NbAiLab/merged_npk_ndla_parallel_paragraphs

About the dataset

This is a dataset consiting of Norwegian Nynorsk and Norwegian Bokmål parallel texts. This dataset is built from two different sources: the Nynorsk Pressekontor (NPK) dataset (https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-47/, https://www.nb.no/sprakbanken/ressurskatalog/oai-nb-no-sbr-80/), and scraped texts from NDLA (https://ndla.no/).

Citation

@inproceedings{kummervold-etal-2026-building,
    title = "Building a One-Million-Pair {B}okm{\r{a}}l{--}{N}ynorsk Translation Corpus: A Quality-First Harvesting and Cleaning Pipeline",
    author = "Kummervold, Per E.  and
      Tollersrud, Thea  and
      Zanardi, Angelina",
    editor = "Piperidis, Stelios  and
      Bel, N{\'u}ria  and
      van den Heuvel, Henk  and
      Ide, Nancy  and
      Krek, Simon  and
      Toral, Antonio",
    booktitle = "Proceedings of the Fifteenth Language Resources and Evaluation Conference",
    month = may,
    year = "2026",
    address = "Palma de Mallorca, Spain",
    publisher = "ELRA Language Resource Association",
    url = "https://aclanthology.org/2026.lrec-1.674/",
    doi = "10.63317/4jidtc8558q6",
    pages = "8551--8555"
}