Article Dans Une Revue Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026) Année : 2026

The Patrologia Graeca Corpus: OCR, Annotation, and Open Release of Noisy Nineteenth-Century Polytonic Greek Editions

Résumé

We present the Patrologia Graeca Corpus, the first large-scale open OCR and linguistic resource for nineteenthcentury editions of Ancient Greek. The collection covers the remaining undigitized volumes of the Patrologia Graeca (PG), printed in complex bilingual (Greek-Latin) layouts and characterized by highly degraded polytonic Greek typography. Through a dedicated pipeline combining YOLO-based layout detection and CRNN-based text recognition, we achieve a character error rate (CER) of 1.05% and a word error rate (WER) of 4.69%, largely outperforming existing OCR systems for polytonic Greek. The resulting corpus contains around six million lemmatized and part-of-speech tagged tokens, aligned with full OCR and layout annotations. Beyond its philological value, this corpus establishes a new benchmark for OCR on noisy polytonic Greek and provides training material for future models, including LLMs.

Fichier principal
Vignette du fichier
PG-final.pdf (4.32 Mo) Télécharger le fichier

Dates et versions

hal-05544910 , version 1 (10-03-2026)

Licence

Identifiants

Citer

Chahan Vidal-Gorène, Bastien Kindt. The Patrologia Graeca Corpus: OCR, Annotation, and Open Release of Noisy Nineteenth-Century Polytonic Greek Editions. Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026), 2026, pp.6519-6527. ⟨10.63317/2gvtaxwvjtc7⟩. ⟨hal-05544910⟩
193 Consultations
68 Téléchargements

Altmetric

Partager

  • More