Automatically Selecting the Best Dependency Annotation Design with Dynamic Oracles

Guillaume Wisniewski, Ophélie Lacroix, François Yvon


Abstract
This work introduces a new strategy to compare the numerous conventions that have been proposed over the years for expressing dependency structures and discover the one for which a parser will achieve the highest parsing performance. Instead of associating each sentence in the training set with a single gold reference we propose to consider a set of references encoding alternative syntactic representations. Training a parser with a dynamic oracle will then automatically select among all alternatives the reference that will be predicted with the highest accuracy. Experiments on the UD corpora show the validity of this approach.
Anthology ID:
N18-2064
Volume:
Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers)
Month:
June
Year:
2018
Address:
New Orleans, Louisiana
Editors:
Marilyn Walker, Heng Ji, Amanda Stent
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
401–406
Language:
URL:
https://aclanthology.org/N18-2064
DOI:
10.18653/v1/N18-2064
Bibkey:
Cite (ACL):
Guillaume Wisniewski, Ophélie Lacroix, and François Yvon. 2018. Automatically Selecting the Best Dependency Annotation Design with Dynamic Oracles. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), pages 401–406, New Orleans, Louisiana. Association for Computational Linguistics.
Cite (Informal):
Automatically Selecting the Best Dependency Annotation Design with Dynamic Oracles (Wisniewski et al., NAACL 2018)
Copy Citation:
PDF:
https://aclanthology.org/N18-2064.pdf