Logo Lanfrica

AraTox

Domaine:

natural language processing

Type de record:

dataset
Créateur:
Als
Hôte:
AraTox is a multi-dialect, multi-label Arabic dataset for toxicity detection. It contains annotated Arabic text from Gulf, Levantine, Nile Basin, North African, Yemeni, and Modern Standard Arabic (MSA). The dataset supports research in toxic language detection, multi-label classification, and benchmarking Arabic NLP models across dialects. { "Text": "شش أقرع", "Appearance": 1, "Cussing": 0, "Hatred": 0,