Better, Faster, Stronger Sequence Tagging Constituent Parsers

David Vilares, Mostafa Abdou, Anders Søgaard

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskningpeer review

18 Citationer (Scopus)
70 Downloads (Pure)

Abstract

Sequence tagging models for constituent parsing are faster, but less accurate than other types of parsers. In this work, we address the following weaknesses of such constituent parsers: (a) high error rates around closing brackets of long constituents, (b) large label sets, leading to sparsity, and (c) error propagation arising from greedy decoding. To effectively close brackets, we train a model that learns to switch between tagging schemes. To reduce sparsity, we decompose the label set and use multi-task learning to jointly learn to predict sublabels. Finally, we mitigate issues from greedy decoding through auxiliary losses and sentence-level fine-tuning with policy gradient. Combining these techniques, we clearly surpass the performance of sequence tagging constituent parsers on the English and Chinese Penn Treebanks, and reduce their parsing time even further. On the SPMRL datasets, we observe even greater improvements across the board, including a new state of the art on Basque, Hebrew, Polish and Swedish.
OriginalsprogEngelsk
TitelProceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)
ForlagAssociation for Computational Linguistics
Publikationsdato2019
Sider3372-3383
DOI
StatusUdgivet - 2019
Begivenhed2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies - NAACL-HLT 2019 - Minneapolis, USA
Varighed: 3 jun. 20197 jun. 2019

Konference

Konference2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies - NAACL-HLT 2019
Land/OmrådeUSA
ByMinneapolis
Periode03/06/201907/06/2019

Citationsformater