HUJI-KU at MRP 2020: Two Transition-based Neural Parsers

Ofir Arviv; Ruixiang Cui; Daniel Hershcovich

doi:10.18653/v1/2020.conll-shared.7

HUJI-KU at MRP 2020: Two Transition-based Neural Parsers

Ofir Arviv, Ruixiang Cui, Daniel Hershcovich

Machine Learning

Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › peer review

7 Citationer (Scopus)

44 Downloads (Pure)

Abstract

This paper describes the HUJI-KU system submission to the shared task on CrossFramework Meaning Representation Parsing (MRP) at the 2020 Conference for Computational Language Learning (CoNLL), employing TUPA and the HIT-SCIR parser, which were, respectively, the baseline system and winning system in the 2019 MRP shared task. Both are transition-based parsers using BERT contextualized embeddings. We generalized TUPA to support the newly-added MRP frameworks and languages, and experimented with multitask learning with the HIT-SCIR parser. We reached 4th place in both the crossframework and cross-lingual tracks.

Originalsprog	Engelsk
Titel	Proceedings of the CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing
Forlag	Association for Computational Linguistics
Publikationsdato	2020
Sider	73-82
DOI	https://doi.org/10.18653/v1/2020.conll-shared.7
Status	Udgivet - 2020
Begivenhed	CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing, - Onlinr Varighed: 19 nov. 2020 → 20 nov. 2020

Konference

Konference	CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing,
Lokation	Onlinr
Periode	19/11/2020 → 20/11/2020

Adgang til dokumentet

10.18653/v1/2020.conll-shared.7Licens: CC BY

OA-HUJI-KU at MRP 2020Forlagets udgivne version, 2,2 MBLicens: CC BY

Citationsformater

Arviv, O, Cui, R & Hershcovich, D 2020, HUJI-KU at MRP 2020: Two Transition-based Neural Parsers. i Proceedings of the CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing. Association for Computational Linguistics, s. 73-82, CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing, 19/11/2020. https://doi.org/10.18653/v1/2020.conll-shared.7

@inproceedings{4dca01904bb54833ab13e862f479e3c1,

title = "HUJI-KU at MRP 2020: Two Transition-based Neural Parsers",

abstract = "This paper describes the HUJI-KU system submission to the shared task on CrossFramework Meaning Representation Parsing (MRP) at the 2020 Conference for Computational Language Learning (CoNLL), employing TUPA and the HIT-SCIR parser, which were, respectively, the baseline system and winning system in the 2019 MRP shared task. Both are transition-based parsers using BERT contextualized embeddings. We generalized TUPA to support the newly-added MRP frameworks and languages, and experimented with multitask learning with the HIT-SCIR parser. We reached 4th place in both the crossframework and cross-lingual tracks.",

author = "Ofir Arviv and Ruixiang Cui and Daniel Hershcovich",

year = "2020",

doi = "10.18653/v1/2020.conll-shared.7",

language = "English",

pages = "73--82",

booktitle = "Proceedings of the CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing",

publisher = "Association for Computational Linguistics",

note = "CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing, ; Conference date: 19-11-2020 Through 20-11-2020",

}

TY - GEN

T1 - HUJI-KU at MRP 2020

T2 - CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing,

AU - Arviv, Ofir

AU - Cui, Ruixiang

AU - Hershcovich, Daniel

PY - 2020

Y1 - 2020

N2 - This paper describes the HUJI-KU system submission to the shared task on CrossFramework Meaning Representation Parsing (MRP) at the 2020 Conference for Computational Language Learning (CoNLL), employing TUPA and the HIT-SCIR parser, which were, respectively, the baseline system and winning system in the 2019 MRP shared task. Both are transition-based parsers using BERT contextualized embeddings. We generalized TUPA to support the newly-added MRP frameworks and languages, and experimented with multitask learning with the HIT-SCIR parser. We reached 4th place in both the crossframework and cross-lingual tracks.

AB - This paper describes the HUJI-KU system submission to the shared task on CrossFramework Meaning Representation Parsing (MRP) at the 2020 Conference for Computational Language Learning (CoNLL), employing TUPA and the HIT-SCIR parser, which were, respectively, the baseline system and winning system in the 2019 MRP shared task. Both are transition-based parsers using BERT contextualized embeddings. We generalized TUPA to support the newly-added MRP frameworks and languages, and experimented with multitask learning with the HIT-SCIR parser. We reached 4th place in both the crossframework and cross-lingual tracks.

U2 - 10.18653/v1/2020.conll-shared.7

DO - 10.18653/v1/2020.conll-shared.7

M3 - Article in proceedings

SP - 73

EP - 82

BT - Proceedings of the CoNLL 2020 Shared Task: Cross-Framework Meaning Representation Parsing

PB - Association for Computational Linguistics

Y2 - 19 November 2020 through 20 November 2020

ER -