Elamur Research⌕
☰
home / papers / 2026 / Towards an Interoperable Corpus of Austrian Historical Newspapers: The case of PressMint-AT

Towards an Interoperable Corpus of Austrian Historical Newspapers: The case of PressMint-AT

Paper · 2026 · Tanja Wissik, Jona Hassenbach, Hannes Pirker et al.

Summary

PressMint-AT is the Austrian part of the CLARIN PressMint project, which builds interoperable, annotated corpora of European newspapers from around 1900. It focuses on the Wiener Abendpost (1863–1921), a Fraktur-printed supplement of the Wiener Zeitung. The authors address two problems. First, the existing ANNO full texts from the Austrian National Library are uncorrected and noisy. Second, ANNO has no metadata marking where the supplement sits inside each Wiener Zeitung issue.

To locate the supplement, they trained image classifiers that detect its title pages. To choose a transcription method, they built a manually corrected line-level ground truth and compared several approaches on it:

  • classic and neural OCR tools,
  • multimodal LLMs, run zero-shot on full page images with German and English prompts,
  • the raw Transkribus output and the ANNO texts.

Despite recent reports favouring LLMs, PERO OCR with its German Fraktur model performed best. Qualitative analysis supported this, showing for example that Transkribus often merged lines across columns. The chosen transcriptions will feed later steps such as POS tagging, NER, entity linking and TEI encoding. This makes OCR choice a key factor in the quality of the whole corpus.

Key points

  • Planned corpus size: more than 17,450 Wiener Abendpost issues, totalling over 90,300 pages.
  • Ground truth: 4 issues from 1914–1918 (20 pages). It was produced with Transkribus Text Titan I and then extensively corrected by hand, keeping hyphenation and abbreviations as printed. It is published in the GitHub repo acdh-oeaw/pressmint-OCR-AI-evaluation.
  • Page spotting: text-based detection failed because the ANNO OCR was too poor, so the authors used two binary classifiers instead. These were fine-tuned from a pretrained ResNet-18 to recognise Abendpost and Amtsblatt title pages, with training data collected through a browser plugin on ANNO. Error was under 10% with 50 positive examples and reached 0% with 500.
  • Systems compared:
    • Dedicated OCR engines: PERO OCR, dots.ocr, Churro OCR, and Google's Cloud Vision OCR.
    • LLMs: Claude Sonnet 4, DeepSeek, GPT-4o, Gemini.
    • Existing outputs: raw Transkribus and ANNO.
  • Result: PERO OCR was judged the most suitable system for PressMint-AT.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🔬Austrian Centre for Digital HumanitiesVienna · Austria3 works · 16 citations🏛️Austrian Academy of SciencesVienna · Austria2 works

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 worksLiterature and Literary Theory · 541 works

Cite

Cite as Tanja Wissik, Jona Hassenbach, Hannes Pirker, Claudia Resch and Stefan Resch. 2026. Towards an Interoperable Corpus of Austrian Historical Newspapers: The case of PressMint-AT. Proceedings of the First Workshop on Creating Interoperable Corpora of Historical Newspapers.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/4ionpc6w2rmt

@inproceedings{wissik2026interoperable,
  title = {{Towards an Interoperable Corpus of Austrian Historical Newspapers: The case of PressMint-AT}},
  author = {Tanja Wissik, Jona Hassenbach, Hannes Pirker, Claudia Resch and Stefan Resch},
  booktitle = {Proceedings of the First Workshop on Creating Interoperable Corpora of Historical Newspapers},
  year = {2026},
  doi = {10.63317/4ionpc6w2rmt},
  url = {https://doi.org/10.63317/4ionpc6w2rmt}
}

