Neural OCR Post-Hoc Correction of Historical Corpora

Author:

Lyu Lijun1,Koutraki Maria2,Krickl Martin3,Fetahu Besnik45

Affiliation:

1. L3S Research Center, Leibniz University of Hannover / Hannover, Germany. lyu@L3S.de

2. L3S Research Center, Leibniz University of Hannover / Hannover, Germany. koutraki@L3S.de

3. Austrian National Library / Vienna, Austria. martin.krickl@onb.ac.at

4. L3S Research Center, Leibniz University of Hannover / Hannover, Germany

5. Amazon / Seattle, WA, USA. besnikf@amazon.com

Abstract

Abstract Optical character recognition (OCR) is crucial for a deeper access to historical collections. OCR needs to account for orthographic variations, typefaces, or language evolution (i.e., new letters, word spellings), as the main source of character, word, or word segmentation transcription errors. For digital corpora of historical prints, the errors are further exacerbated due to low scan quality and lack of language standardization. For the task of OCR post-hoc correction, we propose a neural approach based on a combination of recurrent (RNN) and deep convolutional network (ConvNet) to correct OCR transcription errors. At character level we flexibly capture errors, and decode the corrected output based on a novel attention mechanism. Accounting for the input and output similarity, we propose a new loss function that rewards the model’s correcting behavior. Evaluation on a historical book corpus in German language shows that our models are robust in capturing diverse OCR transcription errors and reduce the word error rate of 32.3% by more than 89%.

Publisher

MIT Press - Journals

Reference35 articles.

1. Using SMT for OCR error correction of historical texts;Afli,2016

2. Neural machine translation by jointly learning to align and translate;Bahdanau,2015

3. Improved transition-based parsing by modeling characters instead of words with lstms;Ballesteros,2015

4. Bootstrapped OCR error detection for a less-resourced language variant;Barbaresi,2016

5. An improved error model for noisy channel spelling correction;Brill,2000

Cited by 6 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Gamified Crowdsourcing in Support of the Semantic Transformation of Digital Libraries;Proceedings of the International Conference on Computer Systems and Technologies 2024;2024-06-14

2. Encoder-Decoder Language Model for Khmer Handwritten Text Recognition in Historical Documents;2022 14th International Conference on Software, Knowledge, Information Management and Applications (SKIMA);2022-12-02

3. CorDeep and the Sacrobosco Dataset: Detection of Visual Elements in Historical Documents;Journal of Imaging;2022-10-15

4. A prototype gutenberg-hathitrust sentence-level parallel corpus for OCR error analysis;Proceedings of the 22nd ACM/IEEE Joint Conference on Digital Libraries;2022-06-20

5. Digital Perspectives in History;Histories;2022-06-04

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3