Automatic punctuation restoration with BERT models
Published in
arXiv.org(2021)External Links:
TL
TL;DR
An approach for automatic punctuation restoration with BERT models for English and Hungarian is presented, which achieves a macro-averaged $F_1$-score of 79.8 in English and 82.2 in Hungarian.
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 $F_1$-score of 79.8 in English and 82.2 in Hungarian. Our code is publicly available.
References23 items
1
Efficient Automatic Punctuation Restoration Using Bidirectional Transformers with Robust Inference
2
Controllable Time-Delay Transformer for Real-Time Punctuation Prediction and Disfluency Detection
3
Question Mark Prediction By Bert
4
2019 22nd Conference of the Oriental COCOSDA International Committee for the Co-ordination and Standardisation of Speech Databases and Assessment Techniques (O-COCOSDA)
5