Datasets:

DANGDOCAO
/

GeneratingQuestions

license: mit
language:
  - vi
tags:
  - ag
  - t5
  - vit5
  - squad-format
  - vietnamese
  - education
  - nlp
pretty_name: vietnamese Question Generation
size_categories:
  - 10K<n<100K

HVU_QA

HVU_QA is an open-source Vietnamese Question–Context–Answer (QCA) corpus for building FAQ-style question generation systems in low-resource languages.
It was created using a fully automated pipeline combining web crawling, semantic tag-based extraction, and AI-assisted filtering to ensure high factual accuracy.

Dataset Description

Language: Vietnamese
Format: SQuAD-style JSON
Size: 30,000 QCA triples
Domains: Social services, labor law, administrative processes, and public service topics

Each data sample contains:

question: The generated or extracted question
context: The supporting passage
answer: The answer span within the context

Dataset Creation

Pipeline:

Selecting relevant QA websites from trusted sources
Automated crawling to collect raw QA webpages
Semantic tag-based extraction to get clean QCA triples
AI-assisted filtering to remove noisy or inconsistent samples

Annotation & Licensing:
All data are collected from public-domain Vietnamese government and service portals, released under CC BY 4.0.

Quality Evaluation

A fine-tuned VietAI/vit5-base model trained on HVU_QA was used to validate the dataset quality.

Automatic metrics:

Metric	Score
BLEU	90.61
Semantic similarity	97.0% (cosine ≥ 0.8)

Human evaluation (1–5 scale):

Aspect	Avg. Score
Grammaticality	4.58 / 5
Usefulness	4.29 / 5

These results confirm that HVU_QA is a high-quality resource for developing robust FAQ-style question generation models.

Setup

Step 1 — Clone repository (optional if local)

git clone https://huggingface.co/datasets/DANGDOCAO/GeneratingQuestions
cd GeneratingQuestions

Step 2 — Install dependencies

Minimal:

python -m pip install --upgrade pip
pip install datasets transformers sentencepiece safetensors

Recommended (for training & evaluation):

pip install datasets transformers sentencepiece safetensors accelerate evaluate sacrebleu rouge-score nltk tensorboard scikit-learn

Install PyTorch (choose CUDA/CPU from https://pytorch.org):

pip install torch

Step 3 — (Optional) Configure environment

pip install -U huggingface_hub
huggingface-cli login
git lfs install

Only needed if you want to push to Hub or download private models.

Step 4 — Download data

from datasets import load_dataset
ds = load_dataset("DANGDOCAO/GeneratingQuestions", split="train")
print(ds[0])

Or from local JSON:

ds = load_dataset("json", data_files="30ktrain.json", split="train")

Step 5 — Verify installation

python -c "import torch, transformers, datasets, sklearn, sentencepiece, safetensors; print('All installed OK!')"

Usage

Fine-tune model

python fine_tune_qg.py

This will:

Load data from 30ktrain.json
Fine-tune VietAI/vit5-base
Save model to t5-viet-qg-finetuned/

Generate questions

python generate_question.py

Example

Input passage:
Iced milk coffee (Cà phê sữa đá) is a famous drink in Vietnam.
Number of questions: 5

Output

What type of coffee is famous in Vietnam?
Why is iced milk coffee popular?
What ingredients are included in iced milk coffee?
Where does iced milk coffee originate from?
How is Vietnamese iced milk coffee prepared?

Adjustable parameters (generate_question.py):
top_k, top_p, temperature, no_repeat_ngram_size, repetition_penalty

Citation

@inproceedings{nguyen2025hvuqa,
  title={A Method to Build QA Corpora for Low-Resource Languages},
  author={Ha Nguyen-Tien and Phuc Le-Hong and Dang Do-Cao and Cuong Nguyen-Hung and Chung Mai-Van},
  booktitle={Proceedings of the International Conference on Knowledge and Systems Engineering (KSE)},
  year={2025}
}