Similar to NoisyHate: Hate Speech Detection with Text Perturbations

Ranked by cosine similarity over each dataset's title, summary and tags. Similarity is computed from descriptions, not from the data itself.

HOT Speech: Hateful, Offensive, Toxic Speech Dataset 86%

Hate Speech and Offensive Language Dataset 85%

TextDetox Multilingual Toxicity Dataset (All Languages) 84%

TextDetox Multilingual Toxicity — English 84%

TextDetox Multilingual Toxicity — Russian 83%

TextDetox Multilingual Toxicity — Hindglish 83%

TextDetox Multilingual Toxicity — German 83%

OffensEval 2020 — English 83%

TextDetox Multilingual Toxicity — Arabic 82%

TextDetox Multilingual Toxicity — Ukrainian 82%