You Are What You Tweet: Profiling Users by Past Tweets to Improve Hate Speech Detection

Hate speech detection research has predominantly focused on purely\ncontent-based methods, without exploiting any additional context. We briefly\ncritique pros and cons of this task formulation. We then investigate profiling\nusers by their past utterances as an informative prior to better predict\nwhether new utterances constitute hate speech. To evaluate this, we augment\nthree Twitter hate speech datasets with additional timeline data, then embed\nthis additional context into a strong baseline model. Promising results suggest\nmerit for further investigation, though analysis is complicated by differences\nin annotation schemes and processes, as well as Twitter API limitations and\ndata sharing policies.\n