chcaa/eno-embs-old-news

Danish Historical Newspaper Articles Dataset

This dataset contains approximately 4.9 million Danish historical newspaper articles (1666–1850) with document embeddings, providing a comprehensive resource for studying Danish language, culture, and history through primary journalistic sources.

Dataset Details

Dataset Description

This dataset comprises digitized newspaper articles from Danish newspapers, featuring full-text content along with metadata including publication dates, newspaper titles, and automated category classifications. The collection spans 28 Danish historical newspapers and periodicals, published between 1666 and 1850 in Denmark-Norway.

The dataset includes document embeddings (pooled representations) created with the Old_News_Segmentation_SBERT_V0.1 model.

Key statistics:

• —Total articles: 4,898,084
• —Dataset size: ~33.7 GB (uncompressed), ~25.1 GB (download)
• —Language: Danish (da)
• —Format: Parquet files

Features:

• —text: Full article text content
• —date: Publication date
• —id: Unique article identifier
• —pwa: Probability-weighted attribute (float)
• —newspaper: Newspaper title
• —pooled: Pre-computed embeddings (list of floats) for semantic analysis
• —predicted_category: Automatically assigned content category, result of a classification task

• —Curated by: Alie Lassche (a.w.lassche@cas.au.dk) & Johan Heinsen (heinsen@dps.aau.dk)
• —Language(s): Danish (da)

Dataset Sources

• —Repository: GitHub
• —Paper: DOI

Uses

This dataset is suitable for:

• —Historical research: Analyzing trends, events, and discourse in Danish history through primary newspaper sources
• —Danish NLP: Training and evaluating language models, text classification systems, and information extraction tools
• —Computational journalism: Studying journalistic practices, media framing, and news coverage patterns
• —Cultural studies: Examining cultural narratives, social movements, and public discourse over time
• —Semantic search: Leveraging pre-computed embeddings for similarity search and article clustering
• —Topic modeling: Discovering themes and topics across time periods and newspapers
• —Named entity recognition: Identifying historical persons, places, and organizations
• —Temporal analysis: Tracking language change and concept evolution

Dataset Structure

The dataset contains a single train split with the following fields:

• —text (string): The full text of the newspaper article
• —date (string): Publication date of the article
• —id (string): Unique identifier for each article
• —pwa (float64): Probability-weighted attribute score
• —newspaper (string): Name of the source newspaper
• —pooled (list[float64]): Pre-computed embedding vector for semantic similarity tasks
• —predicted_category (string): Automatically assigned category label

Dataset Creation

Curation Rationale

This dataset was created to preserve and provide access to Danish historical newspapers in a structured, machine-readable format. The digitization and compilation of these articles serves multiple purposes:

1. 1.Preserving cultural heritage and making historical sources accessible
2. 2.Enabling large-scale computational analysis of Danish history and culture
3. 3.Supporting Danish language processing research with substantial historical data
4. 4.Facilitating interdisciplinary research combining computational methods and humanities scholarship

Source Data

Data Collection and Processing

The newspapers are kept in extraordinarily well-preserved collections housed by the national libraries of Denmark and Norway. The processing pipeline is described in the paper cited at the bottom of this dataset card.

Known Limitations

• —OCR errors: Text extracted from historical newspapers may contain errors due to print quality, paper degradation, or OCR limitations
• —Historical bias: Content reflects the perspectives, biases, and social attitudes of the time period, which may include prejudiced views
• —Coverage gaps: Digitization may not be comprehensive across all time periods or newspapers
• —Language evolution: Danish language, spelling, and orthography have changed over time
• —Selection bias: Available newspapers may not represent all regions, political viewpoints, or social classes equally

Citation

BibTeX:

@inproceedings{lassche-etal-2026-evaluating,
  title = {Evaluating Embedding Models on Danish Historical Newspapers: A Corpus and Benchmark Resource},
  author = {Lassche, Alie and Feldkamp, Pascale and Bizzoni, Yuri and Baunvig, Katrine and Nielbo, Kristoffer and Heinsen, Johan},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)},
  month = {May},
  year = {2026},
  pages = {3577--3589},
  address = {Palma, Mallorca, Spain},
  publisher = {European Language Resources Association (ELRA)},
  doi = {10.63317/2hzf85aou6dw},
  abstract = {We present an enriched dataset of almost five million Danish historical newspaper articles from the late seventeenth to nineteenth century, augmented with semantic embeddings and an annotated subset, to enable semi-automated classification as well as thematic and linguistic exploration. Through three historical benchmark tasks that evaluate the performance of Danish and multilingual embedding models on this historical Danish corpus, we discuss how the choice for an embedding model depends on the type of task, and enrich our corpus with embeddings from the overall best performing model. As a showcase experiment, we look at the distribution of article categories in the three subgenres that can be observed in the corpus. This experiment highlights the corpus and article-level embeddings’ potential for further exploration and analysis of the Danish historical mediascape. The resource is freely available for research use and aims to foster reproducible, data-driven studies of language and culture in the Danish nineteenth century.}
}

APA:

Lassche, A., Feldkamp, P., Bizzoni, Y., Baunvig, K., Nielbo, K., & Heinsen, J. (2026). Evaluating Embedding Models on Danish Historical Newspapers: A Corpus and Benchmark Resource. In Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026) (pp. 3577–3589). European Language Resources Association (ELRA). https://doi.org/10.63317/2hzf85aou6dw.

Dataset Card Author

Alie Lassche, Center for Humanities Computing, Aarhus University