UWaterloo researchers develop new method to detect hate speech on social media

The University of Waterloo says a group of researchers at the school have developed a program which will detect hate speech on social media platforms.

The University of Waterloo says a group of researchers at the school have developed a program which will detect hate speech on social media platforms.

The school says that Multi-Modal Discussion Transformer (mDT), which is currently working at an 88 per cent accuracy rate, will make life easier for those who are tasked with flagging hate speech.

“We really hope this technology can help reduce the emotional cost of having humans sift through hate speech manually,” said Liam Hebert, a Waterloo computer science PhD student and the first author of the study.

“We believe that by taking a community-centred approach in our applications of AI, we can help create safer online spaces for all.”

The school says the mDT can understand the relationship between text and images while also reasoning the greater context surrounding comments.

The program also reduces the number of false positives as it can deduce comments which have been incorrectly flagged as hate speech because they contain culturally sensitive language.

The school says that understanding allows mDT to be much more accurate than previous models which had not been able to understand some nuances of language.

“Context is very important when understanding hate speech,” Hebert explained.