Showing 1–1 of 1 results for author: Grenon-Godbout, N
-
ToxBuster: In-game Chat Toxicity Buster with BERT
Authors:
Zachary Yang,
Yasmine Maricar,
MohammadReza Davari,
Nicolas Grenon-Godbout,
Reihaneh Rabbany
Abstract:
Detecting toxicity in online spaces is challenging and an ever more pressing problem given the increase in social media and gaming consumption. We introduce ToxBuster, a simple and scalable model trained on a relatively large dataset of 194k lines of game chat from Rainbow Six Siege and For Honor, carefully annotated for different kinds of toxicity. Compared to the existing state-of-the-art, ToxBu…
▽ More
Detecting toxicity in online spaces is challenging and an ever more pressing problem given the increase in social media and gaming consumption. We introduce ToxBuster, a simple and scalable model trained on a relatively large dataset of 194k lines of game chat from Rainbow Six Siege and For Honor, carefully annotated for different kinds of toxicity. Compared to the existing state-of-the-art, ToxBuster achieves 82.95% (+7) in precision and 83.56% (+57) in recall. This improvement is obtained by leveraging past chat history and metadata. We also study the implication towards real-time and post-game moderation as well as the model transferability from one game to another.
△ Less
Submitted 21 May, 2023;
originally announced May 2023.