References

24 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. history-reconstructing-esperanto-newspaper-excerpts-from-the-hachette-collection-1898-1915/. find
  2. Ratcliff, John W. and Metzener, D. (1988). "Pattern Matching: The Gestalt Approach". Dr. Dobb's Journal (46). find
  3. READ-COOP SCE. (n.d.) Transkribus. Innsbruck: READ-COOP SCE, [30.07.2025]. https://transkribus.eu/. ↗
  4. Resch, C. (2023). Volltextoptimierung für die historische Wiener Zeitung Mit einem Anwendungsszenario aus der germanistischen Sprachgeschichte. In Bunout, Ehrmann, M., Clavert, F. (Eds.), Digitised Newspapers – A New Eldorado for Historians? Reflections on Tools, Methods and Epistemology. Berlin, Bo… — [doi:10.1515/9783110729214](https://doi.org/10.1515/9783110729214) find
  5. Resch, C. and Kampkaspar D. (2019). DIGITARIUM – Unlocking the Treasure Trove of 18th-Century Newspapers for Digital Times. In: Wallnig, T, Romberg M. and Weis J. (Eds.): Digital Eighteenth Century: Central European Perspectives. Wien / Köln / Weimar: Böhlau Verlag, 49-64. find
  6. Semnani, S., Han Zhang, H., Xinyan He, X., Tekgurler, M. and Lam, M. (2025). CHURRO: Making History Readable with an Open-Weight Large Vision-Language Model for High-Accuracy, Low-Cost Historical Text Recognition. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Process… find
  7. Kodym, O. and Hradiš, M. (2021). Page Layout Analysis System for Unconstrained Historic Documents. ICDAR, https://arxiv.org/abs/2102.11838. — [arXiv:2102.11838](https://arxiv.org/abs/2102.11838) ↗
  8. Greif, G., Griesshaber, N. and Greif, R. (2025). Multimodal LLMs for OCR, OCR Post-Correction, and Named Entity Recognition in Historical Documents. https://arxiv.org/abs/2504.00414 — [arXiv:2504.00414](https://arxiv.org/abs/2504.00414) ↗
  9. Venglarova, K., Adam, R., Mölzer, W., Balasubramanian, S., Kleinert, J., Füllsack, M. and Vogeler, G. (2024). Who Advertises in Newspapers? Data Criticism in Mining Historical Job Ads. In: Proceedings of the Computational Humanities Research Conference 2024. find
  10. Aarhus, Denmark. CEUR. 2024. 788-801. find
  11. Boros, E., Ehrmann, M., Romanello, M., Najem-Meyer, S. and Kaplan, F. (2024). Postcorrection of historical text transcripts with large language models: An exploratory study. In Proceedings of the 8th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanitie… find
  12. Crosilla, G., Klic, L. and Colavizza, G. (2025). Benchmarking large language models for handwritten text recognition. https://arxiv.org/pdf/2503.15195. — [arXiv:2503.15195](https://arxiv.org/abs/2503.15195) ↗
  13. Ehrmann, M, Bunout, E. and Clavert, F. (2023). "Digitised Historical Newspapers: A Changing Research Landscape". Digitised Newspapers – A New Eldorado for Historians? Reflections on Tools, Methods and Epistemology, edited by Estelle Bunout, Maud Ehrmann and Frédéric Clavert, Berlin, Boston: De Gruyt… — [doi:10.1515/9783110729214-001](https://doi.org/10.1515/9783110729214-001) find
  14. Fišer, D. and Lenardič, J. (2018). CLARIN Resources Families / Newspaper Corpora. https://www.clarin.eu/resource-families/newspaper-corpora. ↗
  15. Humphries, M., Leddy, L. C., Downton, Q., Legace, M., McConnell, J., Murray, I. and Spence, E. (2024). Unlocking the archives: Using large language models to transcribe handwritten historical documents. find
  16. Kmainasi, M.B., Khan, R., Shahroor, A.E., Bendou, B., Hasanain, M. and Alam, F. (2025). Native vs Non-native Language Prompting: A Comparative Analysis. In: Barhamgi, M., Wang, H., Wang, X. (eds) Web Information Systems Engineering – WISE 2024. find
  17. WISE 2024. Lecture Notes in Computer Science, vol 15440. Singapore: Springer. https://doi.org/10.1007/978-981-96-0576-7_30. — [doi:10.1007/978-981-96-0576-7_30](https://doi.org/10.1007/978-981-96-0576-7_30) ↗
  18. Kim, S., Baudru, J., Ryckbosch, W., Bersini, H. and Ginis, V. (2025). Early evidence of how llms outperform traditional systems on ocr/htr tasks for historical records. https://arxiv.org/abs/2501.11623. — [arXiv:2501.11623](https://arxiv.org/abs/2501.11623) ↗
  19. Li, L. (2024). Handwriting Recognition in Historical Documents with Multimodal LLM. In CHR 2024: Computational Humanities Research Conference, December 4–6, 2024, Aarhus, Denmark https://arxiv.org/html/2410.24034v1. ↗
  20. Li, Y., Yang, G., Liu, H., Wang, B. and Zhang, C. (2025). dots.ocr: Multilingual document layout parsing in a single vision-language model. arXiv. https://arxiv.org/abs/2512.02498. — [arXiv:2512.02498](https://arxiv.org/abs/2512.02498) ↗
  21. Mayer, S. (2024). Reviving History: Reconstructing Esperanto Newspaper Excerpts from the Hachette Collection (1898-1915). ESF Connected https://esfconnected.org/2024/10/14/reviving-\#\# 10. Language Resource References Couranten Corpus (version 2.0) (2025) [Online Service]. Available at the Dutch Lang… ↗
  22. Dobranić, F., Evkovski, B. and Ljubešić, N. (2024). A Lightweight Approach to a Giga-Corpus of Historical Periodicals: The Story of a Slovenian Historical Newspaper Collection. find
  23. Resch, C. and Kampkaspar, D. (Eds) (2020). Wienerisches DIGITARIUM - Digitale Ausgabe des "Wien[n]erischen Diarium" (322 Ausgaben im Volltext). Resch, C. and Kampkaspar, D. (Eds) (2022). German Fraktur 18th Century – WrDiarium_M9. https://www.transkribus.org/models/german-fraktur-18th-century. ↗
  24. Weil, S. (2021). Tesseract OCR models for historic prints based on Latin script. Zenodo. https://doi.org/10.5281/zenodo.10125246. — [doi:10.5281/zenodo.10125246](https://doi.org/10.5281/zenodo.10125246) ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wissik-2026-interoperable-corpus-austrian-historical-newspapers/