A Korean news summarization dataset of 27,400 (document, summary) pairs, crawled from
Naver News over a ten-day window in July 2022. It was originally built for a
Korean NLP hands-on lab and has been publicly hosted on the Hugging Face Hub since January 2023.
A technical report documenting the collection protocol, corpus statistics, contamination analysis, and
reproducible baselines is available on arXiv: arXiv:2607.20442.… See the full description on the dataset page:
https://huggingface.co/datasets/daekeun-ml/naver-news-summarization-ko.
