Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models

Elliot Schumacher; James Mayfield; Mark Dredze

doi:10.18653/v1/2022.mrl-1.4

Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models

Elliot Schumacher, James Mayfield, Mark Dredze

Whiting School of Engineering

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

Abstract

Most entity linking systems, whether mono or multilingual, link mentions to a single English knowledge base. Few have considered linking non-English text to a non-English KB, and therefore, transferring an English entity linking model to both a new document and KB language. We consider the task of zero-shot cross-language transfer of entity linking systems to a new language and KB. We find that a system trained with multilingual representations does reasonably well, and propose improvements to system training that lead to improved recall in most datasets, often matching the in-language performance. We further conduct a detailed evaluation to elucidate the challenges of this setting.

Original language	English (US)
Title of host publication	MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop
Publisher	Association for Computational Linguistics (ACL)
Pages	38-51
Number of pages	14
ISBN (Electronic)	9781959429166
DOIs	https://doi.org/10.18653/v1/2022.mrl-1.4
State	Published - 2022
Event	2nd Workshop on Multi-Lingual Representation Learning, MRL 2022 - Abu Dhabi, United Arab Emirates Duration: Dec 8 2022 → …

Publication series

Name	MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop

Conference

Conference	2nd Workshop on Multi-Lingual Representation Learning, MRL 2022
Country/Territory	United Arab Emirates
City	Abu Dhabi
Period	12/8/22 → …

ASJC Scopus subject areas

Language and Linguistics
Linguistics and Language

Access to Document

10.18653/v1/2022.mrl-1.4

Cite this

Schumacher, E., Mayfield, J., & Dredze, M. (2022). Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models. In MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop (pp. 38-51). (MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/2022.mrl-1.4

Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models. / Schumacher, Elliot; Mayfield, James; Dredze, Mark.
MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop. Association for Computational Linguistics (ACL), 2022. p. 38-51 (MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop).

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

Schumacher, E, Mayfield, J & Dredze, M 2022, Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models. in MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop. MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop, Association for Computational Linguistics (ACL), pp. 38-51, 2nd Workshop on Multi-Lingual Representation Learning, MRL 2022, Abu Dhabi, United Arab Emirates, 12/8/22. https://doi.org/10.18653/v1/2022.mrl-1.4

Schumacher E, Mayfield J, Dredze M. Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models. In MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop. Association for Computational Linguistics (ACL). 2022. p. 38-51. (MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop). doi: 10.18653/v1/2022.mrl-1.4

Schumacher, Elliot ; Mayfield, James ; Dredze, Mark. / Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models. MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop. Association for Computational Linguistics (ACL), 2022. pp. 38-51 (MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop).

@inproceedings{b6e2ad1af7a84de18e9a73cf3226fe3f,

title = "Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models",

abstract = "Most entity linking systems, whether mono or multilingual, link mentions to a single English knowledge base. Few have considered linking non-English text to a non-English KB, and therefore, transferring an English entity linking model to both a new document and KB language. We consider the task of zero-shot cross-language transfer of entity linking systems to a new language and KB. We find that a system trained with multilingual representations does reasonably well, and propose improvements to system training that lead to improved recall in most datasets, often matching the in-language performance. We further conduct a detailed evaluation to elucidate the challenges of this setting.",

author = "Elliot Schumacher and James Mayfield and Mark Dredze",

note = "Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; 2nd Workshop on Multi-Lingual Representation Learning, MRL 2022 ; Conference date: 08-12-2022",

year = "2022",

doi = "10.18653/v1/2022.mrl-1.4",

language = "English (US)",

series = "MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop",

publisher = "Association for Computational Linguistics (ACL)",

pages = "38--51",

booktitle = "MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop",

address = "United States",

}

TY - GEN

T1 - Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models

AU - Schumacher, Elliot

AU - Mayfield, James

AU - Dredze, Mark

PY - 2022

Y1 - 2022

N2 - Most entity linking systems, whether mono or multilingual, link mentions to a single English knowledge base. Few have considered linking non-English text to a non-English KB, and therefore, transferring an English entity linking model to both a new document and KB language. We consider the task of zero-shot cross-language transfer of entity linking systems to a new language and KB. We find that a system trained with multilingual representations does reasonably well, and propose improvements to system training that lead to improved recall in most datasets, often matching the in-language performance. We further conduct a detailed evaluation to elucidate the challenges of this setting.

AB - Most entity linking systems, whether mono or multilingual, link mentions to a single English knowledge base. Few have considered linking non-English text to a non-English KB, and therefore, transferring an English entity linking model to both a new document and KB language. We consider the task of zero-shot cross-language transfer of entity linking systems to a new language and KB. We find that a system trained with multilingual representations does reasonably well, and propose improvements to system training that lead to improved recall in most datasets, often matching the in-language performance. We further conduct a detailed evaluation to elucidate the challenges of this setting.

UR - http://www.scopus.com/inward/record.url?scp=85151097528&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85151097528&partnerID=8YFLogxK

U2 - 10.18653/v1/2022.mrl-1.4

DO - 10.18653/v1/2022.mrl-1.4

M3 - Conference contribution

AN - SCOPUS:85151097528

T3 - MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop

SP - 38

EP - 51

BT - MRL 2022 - 2nd Workshop on Multi-Lingual Representation Learning, Proceedings of the Workshop

PB - Association for Computational Linguistics (ACL)

T2 - 2nd Workshop on Multi-Lingual Representation Learning, MRL 2022

Y2 - 8 December 2022

ER -

Zero-shot Cross-Language Transfer of Monolingual Entity Linking Models

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this