FinArcNER is a named entity recognition model trained with mostly archival data. This repository has been produced as part of the
FIN-CLARIAH infrastructure project. The model training and paper writing has been conducted in cooperation with the
National Archives of Finland.
1from transformers import pipeline
2
3model_checkpoint = "jyu-digihum/finarcner"
4token_classifier = pipeline(
5 "token-classification", model=model_checkpoint, aggregation_strategy="simple"
6)
7predictions = token_classifier("'Helsingistä tuli Suomen suuriruhtinaskunnan pääkaupunki vuonna 1812.")
8print(predictions)
In addition to the model and full paper, we publish annotation guidelines
in JYX. For the guidelines, you may use the reference below:
1@misc{poso2025ner,
2title={NER annotation guidelines for archival data},
3author={Poso, Venla and Välisalo, Tanja and Toivanen, Ida and Lipsanen, Mikko and Kukkohovi, Laura and Kytöaho, Roosa and Palander, Satu and Pohjola, Maiju and Laitinen, Vesa and Föhr, Atte and Abdelamir, Amir and Niemi, Joonas},
4journal={JYX Digital Repository},
5year={2025},
6publisher={University of Jyväskylä},
7url={http://urn.fi/URN:NBN:fi:jyu-202501291584}
8}
1@article{toivanen2025developing,
2title={Developing named-entity recognition for state authority archives},
3author={Toivanen, Ida and Poso, Venla and Lipsanen, Mikko and Välisalo, Tanja},
4journal={Digital Humanities in the Nordic and Baltic Countries Publications},
5number={3},
6year={2025},
7publisher={University of Oslo Library},
8DOI={https://doi.org/10.5617/dhnbpub.12262}
9}
