Managing the Evaluation of OCR Quality in Flemish Newspaper Collections

The OCR quality of digitised historical newspapers can greatly impact their findability and usability. In 2021 the Flanders Heritage Libraries has worked together with meemoo, the Staatsbibliothek zu Berlin, the Swiss firm Odoma and ten other partner organisations in order to evaluate the current OCR quality of Flemish digitised newspaper collections, as well as their potential for improvement. This presentation gives insight into the coordination of this project. More concretely it will focus on the project workflow, the creation of a ground truth dataset, interpreting the results, the challe... Mehr ...

Verfasser: Denys, Montaine
Dokumenttyp: lecture
Erscheinungsdatum: 2023
Schlagwörter: Evaluation OCR Quality / Flemish Newspaper Collections
Sprache: Englisch
Permalink: https://search.fid-benelux.de/Record/base-26706168
Datenquelle: BASE; Originalkatalog
Powered By: BASE
Link(s) : https://zenodo.org/record/7572443

The OCR quality of digitised historical newspapers can greatly impact their findability and usability. In 2021 the Flanders Heritage Libraries has worked together with meemoo, the Staatsbibliothek zu Berlin, the Swiss firm Odoma and ten other partner organisations in order to evaluate the current OCR quality of Flemish digitised newspaper collections, as well as their potential for improvement. This presentation gives insight into the coordination of this project. More concretely it will focus on the project workflow, the creation of a ground truth dataset, interpreting the results, the challenges faced and the lessons learned. ; This OCR evaluation is part of Nieuwe Tijdingen, a three-year initiative (2021-2023) by the Flanders Heritage Libraries in collaboration with meemoo and more than 50 partner organizations. Nieuwe Tijdingen lays the groundwork for a large-scale programme for the digitization, presentation and preservation of Flemish newspaper collections.