Impresso - Media Monitoring of the Past
Published September 29, 2025 | Version v1
Computational notebook Open

Impresso Datalab - News Agencies Recognition and Linking with Impresso BERT models

Authors/Creators

  • 1. ROR icon École Polytechnique Fédérale de Lausanne
  • 2. ROR icon Luxembourg Centre for Contemporary and Digital History

Description

The Impresso project processes large collections of media archives and develops the Impresso Web App and Datalab for their exploration.

In the context of the Impresso Datalab, this notebook demonstrates how to find mentions of news agencies in historical newspaper articles by loading a pre-trained model from Hugging Face. The model, part of the Impresso Project, is designed to recognize agency names such as Reuters, AFP, Havas, or DPA in texts written in French and German.

Files

newsagency-processing_ImpressoHF.ipynb

Files (26.7 kB)

Name Size Download all
md5:53c5a413e18cc55b67be2b56a29cd237
26.7 kB Preview Download

Additional details

Funding

Swiss National Science Foundation
Impresso - Media Monitoring of the Past II. Beyond Borders: Connecting Historical Newspapers and Radio. 213585

Software

Repository URL
https://github.com/impresso/impresso-datalab-notebooks/blob/main/annotate/newsagency-processing_ImpressoHF.ipynb
Programming language
Python
Development Status
Active