wolof-sentiments-corpus
320,000 automatically labeled texts
Official site consultedAI: datasets
Last verified: 1 October 2026
Corpus of 320,609 Wolof texts (14.5 MB) with a binary sentiment label, 55.8% positive and 44.2% negative. Labels come from automatic processing with an English DistilBERT model, after deduplication. It belongs to a collection of African-language sentiment datasets.
Services
- 179,014 positive and 141,595 negative texts
- Training of multilingual sentiment models
Access requirements
Free download on Hugging Face, MIT license; labels are automatic, not human-verified.
Sources
- wolof-sentiments-corpus (new tab), verified on 1 October 2026Official site consulted