# The BERTić* [bert-ich] /bɜrtitʃ/ model fine-tuned for the task of named entity recognition in Bosnian, Croatian, Montenegrin and Serbian (BCMS)

* The name should resemble the facts (1) that the model was trained in Zagreb, Croatia, where diminutives ending in -ić (as in fotić, smajlić, hengić etc.) are very popular, and (2) that most surnames in the countries where these languages are spoken end in -ić (with diminutive etymology as well).

This is a fine-tuned version of the BERTić model for the task of named entity recognition (PER, LOC, ORG, MISC). The fine-tuning was performed on the following datasets:

• the hr500k dataset, 500 thousand tokens in size, standard Croatian
• the SETimes.SR dataset, 87 thousand tokens in size, standard Serbian
• the ReLDI-hr dataset, 89 thousand tokens in size, Internet (Twitter) Croatian
• the ReLDI-sr dataset, 92 thousand tokens in size, Internet (Twitter) Serbian

The data was augmented with missing diacritics and standard data was additionally over-represented. The F1 obtained on dev data (train and test was merged into train) is 91.38. For a more detailed per-dataset evaluation of the BERTić model on the NER task have a look at the main model page.

If you use this fine-tuned model, please cite the following paper:

@inproceedings{ljubesic-lauc-2021-bertic,
title = "{BERT}i{\'c} - The Transformer Language Model for {B}osnian, {C}roatian, {M}ontenegrin and {S}erbian",
author = "Ljube{\v{s}}i{\'c}, Nikola  and Lauc, Davor",
booktitle = "Proceedings of the 8th Workshop on Balto-Slavic Natural Language Processing",
month = apr,
year = "2021",