Is it Possible to Re-educate RoBERTa? Expert-driven Machine
Learning for Punctuation Correction

J 2023

Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction

MACHURA, Jakub, Hana ŽIŽKOVÁ, Adam FRÉMUND and Jan ŠVEC

Basic information

Original name

Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction

Authors

MACHURA, Jakub (203 Czech Republic, belonging to the institution), Hana ŽIŽKOVÁ (203 Czech Republic, guarantor, belonging to the institution), Adam FRÉMUND (203 Czech Republic) and Jan ŠVEC (203 Czech Republic)

Edition

Jazykovedný časopis, Bratislava, Jazykovedný ústav Ľudovíta Štúra Slovenskej akadémie vied, 2023, 0021-5597

Other information

Language

English

Type of outcome

Článek v odborném periodiku

Field of Study

60203 Linguistics

Country of publisher

Slovakia

Confidentiality degree

není předmětem státního či obchodního tajemství

References:

URL

RIV identification code

RIV/00216224:14210/23:00131927

Organization unit

Faculty of Arts

DOI

http://dx.doi.org/10.2478/jazcas-2023-0052

Keywords in English

comma; Czech; vocative; machine learning; RoBERTa

Abstract

V originále

Although Czech rule-based tools for automatic punctuation insertion rely on extensive grammar and achieve respectable precision, the pre-trained Transformers outperform rule-based systems in precision and recall [hidden reference]. The Czech pre-trained RoBERTa model achieves excellent results, yet a certain level of phenomena is ignored, and the model partially makes errors. This paper aims to investigate whether it is possible to retrain the RoBERTa language model to increase the number of sentence commas the model correctly detects. We have chosen a very specific and narrow type of sentence comma, namely the sentence comma delimiting vocative phrases, which is clearly defined in the grammar and is very often omitted by writers. The chosen approaches were further tested and evaluated on different types of texts.

Links

MUNI/A/1249/2022, interní kód MU

Name: Lexikon a gramatika češtiny III - 2023

Investor: Masaryk University

Citovat

MACHURA, Jakub, Hana ŽIŽKOVÁ, Adam FRÉMUND and Jan ŠVEC. Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction. Jazykovedný časopis. Bratislava: Jazykovedný ústav Ľudovíta Štúra Slovenskej akadémie vied, 2023, vol. 74, No 1, p. 357-368. ISSN 0021-5597. Available from: https://dx.doi.org/10.2478/jazcas-2023-0052.

@article{2326659,
   author = {Machura, Jakub and Žižková, Hana and Frémund, Adam and Švec, Jan},
   article_location = {Bratislava},
   article_number = {1},
   doi = {http://dx.doi.org/10.2478/jazcas-2023-0052},
   keywords = {comma; Czech; vocative; machine learning; RoBERTa},
   language = {eng},
   issn = {0021-5597},
   journal = {Jazykovedný časopis},
   title = {Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction},
   url = {https://www.juls.savba.sk/ediela/jc/2023/1/jc23-01.pdf},
   volume = {74},
   year = {2023}
}

TY  - JOUR
ID  - 2326659
AU  - Machura, Jakub - Žižková, Hana - Frémund, Adam - Švec, Jan
PY  - 2023
TI  - Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction
JF  - Jazykovedný časopis
VL  - 74
IS  - 1
SP  - 357-368
EP  - 357-368
PB  - Jazykovedný ústav Ľudovíta Štúra Slovenskej akadémie vied
SN  - 00215597
KW  - comma
KW  - Czech
KW  - vocative
KW  - machine learning
KW  - RoBERTa
UR  - https://www.juls.savba.sk/ediela/jc/2023/1/jc23-01.pdf
N2  - Although Czech rule-based tools for automatic punctuation insertion rely on extensive grammar and achieve respectable precision, the pre-trained Transformers outperform rule-based systems in precision and recall [hidden reference]. The Czech pre-trained RoBERTa model achieves excellent results, yet a certain level of phenomena is ignored, and the model partially makes errors. This paper aims to investigate whether it is possible to retrain the RoBERTa language model to increase the number of sentence commas the model correctly detects. We have chosen a very specific and narrow type of sentence comma, namely the sentence comma delimiting vocative phrases, which is clearly defined in the grammar and is very often omitted by writers. The chosen approaches were further tested and evaluated on different types of texts.
ER  -

MACHURA, Jakub, Hana ŽIŽKOVÁ, Adam FRÉMUND and Jan ŠVEC. Is it Possible to Re-educate RoBERTa? Expert-driven Machine Learning for Punctuation Correction. \textit{Jazykovedný časopis}. Bratislava: Jazykovedný ústav Ľudovíta Štúra Slovenskej akadémie vied, 2023, vol.~74, No~1, p.~357-368. ISSN~0021-5597. Available from: https://dx.doi.org/10.2478/jazcas-2023-0052.

Detailed Information on Publication Record