Hateful Symbols or Hateful People? Predictive Features for Hate Speech Detection on Twitter

TLDR

Hate speech, often racist or sexist, is common on social media, but its definition varies and detection is largely manual. The study aims to create a critical race theory–based annotation scheme and a dictionary of indicative words for a 16k‑tweet corpus to improve hate‑speech detection. The authors analyze how extra‑linguistic features combined with character n‑grams affect hate‑speech detection performance.

Abstract

Hate speech in the form of racist and sexist remarks are a common occurrence on social media. For that reason, many social media services address the problem of identifying hate speech, but the definition of hate speech varies markedly and is largely a manual effort. We provide a list of criteria founded in critical race theory, and use them to annotate a publicly available corpus of more than 16k tweets. We analyze the impact of various extra-linguistic features in conjunction with character n-grams for hate-speech detection. We also present a dictionary based the most indicative words in our data.