KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task

Sugyeong Eo; Chanjun Park; Hyeonseok Moon; Jaehyung Seo; Heuiseok Lim

KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task

Sugyeong Eo, Chanjun Park, Hyeonseok Moon, Jaehyung Seo, Heuiseok Lim

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

This paper presents KU X Upstage's submission to the quality estimation (QE): critical error detection (CED) shared task in WMT22. We leverage the XLM-RoBERTa large model without utilizing any additional parallel data. To the best of our knowledge, we apply prompt-based fine-tuning to the QE task for the first time. To maximize the model's language understanding capability, we reformulate the CED task to be similar to the masked language model objective, which is a pre-training strategy of the language model. We design intuitive templates and label words, and include auxiliary descriptions such as demonstration or Google Translate results in the input sequence. We further improve the performance through the template ensemble, and as a result of the shared task, our approach achieve the best performance for both English-German and Portuguese-English language pairs in an unconstrained setting.

Original language	English
Title of host publication	WMT 2022 - 7th Conference on Machine Translation, Proceedings of the Conference
Publisher	Association for Computational Linguistics
Pages	606-614
Number of pages	9
ISBN (Electronic)	9781959429296
Publication status	Published - 2022
Event	7th Conference on Machine Translation, WMT 2022 - Abu Dhabi, United Arab Emirates Duration: 2022 Dec 7 → 2022 Dec 8

Publication series

Name	Conference on Machine Translation - Proceedings
ISSN (Electronic)	2768-0983

Conference

Conference	7th Conference on Machine Translation, WMT 2022
Country/Territory	United Arab Emirates
City	Abu Dhabi
Period	22/12/7 → 22/12/8

Bibliographical note

Funding Information:
This research was supported by the Ministry of Science and ICT (MSIT), Korea, under the Information Technology Research Center (ITRC) support program (IITP-2022-2018-0-01405) supervised by the Institute for Information & Communications Technology Planning & Evaluation(IITP) and this work was supported by an IITP grant funded by the MSIT (No. 2020-0-00368, A Neural-Symbolic Model for Knowledge Acquisition and Inference Techniques) and this research was supported by the MSIT, Korea, under the ICT Creative Consilience program (IITP-2022-2020-0-01819) supervised by the IITP.

Publisher Copyright:
© 2022 Association for Computational Linguistics.

ASJC Scopus subject areas

Language and Linguistics
Human-Computer Interaction
Software

Cite this

KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task. / Eo, Sugyeong; Park, Chanjun; Moon, Hyeonseok et al.
WMT 2022 - 7th Conference on Machine Translation, Proceedings of the Conference. Association for Computational Linguistics, 2022. p. 606-614 (Conference on Machine Translation - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Eo, S, Park, C, Moon, H, Seo, J & Lim, H 2022, KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task. in WMT 2022 - 7th Conference on Machine Translation, Proceedings of the Conference. Conference on Machine Translation - Proceedings, Association for Computational Linguistics, pp. 606-614, 7th Conference on Machine Translation, WMT 2022, Abu Dhabi, United Arab Emirates, 22/12/7.

@inproceedings{af9e321ebd684ea49293de6d113a1bb1,

title = "KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task",

abstract = "This paper presents KU X Upstage's submission to the quality estimation (QE): critical error detection (CED) shared task in WMT22. We leverage the XLM-RoBERTa large model without utilizing any additional parallel data. To the best of our knowledge, we apply prompt-based fine-tuning to the QE task for the first time. To maximize the model's language understanding capability, we reformulate the CED task to be similar to the masked language model objective, which is a pre-training strategy of the language model. We design intuitive templates and label words, and include auxiliary descriptions such as demonstration or Google Translate results in the input sequence. We further improve the performance through the template ensemble, and as a result of the shared task, our approach achieve the best performance for both English-German and Portuguese-English language pairs in an unconstrained setting.",

