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

TL;DR

This paper introduces the Patrologia Graeca Corpus, a large-scale, open OCR and linguistic resource for nineteenth-century Greek editions, achieving high accuracy and providing valuable data for philology and future AI models.

Contribution

It presents a novel OCR pipeline for complex Greek texts, creating a comprehensive annotated corpus that sets new benchmarks and supports future research.

Findings

Achieved 1.05% CER and 4.69% WER in OCR accuracy.

Generated a corpus with six million lemmatized and POS-tagged tokens.

Established a new benchmark for OCR on noisy polytonic Greek texts.

Abstract

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.