveyed on the network, their origin and their spread – the researcher goes on – The data analyzed provided us with a sadly important representation of the phenomenon of Islamophobia on the web, with over three million contents identified in a few months. These data were later on used for the creation of an artificial intelligence-based platform capable of suggesting possible answers to NGO operators, who can thus act on online chats trying to soften their tones. We could see – Sara Tonelli concluded – that targeted intervention in chats actually produces results, with many haters abandoning discussions and with a general drop in the hatred levels in chats“.

Hatemeter joins two other projects launched by Fondazione Bruno Kessler to stem violence and discrimination, not only online: DicodiNo, an initiative launched with schools on stereotypes, prejudices, discrimination and verbal violence against religious minorities, and the CREEP project, which addresses the issue of cyberbullying, suggesting prevention tools through the creation of new technologies.





The author/s