author = "Sugyeong Eo and Chanjun Park and Hyeonseok Moon and Jaehyung Seo and Heuiseok Lim",

note = "Funding Information: This research was supported by the Ministry of Science and ICT (MSIT), Korea, under the Information Technology Research Center (ITRC) support program (IITP-2022-2018-0-01405) supervised by the Institute for Information & Communications Technology Planning & Evaluation(IITP) and this work was supported by an IITP grant funded by the MSIT (No. 2020-0-00368, A Neural-Symbolic Model for Knowledge Acquisition and Inference Techniques) and this research was supported by the MSIT, Korea, under the ICT Creative Consilience program (IITP-2022-2020-0-01819) supervised by the IITP. Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; 7th Conference on Machine Translation, WMT 2022 ; Conference date: 07-12-2022 Through 08-12-2022",

year = "2022",

language = "English",

series = "Conference on Machine Translation - Proceedings",

publisher = "Association for Computational Linguistics",

pages = "606--614",

booktitle = "WMT 2022 - 7th Conference on Machine Translation, Proceedings of the Conference",

}

TY - GEN

T1 - KU X Upstage's submission for the WMT22 Quality Estimation

T2 - 7th Conference on Machine Translation, WMT 2022

AU - Eo, Sugyeong

AU - Park, Chanjun

AU - Moon, Hyeonseok

AU - Seo, Jaehyung

AU - Lim, Heuiseok

N1 - Funding Information: This research was supported by the Ministry of Science and ICT (MSIT), Korea, under the Information Technology Research Center (ITRC) support program (IITP-2022-2018-0-01405) supervised by the Institute for Information & Communications Technology Planning & Evaluation(IITP) and this work was supported by an IITP grant funded by the MSIT (No. 2020-0-00368, A Neural-Symbolic Model for Knowledge Acquisition and Inference Techniques) and this research was supported by the MSIT, Korea, under the ICT Creative Consilience program (IITP-2022-2020-0-01819) supervised by the IITP. Publisher Copyright: © 2022 Association for Computational Linguistics.

PY - 2022

Y1 - 2022

N2 - This paper presents KU X Upstage's submission to the quality estimation (QE): critical error detection (CED) shared task in WMT22. We leverage the XLM-RoBERTa large model without utilizing any additional parallel data. To the best of our knowledge, we apply prompt-based fine-tuning to the QE task for the first time. To maximize the model's language understanding capability, we reformulate the CED task to be similar to the masked language model objective, which is a pre-training strategy of the language model. We design intuitive templates and label words, and include auxiliary descriptions such as demonstration or Google Translate results in the input sequence. We further improve the performance through the template ensemble, and as a result of the shared task, our approach achieve the best performance for both English-German and Portuguese-English language pairs in an unconstrained setting.

AB - This paper presents KU X Upstage's submission to the quality estimation (QE): critical error detection (CED) shared task in WMT22. We leverage the XLM-RoBERTa large model without utilizing any additional parallel data. To the best of our knowledge, we apply prompt-based fine-tuning to the QE task for the first time. To maximize the model's language understanding capability, we reformulate the CED task to be similar to the masked language model objective, which is a pre-training strategy of the language model. We design intuitive templates and label words, and include auxiliary descriptions such as demonstration or Google Translate results in the input sequence. We further improve the performance through the template ensemble, and as a result of the shared task, our approach achieve the best performance for both English-German and Portuguese-English language pairs in an unconstrained setting.

UR - http://www.scopus.com/inward/record.url?scp=85160568003&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85160568003

T3 - Conference on Machine Translation - Proceedings

SP - 606

EP - 614

BT - WMT 2022 - 7th Conference on Machine Translation, Proceedings of the Conference

PB - Association for Computational Linguistics

Y2 - 7 December 2022 through 8 December 2022

ER -

KU X Upstage's submission for the WMT22 Quality Estimation: Critical Error Detection Shared Task

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this