Minimalist Entity Disambiguation for Mid-Resource Languages

Benno Kruit

doi:10.18653/v1/2023.sustainlp-1.24

Minimalist Entity Disambiguation for Mid-Resource Languages

Benno Kruit^*

^*Corresponding author for this work

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

Abstract

For many languages and applications, even though enough data is available for training Named Entity Disambiguation (NED) systems, few off-the-shelf models are available for use in practice. This is due to both the large size of state-of-the-art models, and to the computational requirements for recreating them from scratch. However, we observe that in practice, acceptable models can be trained and run with far fewer resources. In this work, we introduce MiniNED, a framework for creating small NED models from medium-sized datasets. The resulting models can be tuned for applicationspecific objectives and trade-offs, depending on practitioners' requirements concerning model size, frequency bias, and out-of-domain generalization. We evaluate the framework in nine languages, and achieve reasonable performance using models that are a fraction of the size of recent work.

Original language	English
Title of host publication	Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP)
Editors	Nafise Sadat Moosavi, Iryna Gurevych, Yufang Hou, Gyuwan Kim, Jin Kim Young, Tal Schuster, Ameeta Agrawal
Publisher	Association for Computational Linguistics (ACL)
Pages	299-306
Number of pages	8
ISBN (Electronic)	9781959429791
DOIs	https://doi.org/10.18653/v1/2023.sustainlp-1.24
Publication status	Published - 2023
Event	4th Workshop on Simple and Efficient Natural Language Processing, SustaiNLP 2023 - Toronto, Canada Duration: 13 Jul 2023 → …

Publication series

Name	Proceedings of the Annual Meeting of the Association for Computational Linguistics
ISSN (Print)	0736-587X

Conference

Conference	4th Workshop on Simple and Efficient Natural Language Processing, SustaiNLP 2023
Country/Territory	Canada
City	Toronto
Period	13/07/23 → …

Bibliographical note

Publisher Copyright:
© 2023 Proceedings of the Annual Meeting of the Association for Computational Linguistics. All rights reserved.

Funding

This research is partially funded by Huawei Amsterdam Research Center. This research is partially funded by Huawei Amsterdam Research Center. I would like to thank Thiviyan Thanapalasingam, Majid Mohammedi and Erman Acar for their early feedback on language-specific models, and the members of the VU Amsterdam Knowledge in AI and Learning & Reasoning groups, Winston Wansleeben, Rens Hassfeld and Mara Spadon for their feedback on drafts of this work.

Funders	Funder number
Huawei Amsterdam Research Center
Knowledge Centre Overweight, EMGO Institute for Health and Care Research, VU University Medical Centre, Amsterdam, The Netherlands. [email protected]

Access to Document

10.18653/v1/2023.sustainlp-1.24Licence: CC BY

Persistent URL (handle)

Persistent link

Cite this

Kruit, B. (2023). Minimalist Entity Disambiguation for Mid-Resource Languages. In N. S. Moosavi, I. Gurevych, Y. Hou, G. Kim, J. K. Young, T. Schuster, & A. Agrawal (Eds.), Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP) (pp. 299-306). (Proceedings of the Annual Meeting of the Association for Computational Linguistics). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/2023.sustainlp-1.24

Kruit, Benno. / Minimalist Entity Disambiguation for Mid-Resource Languages. Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP). editor / Nafise Sadat Moosavi ; Iryna Gurevych ; Yufang Hou ; Gyuwan Kim ; Jin Kim Young ; Tal Schuster ; Ameeta Agrawal. Association for Computational Linguistics (ACL), 2023. pp. 299-306 (Proceedings of the Annual Meeting of the Association for Computational Linguistics).

@inproceedings{abcef3258e7e43f092c68621294651b7,

title = "Minimalist Entity Disambiguation for Mid-Resource Languages",

abstract = "For many languages and applications, even though enough data is available for training Named Entity Disambiguation (NED) systems, few off-the-shelf models are available for use in practice. This is due to both the large size of state-of-the-art models, and to the computational requirements for recreating them from scratch. However, we observe that in practice, acceptable models can be trained and run with far fewer resources. In this work, we introduce MiniNED, a framework for creating small NED models from medium-sized datasets. The resulting models can be tuned for applicationspecific objectives and trade-offs, depending on practitioners' requirements concerning model size, frequency bias, and out-of-domain generalization. We evaluate the framework in nine languages, and achieve reasonable performance using models that are a fraction of the size of recent work.",

author = "Benno Kruit",

note = "Publisher Copyright: {\textcopyright} 2023 Proceedings of the Annual Meeting of the Association for Computational Linguistics. All rights reserved.; 4th Workshop on Simple and Efficient Natural Language Processing, SustaiNLP 2023 ; Conference date: 13-07-2023",

year = "2023",

doi = "10.18653/v1/2023.sustainlp-1.24",

language = "English",

series = "Proceedings of the Annual Meeting of the Association for Computational Linguistics",

