Finnish NER

We have trained an NER system based on FinBERT and a new NER annotation layer of the UD_Finnish-TDT treebank. In comparisons, the NER system surpassed the state-of-the-art.

Links and resources:

Note: the latest model available for download from uses OntoNotes NE types:

Type Description
PERSON People, including fictional
NORP Nationalities or religious or political groups
FAC Buildings, airports, highways, bridges, etc.
ORG Companies, agencies, institutions, etc.
GPE Countries, cities, states
LOC Non-GPE locations, mountain ranges, bodies of water
PRODUCT Objects, vehicles, foods, etc. (Not services.)
EVENT Named hurricanes, battles, wars, sports events, etc.
WORK_OF_ART Titles of books, songs, etc.
LAW Named documents made into laws
LANGUAGE Any named language
DATE Absolute or relative dates or periods
TIME Times smaller than a day
PERCENT Percentage, including “%”
MONEY Monetary values, including unit
QUANTITY Measurements, as of weight or distance
ORDINAL “first”, “second”
CARDINAL Numerals that do not fall under another type

To run a server version of the tagger locally (in UNIX):

git clone
cd keras-bert-ner/
git submodule init
git submodule update
tar xvzf combined-ext-model-130220.tar.gz
python3 --ner_model_dir combined-ext-model

and then try



curl -G --data-urlencode "text=Turun yliopisto"

If you use this system or data, please cite:

Jouni Luoma, Miika Oinonen, Maria Pyykönen, Veronika Laippala, Sampo Pyysalo. 2020. A Broad-coverage Corpus for Finnish Named Entity Recognition. In Proceedings of The 12th Language Resources and Evaluation Conference (LREC’2020). BibTeX