Pierre Voué

Tom De Smedt, Pierre Voué, Sylvia Jaki et al.

This report provides an engagement analysis of counternarratives against online toxicity. Between February 2020 and July 2021, we observed over 15 million toxic messages on social media identified by our fine-grained, multilingual detection AI. Over 1,000 dashboard users responded to toxic messages with combinations of visual memes, text, or AI-generated text, or they reported content. This leads to new, real-life insights on self-regulatory approaches for the mitigation of online hate.

Pierre Voué, Tom De Smedt, Guy De Pauw

We have collected over 30M messages from the publicly available /pol/ message boards on 4chan and 8chan, and compiled them into a model of toxic language use. The trained word embeddings (0.4GB) are released for free and may be useful for further study on toxic discourse or to boost hate speech detection systems: https://textgain.com/8chan.