Title: Understanding Online Discourse through Social Context and Structured Pragmatics

Abstract: In an increasingly online world, understanding discourse on social media is akin to understanding our society. However, when it comes to social media discourse, a disproportionate amount of focus has been laid on content moderation via hate speech detection. In this talk, I will address a key limitation of this application: existing hate speech detection systems are riddled with racial biases introduced during annotation, which are reinforced and propagated by models trained on such data. I will present the inadequacies of current methods for debiasing hate speech detection and show how the subjectivity of this task design leads to debiasing failures. Next, I will focus on uncovering the origin of bias in toxic language detection. I will demonstrate how annotators’ demographics and beliefs influence their toxicity ratings, and how ignoring such societal context can lead to biased outcomes. Finally, I will present some ongoing work on understanding online discourse on homelessness, which presents some unique challenges. Overall, I will argue for the value of rethinking traditional the hate speech classification task, and the need for richer context and nuance when considering online discourse.