README.md

---
pretty_name: ViHSD Explainable
language:
- vi

license: cc-by-4.0

multilinguality:
- monolingual

task_categories:
- text-classification
- text-generation

task_ids:
- hate-speech-detection
- explanation-generation

dataset_type: json

size_categories:
- 10K<n<100K

annotations_creators:
- machine-generated
- expert-generated

source_datasets:
- htdung167/ViHSD

modalities:
- text

formats:
- json

libraries:
- datasets
- pandas
- polars

tags:
- ViHSD
- Vietnamese
- Hate Speech Detection
- Toxicity Moderation
- Explainable AI
---

# vihsd-explainable

**ViHSD (extended)** — Vietnamese toxic/offensive dataset **with explanations & evidence**.

This dataset extends the original `htdung167/ViHSD` examples by adding:
- `explanation`: a short Vietnamese rationale (why the gold label applies)
- `evidence`: verbatim substrings extracted from the text that justify the label

## Splits
- `train`, `validation`, `test` — same splits as original ViHSD

## Schema (per sample)
- `text` (string): original sentence
- `label_id` (int): 0=CLEAN, 1=OFFENSIVE, 2=HATE
- `label` (string): human-readable label
- `explanation` (string): short Vietnamese rationale
- `evidence` (list[string]): verbatim substrings (empty for CLEAN)

## Provenance
- Original labels from `htdung167/ViHSD`.
- `explanation` and `evidence` were generated using an LLM (GPT-4o-mini on Azure) as a teacher, and reviewed/curated by authors.

## License
Dataset license: **CC-BY-4.0**.