Automatically Selecting the Best Dependency Annotation Design with Dynamic Oracles

Guillaume Wisniewski (1) , Ophélie Lacroix (2) , François Yvon (1)

1 TLP - Traitement du Langage Parlé
2 UCPH - University of Copenhagen = Københavns Universitet

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.

Mots clés: dependency parsing

Domaines: Informatique [cs]; Informatique et langage [cs.CL]

Guillaume Wisniewski, Ophélie Lacroix, François Yvon. Automatically Selecting the Best Dependency Annotation Design with Dynamic Oracles. Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Association for Computational Linguistics, Jun 2018, New Orleans, United States. pp.401 - 406, ⟨10.18653/v1/N18-2064⟩. ⟨hal-01813395⟩