800,000 digitized pages of Luxembourgish historical newspapers and monographs converted into a structured text dataset. The collection includes full-text extractions and associated metadata for individual articles, sections, and advertisements from public domain sources.
Use Cases
- Train historical language models for Luxembourgish using the extracted full-text content
- Perform longitudinal sentiment analysis on public discourse using the article and section metadata
- Analyze historical marketing trends by categorizing and processing the advertisement entries
- Build a search index for historical Luxembourgish genealogy or events using the metadata and text fields
Strengths
- 800,000 digitized pages from the National Library of Luxembourg (BnL)
- Granular extraction of content types including articles, sections, and advertisements
- Includes full-text content and associated metadata for every extracted item
- Focuses exclusively on public domain newspapers and monographs