visolex/

VN-HSD

	Dataset Card for ViSoLex‑HSD




	
	1. Dataset Summary

ViSoLex‑HSD is a unified Vietnamese hate‐speech detection corpus, combining three benchmark datasets:

ViHSD (Son et al., 2021): 33K comments labeled CLEAN, OFFENSIVE, or HATE UIT‑ViCTSD (Nguyen et al., 2020): 10K comments annotated for TOXIC (mapped to HATE) or CLEAN ViHOS (Hoang et al., 2023): span‐level labels aggregated into comment‐level HATE/CLEAN

After renaming and mapping labels (0 = CLEAN; 1 = OFFENSIVE; 2 =… See the full description on the dataset page: https://huggingface.co/datasets/visolex/VN-HSD.