Data Centric Domain Adaptation for Historical Text with OCR Errors

Marz, Luisa; Schweter, Stefan; Poerner, Nina; Roth, Benjamin und Schütze, Hinrich (2. September 2021): Data Centric Domain Adaptation for Historical Text with OCR Errors. International Conference on Document Analysis and Recognition, Lausanne, Switzerland, September 2021. Lladós, J.; Lopresti, D. und Uchida, S. (eds.) : In: Document Analysis and Recognition – ICDAR 2021. 16th International Conference, Lausanne, Switzerland, September 5–10, 2021, Proceedings, Part II, Cham: Springer. pp. 748-761

Abstract

We propose new methods for in-domain and cross-domain Named Entity Recognition (NER) on historical data for Dutch and French. For the cross-domain case, we address domain shift by integrating unsupervised in-domain data via contextualized string embeddings; and OCR errors by injecting synthetic OCR errors into the source domain and address data centric domain adaptation. We propose a general approach to imitate OCR errors in arbitrary input data. Our cross-domain as well as our in-domain results outperform several strong baselines and establish state-of-the-art results. We publish preprocessed versions of the French and Dutch Europeana NER corpora.

• Item Type: Conference or Workshop Item (Paper)
• EU Funded Grant Agreement Number: 740516
• EU Projects: Horizon 2020 > ERC Grants > ERC Advanced Grant > ERC Grant 740516: NonSequeToR - Non-sequence models for tokenization replacement
• Form of publication: Preprint
• Research Centers: Center for Information and Language Processing (CIS)
• Subjects: 000 Computer science, information and general works > 000 Computer science, knowledge, and systems; 400 Language > 400 Language; 400 Language > 410 Linguistics
• Place of Publication: Cham
• Language: English