English

Automatic punctuation restoration with BERT models

Computation and Language 2021-01-20 v1

Abstract

We present an approach for automatic punctuation restoration with BERT models for English and Hungarian. For English, we conduct our experiments on Ted Talks, a commonly used benchmark for punctuation restoration, while for Hungarian we evaluate our models on the Szeged Treebank dataset. Our best models achieve a macro-averaged F1F_1-score of 79.8 in English and 82.2 in Hungarian. Our code is publicly available.

Keywords

Cite

@article{arxiv.2101.07343,
  title  = {Automatic punctuation restoration with BERT models},
  author = {Attila Nagy and Bence Bial and Judit Ács},
  journal= {arXiv preprint arXiv:2101.07343},
  year   = {2021}
}

Comments

11 pages, 6 figures, source code at https://github.com/attilanagy234/neural-punctuator

R2 v1 2026-06-23T22:17:40.535Z