Dataset Card for HuggingFace-CJK-Metadata

Dataset Summary

This dataset provides structured metadata and documentation extracted from the top 700 most downloaded datasets per language on the Hugging Face Hub for Chinese (zh), Japanese (ja), Korean (ko), and English (en, as a reference). The collection includes both high-level metadata (e.g., size, license, task type) and raw dataset card contents, enabling large-scale, cross-linguistic analysis of data curation, documentation quality, and cultural development patterns in East Asian NLP communities. All download statistics were recorded on January 28, 2025, and the dataset includes a total of 3,300+ entries spanning metadata and documentation fields.

Supported Tasks

This dataset is intended for:

• Dataset ecosystem analysis
• Meta-evaluation of documentation quality
• Cultural and institutional analysis of NLP practices across languages
• Visualization and benchmarking of dataset trends

Languages

• zh – Chinese
• ja – Japanese
• ko – Korean
• en – English (reference baseline)

---

Dataset Structure

The dataset contains two main components: structured metadata and full dataset card content, each organized by language (English, Chinese, Japanese, Korean).

huggingface-cjk-metadata/
└── data/
├── dataset_card/
│ ├── dataset_cards_en.csv
│ ├── dataset_cards_ko.csv
│ ├── dataset_cards_ja.csv
│ └── dataset_cards_zh.csv
└── dataset_meta/
├── dataset_meta_en.csv
├── dataset_meta_ko.csv
├── dataset_meta_ja.csv
└── dataset_meta_zh.csv

📘 dataset_meta files

Each row corresponds to a Hugging Face dataset and includes structured metadata fields:

Field	Description
id	Hugging Face dataset ID (e.g., skt/kogpt2)
author	Dataset creator (user or organization)
created_at	Timestamp when the dataset repo was created
lastModified	Timestamp of the latest commit
sha	Git commit SHA
downloads_30	Number of downloads in the past 30 days
downloads_alltime	Total number of downloads
likes	Number of likes on the dataset page
tags	Associated tags
tasks	NLP tasks associated with the dataset
description	Short dataset summary
citation	Citation information
languages	Languages covered (e.g., ko, en, zh)
language_category	One of: monolingual, en-paired, multilingual
size_categories	Estimated dataset size (e.g., 10K<n<100K)
paperswithcode_id	Linked PapersWithCode ID (if any)
private	Boolean indicating if repo is private
gated	Boolean for gated access
disabled	Boolean for deactivated datasets
license	License name (e.g., apache-2.0, cc-by-nc-4.0)
arxiv_id	arXiv paper ID (if applicable)
url	Hugging Face dataset URL
task_ids	Internal Hugging Face task identifiers

📄 dataset_card files

Each file contains raw Hugging Face dataset card contents in two fields:

Field	Description
dataset_id	Dataset identifier (same as id in meta)
yaml_metadata	Structured YAML block from the top of the README
markdown_content	The full free-text markdown body of the dataset card

These cards enable deeper qualitative analyses of documentation quality, structure, and cultural content across languages.

---

Repository

All scraping code and analysis notebooks can be found at:
👉 GitHub: https://github.com/Dasol-Choi/cjk-huggingface-analysis

Citation

@misc{choi2025languagedataleftbehind,
      title={No Language Data Left Behind: A Comparative Study of CJK Language Datasets in the Hugging Face Ecosystem}, 
      author={Dasol Choi and Woomyoung Park and Youngsook Song},
      year={2025},
      eprint={2507.04329},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2507.04329}, 
}

Contact

• dasolchoi@yonsei.ac.kr