Machine-Learning Method Detects Hate Speech on Social Media Platforms

A team of researchers at the University of Waterloo has developed an innovative machine-learning method that significantly improves hate speech detection on social media platforms. The method, known as the Multi-Modal Discussion Transformer (mDT), achieves an impressive 88% accuracy, surpassing previous approaches and potentially saving employees countless hours of emotionally taxing work.

[block_2]

Understanding Context and Reducing False Positives