SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation

Junpei Zhou; Zhisong Zhang; Zecong Hu

doi:10.18653/v1/w19-5411

SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation

Junpei Zhou (Co-first Author), Zhisong Zhang (Co-first Author), Zecong Hu (Co-first Author)

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

9 Citations (Scopus)

11 Downloads (CityUHK Scholars)

Abstract

Quality estimation (QE) of machine translation (MT) systems is a task of growing importance. It reduces the cost of post-editing, allowing machine-translated text to be used in formal occasions. In this work, we describe our submission system in WMT 2019 sentence-level QE task. We mainly explore the utilization of pre-trained translation models in QE and adopt a bi-directional translation-like strategy. The strategy is similar to ELMo, but additionally conditions on source sentences. Experiments on WMT QE dataset show that our strategy, which makes the pre-training slightly harder, can bring improvements for QE. In WMT-2019 QE task, our system ranked in the second place on En-De NMT dataset and the third place on En-Ru NMT dataset. ©2019 Association for Computational Linguistics.

Original language	English
Title of host publication	Proceedings of the Fourth Conference on Machine Translation
Publisher	Association for Computational Linguistics
Pages	106-111
Number of pages	6
Volume	3: Shared Task Papers, Day 2
ISBN (Print)	9781950737277
DOIs	https://doi.org/10.18653/v1/w19-5411
Publication status	Published - Aug 2019
Externally published	Yes
Event	4th Conference on Machine Translation (WMT 2019) - Florence, Italy Duration: 1 Aug 2019 → 2 Aug 2019 https://aclanthology.org/volumes/W19-52/

Conference

Conference	4th Conference on Machine Translation (WMT 2019)
Place	Italy
City	Florence
Period	1/08/19 → 2/08/19
Internet address	https://aclanthology.org/volumes/W19-52/

Publisher's Copyright Statement

This full text is made available under CC-BY 4.0. https://creativecommons.org/licenses/by/4.0/

Access Output

10.18653/v1/w19-5411

316818163.pdfFinal Published version, 356 KBLicence: CC BY

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inproceedings{931928baa3cc41ed8cbeb1d9c5ce927e,

title = "SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation",

abstract = "Quality estimation (QE) of machine translation (MT) systems is a task of growing importance. It reduces the cost of post-editing, allowing machine-translated text to be used in formal occasions. In this work, we describe our submission system in WMT 2019 sentence-level QE task. We mainly explore the utilization of pre-trained translation models in QE and adopt a bi-directional translation-like strategy. The strategy is similar to ELMo, but additionally conditions on source sentences. Experiments on WMT QE dataset show that our strategy, which makes the pre-training slightly harder, can bring improvements for QE. In WMT-2019 QE task, our system ranked in the second place on En-De NMT dataset and the third place on En-Ru NMT dataset. {\textcopyright}2019 Association for Computational Linguistics.",

author = "Junpei Zhou and Zhisong Zhang and Zecong Hu",

year = "2019",

month = aug,

doi = "10.18653/v1/w19-5411",

language = "English",

isbn = "9781950737277",

volume = "3: Shared Task Papers, Day 2",

pages = "106--111",

booktitle = "Proceedings of the Fourth Conference on Machine Translation",

publisher = "Association for Computational Linguistics",

address = "United States",

note = "4th Conference on Machine Translation (WMT 2019) ; Conference date: 01-08-2019 Through 02-08-2019",

url = "https://aclanthology.org/volumes/W19-52/",

}

Zhou, J, Zhang, Z & Hu, Z 2019, SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation. in Proceedings of the Fourth Conference on Machine Translation. vol. 3: Shared Task Papers, Day 2, Association for Computational Linguistics, pp. 106-111, 4th Conference on Machine Translation (WMT 2019), Florence, Italy, 1/08/19. https://doi.org/10.18653/v1/w19-5411

SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation. / Zhou, Junpei (Co-first Author); Zhang, Zhisong (Co-first Author); Hu, Zecong (Co-first Author).
Proceedings of the Fourth Conference on Machine Translation. Vol. 3: Shared Task Papers, Day 2 Association for Computational Linguistics, 2019. p. 106-111.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - SOURCE

T2 - 4th Conference on Machine Translation (WMT 2019)

AU - Zhou, Junpei

AU - Zhang, Zhisong

AU - Hu, Zecong

PY - 2019/8

Y1 - 2019/8

N2 - Quality estimation (QE) of machine translation (MT) systems is a task of growing importance. It reduces the cost of post-editing, allowing machine-translated text to be used in formal occasions. In this work, we describe our submission system in WMT 2019 sentence-level QE task. We mainly explore the utilization of pre-trained translation models in QE and adopt a bi-directional translation-like strategy. The strategy is similar to ELMo, but additionally conditions on source sentences. Experiments on WMT QE dataset show that our strategy, which makes the pre-training slightly harder, can bring improvements for QE. In WMT-2019 QE task, our system ranked in the second place on En-De NMT dataset and the third place on En-Ru NMT dataset. ©2019 Association for Computational Linguistics.

AB - Quality estimation (QE) of machine translation (MT) systems is a task of growing importance. It reduces the cost of post-editing, allowing machine-translated text to be used in formal occasions. In this work, we describe our submission system in WMT 2019 sentence-level QE task. We mainly explore the utilization of pre-trained translation models in QE and adopt a bi-directional translation-like strategy. The strategy is similar to ELMo, but additionally conditions on source sentences. Experiments on WMT QE dataset show that our strategy, which makes the pre-training slightly harder, can bring improvements for QE. In WMT-2019 QE task, our system ranked in the second place on En-De NMT dataset and the third place on En-Ru NMT dataset. ©2019 Association for Computational Linguistics.

UR - http://www.scopus.com/inward/record.url?scp=85095005083&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85095005083&origin=recordpage

U2 - 10.18653/v1/w19-5411

DO - 10.18653/v1/w19-5411

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9781950737277

VL - 3: Shared Task Papers, Day 2

SP - 106

EP - 111

BT - Proceedings of the Fourth Conference on Machine Translation

PB - Association for Computational Linguistics

Y2 - 1 August 2019 through 2 August 2019

ER -

SOURCE: SOURce-conditional elmo-style model for machine translation quality estimation

Abstract

Conference

Publisher's Copyright Statement

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this