Graph Representations for Machine Translation in Dialogue Settings

Lea Krause; Selene Baez Santamaria; Jan Christoph Kalo

doi:10.18653/v1/2024.wmt-1.106

Graph Representations for Machine Translation in Dialogue Settings

Lea Krause^*, Selene Baez Santamaria, Jan Christoph Kalo

^*Corresponding author for this work

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

Abstract

In this paper, we present our approach to the WMT24 - Chat Task, addressing the challenge of translating chat conversations. Chat conversations are characterised by their informal, ungrammatical nature and strong reliance on context posing significant challenges for machine translation systems. To address these challenges, we augment large language models with explicit memory mechanisms designed to enhance coherence and consistency across dialogues. Specifically, we employ graph representations to capture and utilise dialogue context, leveraging concept connectivity as a compressed memory. Our approach ranked second place for Dutch and French, and third place for Portuguese and German, based on COMET-22 scores and human evaluation.

Original language	English
Title of host publication	Proceedings of the Ninth Conference on Machine Translation
Editors	Barry Haddow, Tom Kocmi, Philipp Koehn, Christof Monz
Publisher	Association for Computational Linguistics
Pages	1038-1046
Number of pages	9
ISBN (Electronic)	9798891761797
DOIs	https://doi.org/10.18653/v1/2024.wmt-1.106
Publication status	Published - 2024
Event	9th Conference on Machine Translation, WMT 2024 - Miami, United States Duration: 15 Nov 2024 → 16 Nov 2024

Publication series

Name	Conference on Machine Translation - Proceedings
Volume	2024-November
ISSN (Electronic)	2768-0983

Conference

Conference	9th Conference on Machine Translation, WMT 2024
Country/Territory	United States
City	Miami
Period	15/11/24 → 16/11/24

Bibliographical note

Publisher Copyright:
©2024 Association for Computational Linguistics.

Funding

This research was funded by the Vrije Universiteit Amsterdam and the Netherlands Organisation for Scientific Research (NWO) through the Hybrid Intelligence Centre via the Zwaartekracht grant (024.004.022), and the Spinoza grant (SPI 63-260) awarded to Piek Vossen.

Funders	Funder number
Nederlandse Organisatie voor Wetenschappelijk Onderzoek
Hybrid Intelligence Centre	024.004.022, SPI 63-260

Access to Document

10.18653/v1/2024.wmt-1.106

Persistent URL (handle)

Persistent link

Cite this

Krause, L., Santamaria, S. B., & Kalo, J. C. (2024). Graph Representations for Machine Translation in Dialogue Settings. In B. Haddow, T. Kocmi, P. Koehn, & C. Monz (Eds.), Proceedings of the Ninth Conference on Machine Translation (pp. 1038-1046). (Conference on Machine Translation - Proceedings; Vol. 2024-November). Association for Computational Linguistics. https://doi.org/10.18653/v1/2024.wmt-1.106

Krause, Lea ; Santamaria, Selene Baez ; Kalo, Jan Christoph. / Graph Representations for Machine Translation in Dialogue Settings. Proceedings of the Ninth Conference on Machine Translation. editor / Barry Haddow ; Tom Kocmi ; Philipp Koehn ; Christof Monz. Association for Computational Linguistics, 2024. pp. 1038-1046 (Conference on Machine Translation - Proceedings).

@inproceedings{efa5447b29ed4135892f144237cd7afa,

title = "Graph Representations for Machine Translation in Dialogue Settings",

abstract = "In this paper, we present our approach to the WMT24 - Chat Task, addressing the challenge of translating chat conversations. Chat conversations are characterised by their informal, ungrammatical nature and strong reliance on context posing significant challenges for machine translation systems. To address these challenges, we augment large language models with explicit memory mechanisms designed to enhance coherence and consistency across dialogues. Specifically, we employ graph representations to capture and utilise dialogue context, leveraging concept connectivity as a compressed memory. Our approach ranked second place for Dutch and French, and third place for Portuguese and German, based on COMET-22 scores and human evaluation.",

