Neural reranking for dependency parsing: An evaluation

PDF
2020.acl-main.379-1.pdf - Published
Download (354kB)

URL:	https://madoc.bib.uni-mannheim.de/55426
Additional URL:	https://www.aclweb.org/anthology/2020.acl-main.379...
URN:	urn:nbn:de:bsz:180-madoc-554262
Document Type:	Conference or workshop publication
Year of publication:	2020
Book title:	The 58th Annual Meeting of the Association for Computational Linguistics - proceedings of the conference : July 5-10, 2020 : ACL 2020
Page range:	4123-4133
Conference title:	ACL 2020
Location of the conference venue:	Online
Date of the conference:	05.-10.07.2020
Publisher:	Jurafsky, Dan
Place of publication:	Stroudsburg, PA
Publishing house:	Association for Computational Linguistics, ACL
ISBN:	978-1-952148-25-5
Related URLs:	https://www.aclweb.org/anthology/volumes...
Publication language:	English
Institution:	Außerfakultäre Einrichtungen > SFB 884
Pre-existing license:	Creative Commons Attribution, Non-Commercial, Share Alike 3.0 Unported (CC BY-NC-SA 3.0)
Subject:	004 Computer science, internet
Abstract:	Recent work has shown that neural rerankers can improve results for dependency parsing over the top k trees produced by a base parser. However, all neural rerankers so far have been evaluated on English and Chinese only, both languages with a configurational word order and poor morphology. In the paper, we re-assess the potential of successful neural reranking models from the literature on English and on two morphologically rich(er) languages, German and Czech. In addition, we introduce a new variation of a discriminative reranker based on graph convolutional networks (GCNs). We show that the GCN not only outperforms previous models on English but is the only model that is able to improve results over the baselines on German and Czech. We explain the differences in reranking performance based on an analysis of a) the gold tree ratio and b) the variety in the k-best lists.
Additional information:	Online-Ressource