Character Transformations for Non-Autoregressive GEC Tagging

Milan Straka, Jakub Náplava, Jana Straková


Abstract
We propose a character-based non-autoregressive GEC approach, with automatically generated character transformations. Recently, per-word classification of correction edits has proven an efficient, parallelizable alternative to current encoder-decoder GEC systems. We show that word replacement edits may be suboptimal and lead to explosion of rules for spelling, diacritization and errors in morphologically rich languages, and propose a method for generating character transformations from GEC corpus. Finally, we train character transformation models for Czech, German and Russian, reaching solid results and dramatic speedup compared to autoregressive systems. The source code is released at https://github.com/ufal/wnut2021_character_transformations_gec.
Anthology ID:
2021.wnut-1.46
Volume:
Proceedings of the Seventh Workshop on Noisy User-generated Text (W-NUT 2021)
Month:
November
Year:
2021
Address:
Online
Venue:
WNUT
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
417–422
Language:
URL:
https://aclanthology.org/2021.wnut-1.46
DOI:
10.18653/v1/2021.wnut-1.46
Bibkey:
Cite (ACL):
Milan Straka, Jakub Náplava, and Jana Straková. 2021. Character Transformations for Non-Autoregressive GEC Tagging. In Proceedings of the Seventh Workshop on Noisy User-generated Text (W-NUT 2021), pages 417–422, Online. Association for Computational Linguistics.
Cite (Informal):
Character Transformations for Non-Autoregressive GEC Tagging (Straka et al., WNUT 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.wnut-1.46.pdf
Code
 ufal/wnut2021_character_transformations_gec
Data
AKCES-GECCoNLL-2014 Shared Task: Grammatical Error Correction