author = "Lea Krause and Santamaria, {Selene Baez} and Kalo, {Jan Christoph}",

note = "Publisher Copyright: {\textcopyright}2024 Association for Computational Linguistics.; 9th Conference on Machine Translation, WMT 2024 ; Conference date: 15-11-2024 Through 16-11-2024",

year = "2024",

doi = "10.18653/v1/2024.wmt-1.106",

language = "English",

series = "Conference on Machine Translation - Proceedings",

publisher = "Association for Computational Linguistics",

pages = "1038--1046",

editor = "Barry Haddow and Tom Kocmi and Philipp Koehn and Christof Monz",

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

}

Krause, L , Santamaria, SB & Kalo, JC 2024, Graph Representations for Machine Translation in Dialogue Settings. in B Haddow, T Kocmi, P Koehn & C Monz (eds), Proceedings of the Ninth Conference on Machine Translation. Conference on Machine Translation - Proceedings, vol. 2024-November, Association for Computational Linguistics, pp. 1038-1046, 9th Conference on Machine Translation, WMT 2024, Miami, United States, 15/11/24. https://doi.org/10.18653/v1/2024.wmt-1.106

Graph Representations for Machine Translation in Dialogue Settings. / Krause, Lea ; Santamaria, Selene Baez ; Kalo, Jan Christoph.
Proceedings of the Ninth Conference on Machine Translation. ed. / Barry Haddow; Tom Kocmi; Philipp Koehn; Christof Monz. Association for Computational Linguistics, 2024. p. 1038-1046 (Conference on Machine Translation - Proceedings; Vol. 2024-November).

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Graph Representations for Machine Translation in Dialogue Settings

AU - Krause, Lea

AU - Santamaria, Selene Baez

AU - Kalo, Jan Christoph

PY - 2024

Y1 - 2024

N2 - In this paper, we present our approach to the WMT24 - Chat Task, addressing the challenge of translating chat conversations. Chat conversations are characterised by their informal, ungrammatical nature and strong reliance on context posing significant challenges for machine translation systems. To address these challenges, we augment large language models with explicit memory mechanisms designed to enhance coherence and consistency across dialogues. Specifically, we employ graph representations to capture and utilise dialogue context, leveraging concept connectivity as a compressed memory. Our approach ranked second place for Dutch and French, and third place for Portuguese and German, based on COMET-22 scores and human evaluation.

AB - In this paper, we present our approach to the WMT24 - Chat Task, addressing the challenge of translating chat conversations. Chat conversations are characterised by their informal, ungrammatical nature and strong reliance on context posing significant challenges for machine translation systems. To address these challenges, we augment large language models with explicit memory mechanisms designed to enhance coherence and consistency across dialogues. Specifically, we employ graph representations to capture and utilise dialogue context, leveraging concept connectivity as a compressed memory. Our approach ranked second place for Dutch and French, and third place for Portuguese and German, based on COMET-22 scores and human evaluation.

UR - http://www.scopus.com/inward/record.url?scp=85213410525&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85213410525&partnerID=8YFLogxK

UR - https://aclanthology.org/volumes/2024.wmt-1/

U2 - 10.18653/v1/2024.wmt-1.106

DO - 10.18653/v1/2024.wmt-1.106

M3 - Conference contribution

AN - SCOPUS:85213410525

T3 - Conference on Machine Translation - Proceedings

SP - 1038

EP - 1046

BT - Proceedings of the Ninth Conference on Machine Translation

A2 - Haddow, Barry

A2 - Kocmi, Tom

A2 - Koehn, Philipp

A2 - Monz, Christof

PB - Association for Computational Linguistics

T2 - 9th Conference on Machine Translation, WMT 2024

Y2 - 15 November 2024 through 16 November 2024

ER -

Krause L , Santamaria SB , Kalo JC. Graph Representations for Machine Translation in Dialogue Settings. In Haddow B, Kocmi T, Koehn P, Monz C, editors, Proceedings of the Ninth Conference on Machine Translation. Association for Computational Linguistics. 2024. p. 1038-1046. (Conference on Machine Translation - Proceedings). Epub 2024 Nov. doi: 10.18653/v1/2024.wmt-1.106

Graph Representations for Machine Translation in Dialogue Settings

Abstract

Publication series

Conference

Bibliographical note

Funding

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this