@inproceedings{danilova-aangenendt-2025-post,
title = "Post-{OCR} Correction of Historical {G}erman Periodicals using {LLM}s",
author = "Danilova, Vera and
Aangenendt, Gijs",
editor = "Holdt, {\v{S}}pela Arhar and
Ilinykh, Nikolai and
Scalvini, Barbara and
Bruton, Micaella and
Debess, Iben Nyholm and
Tudor, Crina Madalina",
booktitle = "Proceedings of the Third Workshop on Resources and Representations for Under-Resourced Languages and Domains (RESOURCEFUL-2025)",
month = mar,
year = "2025",
address = "Tallinn, Estonia",
publisher = "University of Tartu Library, Estonia",
url = "https://aclanthology.org/2025.resourceful-1.26/",
pages = "120--129",
ISBN = "978-9908-53-121-2",
abstract = "Optical Character Recognition (OCR) is critical for accurate access to historical corpora, providing a foundation for processing pipelines and the reliable interpretation of historical texts. Despite advances, the quality of OCR in historical documents remains limited, often requiring post-OCR correction to address residual errors. Building on recent progress with instruction-tuned Llama 2 models applied to English historical newspapers, we examine the potential of German Llama 2 and Mistral models for post-OCR correction of German medical historical periodicals. We perform instruction tuning using two configurations of training data, augmenting our small annotated dataset with two German datasets from the same time period. The results demonstrate that German Mistral enhances the raw OCR output, achieving a lower average word error rate (WER). However, the average character error rate (CER) either decreases or remains unchanged across all models considered. We perform an analysis of performance within the error groups and provide an interpretation of the results."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="danilova-aangenendt-2025-post">
<titleInfo>
<title>Post-OCR Correction of Historical German Periodicals using LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Vera</namePart>
<namePart type="family">Danilova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gijs</namePart>
<namePart type="family">Aangenendt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-03</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Third Workshop on Resources and Representations for Under-Resourced Languages and Domains (RESOURCEFUL-2025)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Špela</namePart>
<namePart type="given">Arhar</namePart>
<namePart type="family">Holdt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikolai</namePart>
<namePart type="family">Ilinykh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Barbara</namePart>
<namePart type="family">Scalvini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Micaella</namePart>
<namePart type="family">Bruton</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Iben</namePart>
<namePart type="given">Nyholm</namePart>
<namePart type="family">Debess</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Crina</namePart>
<namePart type="given">Madalina</namePart>
<namePart type="family">Tudor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>University of Tartu Library, Estonia</publisher>
<place>
<placeTerm type="text">Tallinn, Estonia</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">978-9908-53-121-2</identifier>
</relatedItem>
<abstract>Optical Character Recognition (OCR) is critical for accurate access to historical corpora, providing a foundation for processing pipelines and the reliable interpretation of historical texts. Despite advances, the quality of OCR in historical documents remains limited, often requiring post-OCR correction to address residual errors. Building on recent progress with instruction-tuned Llama 2 models applied to English historical newspapers, we examine the potential of German Llama 2 and Mistral models for post-OCR correction of German medical historical periodicals. We perform instruction tuning using two configurations of training data, augmenting our small annotated dataset with two German datasets from the same time period. The results demonstrate that German Mistral enhances the raw OCR output, achieving a lower average word error rate (WER). However, the average character error rate (CER) either decreases or remains unchanged across all models considered. We perform an analysis of performance within the error groups and provide an interpretation of the results.</abstract>
<identifier type="citekey">danilova-aangenendt-2025-post</identifier>
<location>
<url>https://aclanthology.org/2025.resourceful-1.26/</url>
</location>
<part>
<date>2025-03</date>
<extent unit="page">
<start>120</start>
<end>129</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Post-OCR Correction of Historical German Periodicals using LLMs
%A Danilova, Vera
%A Aangenendt, Gijs
%Y Holdt, Špela Arhar
%Y Ilinykh, Nikolai
%Y Scalvini, Barbara
%Y Bruton, Micaella
%Y Debess, Iben Nyholm
%Y Tudor, Crina Madalina
%S Proceedings of the Third Workshop on Resources and Representations for Under-Resourced Languages and Domains (RESOURCEFUL-2025)
%D 2025
%8 March
%I University of Tartu Library, Estonia
%C Tallinn, Estonia
%@ 978-9908-53-121-2
%F danilova-aangenendt-2025-post
%X Optical Character Recognition (OCR) is critical for accurate access to historical corpora, providing a foundation for processing pipelines and the reliable interpretation of historical texts. Despite advances, the quality of OCR in historical documents remains limited, often requiring post-OCR correction to address residual errors. Building on recent progress with instruction-tuned Llama 2 models applied to English historical newspapers, we examine the potential of German Llama 2 and Mistral models for post-OCR correction of German medical historical periodicals. We perform instruction tuning using two configurations of training data, augmenting our small annotated dataset with two German datasets from the same time period. The results demonstrate that German Mistral enhances the raw OCR output, achieving a lower average word error rate (WER). However, the average character error rate (CER) either decreases or remains unchanged across all models considered. We perform an analysis of performance within the error groups and provide an interpretation of the results.
%U https://aclanthology.org/2025.resourceful-1.26/
%P 120-129
Markdown (Informal)
[Post-OCR Correction of Historical German Periodicals using LLMs](https://aclanthology.org/2025.resourceful-1.26/) (Danilova & Aangenendt, RESOURCEFUL 2025)
ACL