Predicting the Success of Domain Adaptation in Text Similarity

Nicolai Pogrebnyakov; Shohreh Shaghaghian

Predicting the Success of Domain Adaptation in Text Similarity

Nicolai Pogrebnyakov^*, Shohreh Shaghaghian

^*Corresponding author for this work

Department of International Economics, Government and Business

Thomson Reuters Canada

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

Transfer learning methods, and in particular domain adaptation, help exploit labeled data in one domain to improve the performance of a certain task in another domain. However, it is still not clear what factors affect the success of domain adaptation. This paper models adaptation success and selection of the most suitable source domains among several candidates in text similarity. We use descriptive domain information and cross-domain similarity metrics as predictive features. While mostly positive, the results also point to some domains where adaptation success was difficult to predict.

Original language	English
Title of host publication	The 6th Workshop on Representation Learning for NLP : Proceedings of the Workshop (REPL4NLP 2021)
Editors	Anna Rogers, Iacer Calixto, Iacer Calixto, Ivan Vulic, Naomi Saphra, Nora Kassner, Oana-Maria Camburu, Trapit Bansal, Vered Shwartz
Number of pages	7
Place of Publication	Stroudsburg, PA
Publisher	Association for Computational Linguistics
Publication date	2021
Pages	206-212
ISBN (Print)	9781954085725
ISBN (Electronic)	9781954085725
Publication status	Published - 2021

Bibliographical note

Part of the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (ACL-IJCNLP 2021).

Access to Document

https://arxiv.org/ftp/arxiv/papers/2106/2106.04641.pdfLicence: Other

Handle.net

Persistent link

Cite this

Pogrebnyakov, N., & Shaghaghian, S. (2021). Predicting the Success of Domain Adaptation in Text Similarity. In A. Rogers, I. Calixto, I. Calixto, I. Vulic, N. Saphra, N. Kassner, O.-M. Camburu, T. Bansal, & V. Shwartz (Eds.), The 6th Workshop on Representation Learning for NLP: Proceedings of the Workshop (REPL4NLP 2021) (pp. 206-212). Association for Computational Linguistics. https://arxiv.org/ftp/arxiv/papers/2106/2106.04641.pdf

Pogrebnyakov, Nicolai ; Shaghaghian, Shohreh. / Predicting the Success of Domain Adaptation in Text Similarity. The 6th Workshop on Representation Learning for NLP: Proceedings of the Workshop (REPL4NLP 2021). editor / Anna Rogers ; Iacer Calixto ; Iacer Calixto ; Ivan Vulic ; Naomi Saphra ; Nora Kassner ; Oana-Maria Camburu ; Trapit Bansal ; Vered Shwartz. Stroudsburg, PA : Association for Computational Linguistics, 2021. pp. 206-212

@inproceedings{6673131bcf6742d3bda1bc0a02f58687,

title = "Predicting the Success of Domain Adaptation in Text Similarity",

abstract = "Transfer learning methods, and in particular domain adaptation, help exploit labeled data in one domain to improve the performance of a certain task in another domain. However, it is still not clear what factors affect the success of domain adaptation. This paper models adaptation success and selection of the most suitable source domains among several candidates in text similarity. We use descriptive domain information and cross-domain similarity metrics as predictive features. While mostly positive, the results also point to some domains where adaptation success was difficult to predict.",

author = "Nicolai Pogrebnyakov and Shohreh Shaghaghian",

note = "Part of the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (ACL-IJCNLP 2021).",

year = "2021",

language = "English",

isbn = "9781954085725",

pages = "206--212",

editor = "Anna Rogers and Iacer Calixto and Iacer Calixto and Ivan Vulic and Naomi Saphra and Nora Kassner and Oana-Maria Camburu and Trapit Bansal and Vered Shwartz",

booktitle = "The 6th Workshop on Representation Learning for NLP",

publisher = "Association for Computational Linguistics",

address = "United States",

}

Pogrebnyakov, N & Shaghaghian, S 2021, Predicting the Success of Domain Adaptation in Text Similarity. in A Rogers, I Calixto, I Calixto, I Vulic, N Saphra, N Kassner, O-M Camburu, T Bansal & V Shwartz (eds), The 6th Workshop on Representation Learning for NLP: Proceedings of the Workshop (REPL4NLP 2021). Association for Computational Linguistics, Stroudsburg, PA, pp. 206-212. <https://arxiv.org/ftp/arxiv/papers/2106/2106.04641.pdf>

Predicting the Success of Domain Adaptation in Text Similarity. / Pogrebnyakov, Nicolai; Shaghaghian, Shohreh.
The 6th Workshop on Representation Learning for NLP: Proceedings of the Workshop (REPL4NLP 2021). ed. / Anna Rogers; Iacer Calixto; Iacer Calixto; Ivan Vulic; Naomi Saphra; Nora Kassner; Oana-Maria Camburu; Trapit Bansal; Vered Shwartz. Stroudsburg, PA: Association for Computational Linguistics, 2021. p. 206-212.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Predicting the Success of Domain Adaptation in Text Similarity

AU - Pogrebnyakov, Nicolai

AU - Shaghaghian, Shohreh

N1 - Part of the Joint Conference of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (ACL-IJCNLP 2021).

PY - 2021

Y1 - 2021

N2 - Transfer learning methods, and in particular domain adaptation, help exploit labeled data in one domain to improve the performance of a certain task in another domain. However, it is still not clear what factors affect the success of domain adaptation. This paper models adaptation success and selection of the most suitable source domains among several candidates in text similarity. We use descriptive domain information and cross-domain similarity metrics as predictive features. While mostly positive, the results also point to some domains where adaptation success was difficult to predict.

AB - Transfer learning methods, and in particular domain adaptation, help exploit labeled data in one domain to improve the performance of a certain task in another domain. However, it is still not clear what factors affect the success of domain adaptation. This paper models adaptation success and selection of the most suitable source domains among several candidates in text similarity. We use descriptive domain information and cross-domain similarity metrics as predictive features. While mostly positive, the results also point to some domains where adaptation success was difficult to predict.

M3 - Article in proceedings

SN - 9781954085725

SP - 206

EP - 212

BT - The 6th Workshop on Representation Learning for NLP

A2 - Rogers, Anna

A2 - Calixto, Iacer

A2 - Vulic, Ivan

A2 - Saphra, Naomi

A2 - Kassner, Nora

A2 - Camburu, Oana-Maria

A2 - Bansal, Trapit

A2 - Shwartz, Vered

PB - Association for Computational Linguistics

CY - Stroudsburg, PA

ER -

Pogrebnyakov N, Shaghaghian S. Predicting the Success of Domain Adaptation in Text Similarity. In Rogers A, Calixto I, Calixto I, Vulic I, Saphra N, Kassner N, Camburu OM, Bansal T, Shwartz V, editors, The 6th Workshop on Representation Learning for NLP: Proceedings of the Workshop (REPL4NLP 2021). Stroudsburg, PA: Association for Computational Linguistics. 2021. p. 206-212