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

pages = "299--306",

editor = "Moosavi, {Nafise Sadat} and Iryna Gurevych and Yufang Hou and Gyuwan Kim and Young, {Jin Kim} and Tal Schuster and Ameeta Agrawal",

booktitle = "Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP)",

address = "United States",

}

Kruit, B 2023, Minimalist Entity Disambiguation for Mid-Resource Languages. in NS Moosavi, I Gurevych, Y Hou, G Kim, JK Young, T Schuster & A Agrawal (eds), Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP). Proceedings of the Annual Meeting of the Association for Computational Linguistics, Association for Computational Linguistics (ACL), pp. 299-306, 4th Workshop on Simple and Efficient Natural Language Processing, SustaiNLP 2023, Toronto, Canada, 13/07/23. https://doi.org/10.18653/v1/2023.sustainlp-1.24

Minimalist Entity Disambiguation for Mid-Resource Languages. / Kruit, Benno.
Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP). ed. / Nafise Sadat Moosavi; Iryna Gurevych; Yufang Hou; Gyuwan Kim; Jin Kim Young; Tal Schuster; Ameeta Agrawal. Association for Computational Linguistics (ACL), 2023. p. 299-306 (Proceedings of the Annual Meeting of the Association for Computational Linguistics).

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

TY - GEN

T1 - Minimalist Entity Disambiguation for Mid-Resource Languages

AU - Kruit, Benno

PY - 2023

Y1 - 2023

N2 - For many languages and applications, even though enough data is available for training Named Entity Disambiguation (NED) systems, few off-the-shelf models are available for use in practice. This is due to both the large size of state-of-the-art models, and to the computational requirements for recreating them from scratch. However, we observe that in practice, acceptable models can be trained and run with far fewer resources. In this work, we introduce MiniNED, a framework for creating small NED models from medium-sized datasets. The resulting models can be tuned for applicationspecific objectives and trade-offs, depending on practitioners' requirements concerning model size, frequency bias, and out-of-domain generalization. We evaluate the framework in nine languages, and achieve reasonable performance using models that are a fraction of the size of recent work.

AB - For many languages and applications, even though enough data is available for training Named Entity Disambiguation (NED) systems, few off-the-shelf models are available for use in practice. This is due to both the large size of state-of-the-art models, and to the computational requirements for recreating them from scratch. However, we observe that in practice, acceptable models can be trained and run with far fewer resources. In this work, we introduce MiniNED, a framework for creating small NED models from medium-sized datasets. The resulting models can be tuned for applicationspecific objectives and trade-offs, depending on practitioners' requirements concerning model size, frequency bias, and out-of-domain generalization. We evaluate the framework in nine languages, and achieve reasonable performance using models that are a fraction of the size of recent work.

UR - http://www.scopus.com/inward/record.url?scp=85175863061&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85175863061&partnerID=8YFLogxK

UR - https://aclanthology.org/volumes/2023.sustainlp-1/

U2 - 10.18653/v1/2023.sustainlp-1.24

DO - 10.18653/v1/2023.sustainlp-1.24

M3 - Conference contribution

AN - SCOPUS:85175863061

T3 - Proceedings of the Annual Meeting of the Association for Computational Linguistics

SP - 299

EP - 306

BT - Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP)

A2 - Moosavi, Nafise Sadat

A2 - Gurevych, Iryna

A2 - Hou, Yufang

A2 - Kim, Gyuwan

A2 - Young, Jin Kim

A2 - Schuster, Tal

A2 - Agrawal, Ameeta

PB - Association for Computational Linguistics (ACL)

T2 - 4th Workshop on Simple and Efficient Natural Language Processing, SustaiNLP 2023

Y2 - 13 July 2023

ER -

Kruit B. Minimalist Entity Disambiguation for Mid-Resource Languages. In Moosavi NS, Gurevych I, Hou Y, Kim G, Young JK, Schuster T, Agrawal A, editors, Proceedings of The Fourth Workshop on Simple and Efficient Natural Language Processing (SustaiNLP). Association for Computational Linguistics (ACL). 2023. p. 299-306. (Proceedings of the Annual Meeting of the Association for Computational Linguistics). Epub 2023 Jul. doi: 10.18653/v1/2023.sustainlp-1.24

Minimalist Entity Disambiguation for Mid-Resource Languages

Abstract

Publication series

Conference

Bibliographical note

Funding

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this