Constructing a Taxonomy of Implicit Hate Speech Grounded in Social Theory with Diyi Yang and David Muchlinksi

David Muchlinski

Posted July 18, 2022

External Article: Diaries of Social Data Research

David Muchlinski, assistant professor in the Sam Nunn School of International Affairs, joined the Diaries of Social Data Research podcast to discuss a paper he co-authored titled "Latent Hatred: A Benchmark for Understanding Implicit Hate Speech." Muchlinksi, alongside co-author Diyi Yang of the School of Interactive Computing, discussed how their annotation process for the paper took two years and incorporated the human aspect of hate speech into it.

An excerpt:

What we understand about some of these groups — especially some of these hate groups — is they're not stupid.... They understand how content moderation policies work, and they understand how to game them. If you continue  doing content moderation based off of explicit references, you will eventually allow hate speech onto your platform in all sorts of disguises.