vanloc1808/buddhist-scholar-test-set

Vietnamese Buddhist Scholar Test Set

Dataset Description

This dataset contains 1008 Vietnamese question-answer pairs focused on Buddhist teachings and literature. The dataset was created to evaluate chatbots' knowledge and understanding of Buddhist concepts, particularly for Vietnamese-speaking users.

Dataset Details

Dataset Summary

• Language: Vietnamese
• Task: Question Answering, Chatbot Evaluation
• Domain: Buddhism, Religious Studies
• Size: 1008 question-answer pairs
• Format: JSON with "question" and "answer" fields

Dataset Structure

Each entry contains:

• question: A question in Vietnamese about Buddhist teachings
• answer: The corresponding answer extracted from Buddhist texts

Statistics

• Total Q&A pairs: 1008
• Average question length: 54.23 characters
• Average answer length: 75.18 characters
• Question length range: 15 - 121 characters
• Answer length range: 2 - 417 characters

Sample Data

[
  {
    "question": "Quyển Ba của tác phẩm có tên là gì?",
    "answer": "Quyển Ba: Bàn rộng để khơi mở và củng cố niềm tin"
  },
  {
    "question": "Bài văn khuyên tu hành nằm ở trang bao nhiêu?",
    "answer": "139"
  },
  {
    "question": "Theo đoạn văn, điều gì được cho là nặng nề hơn cả tham dục?",
    "answer": "Chướng ngại của lý còn nặng nề hơn cả tham dục"
  }
]

Data Sources

The questions and answers were generated from Vietnamese Buddhist texts and literature, including:

• Classical Buddhist scriptures translated into Vietnamese
• Buddhist scholarly works
• Religious teachings and commentaries

Intended Use

Primary Use Cases

1. Chatbot Evaluation: Test Vietnamese chatbots' understanding of Buddhist concepts
2. Question Answering Models: Train or evaluate QA models on religious/cultural content
3. Educational Tools: Develop learning applications for Buddhist studies
4. Cultural AI: Improve AI systems' understanding of Vietnamese Buddhist culture

Considerations

• The dataset focuses specifically on Buddhist teachings and may not be suitable for general knowledge evaluation
• Answers are based on traditional Buddhist texts and interpretations
• Users should be aware of the religious and cultural context when using this dataset

Data Collection Process

The dataset was created by:

1. Processing Vietnamese Buddhist texts and literature
2. Extracting relevant passages using document processing tools
3. Generating question-answer pairs using AI models
4. Manual review and quality control
5. Formatting and validation

Dataset Creation

Created: 2025-07-23 Creator: vanloc1808 Processing Tools: Azure AI Inference, Docling, Custom scripts

Loading the Dataset

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("vanloc1808/buddhist-scholar-test-set")

# Access the data
for item in dataset['train']:
    print(f"Q: {item['question']}")
    print(f"A: {item['answer']}")
    print("---")

Evaluation Metrics

When using this dataset for evaluation, consider:

• Semantic Similarity: How well answers capture the meaning of reference answers
• Cultural Accuracy: Correctness within Buddhist and Vietnamese cultural context
• Language Quality: Fluency and naturalness of Vietnamese responses
• Factual Correctness: Accuracy of Buddhist teachings and concepts

Limitations

• Limited to Vietnamese language and Buddhist domain
• May contain biases present in source materials
• Answers reflect traditional interpretations and may not cover modern perspectives
• Quality may vary across different topics within Buddhism

Citation

If you use this dataset in your research, please cite:

@dataset{buddhist_scholar_vietnamese_2024,
  title={Vietnamese Buddhist Scholar Test Set},
  author={vanloc1808},
  year={2024},
  url={https://huggingface.co/datasets/vanloc1808/buddhist-scholar-test-set}
}

License

This dataset is released under the MIT License. See LICENSE for more details.

Contact

For questions or issues regarding this dataset, please contact the creator through Hugging Face.