Named Entity Recognition in Estonian 19th Century Parish Court Records

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 5304–5313 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 5304 Named Entity Recognition in Estonian 19th Century Parish Court Records Siim Orasmaa, Kadri Muischnek, Kristjan Poska, Anna Edela Institute of Computer Science, University of Tartu, Narva mnt 18, 51009 Tartu, Estonia, Institute of Estonian and General Linguistics, University of Tartu, Jakobi 2, 51005, Tartu, Estonia, {siim.orasmaa, kadri.muischnek}@ut.ee , {kristjanposka, anna.edela}@gmail.com Abstract This paper presents a new historical language resource, a corpus of Estonian Parish Court records from the years 1821-1920, annotated for named entities (NE), and reports on named entity recognition (NER) experiments using this corpus. The hand-written records have been transcribed manually via a crowdsourcing project, so the transcripts are of high quality, but the variation of language and spelling is high in these documents due to dialectal variation and the fact that there was a considerable change in Estonian spelling conventions during the time of their writing. The typology of NEs for manual annotation includes 7 categories, but the inter-annotator agreement is as good as 95.0 (mean F1-score). We experimented with fine-tuning BERT-like transfer learning approaches for NER, and found modern Estonian BERT models highly applicable, despite the difficulty of the historical material. Our best model, finetuned Est-RoBERTa, achieved microaverage F1 score of 93.6, which is comparable to state-of-the-art NER performance on the contemporary Estonian. Keywords: historical language processing, named entity recognition, digital humanities, corpus annotation 1. Introduction In this paper we present a new resource – a corpus of 19th century Parish Court records annotated for named entities (NE) in Estonian – and report on experiments on named entity recognition (NER) using this dataset. The corpus contains a subpart of the Estonian Parish Court records from the 1820s to the beginning of the 20th century, the majority of texts originate from the period 1860-1890. The original documents are hand- written and the digitization was done manually by vol- unteers. Although the digitized documents are not completely error-free, manual digitization gives better results than handwritten text recognition or even OCR. The texts of the Parish Court records are very hetero- geneous in terms of spelling conventions, capitalization and dialectal variation, which can hamper both manual NE annotation and NER. This corpus is a valuable source of information for his- torians, linguists interested in language history and the historical development of written language, and also for the public at large. Of course it is also a valuable resource for historical language processing. Annotating it with NEs enables better search queries for the users, also various data analysis and visualiza- tion procedures. By developing a specific NER tool for these texts we make it possible to annotate the NEs in all Parish Court records that are being constantly digi- tized. While NER for contemporary texts is an established task in information retrieval and information extrac- tion, there has also been an increasing amount of work done for NER in historical documents during …

Content truncated to 500 words