Lex Rosetta: Transfer of Predictive Models Across Languages,
Jurisdictions, and Legal Domains

D 2021

Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains

ŠAVELKA, Jaromír, Hannes WESTERMANN, Karim BENYEKHLEF, Charlotte S. ALEXANDER, Jayla C. GRANT et. al.

Basic information

Original name

Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains

Authors

ŠAVELKA, Jaromír (203 Czech Republic, guarantor), Hannes WESTERMANN, Karim BENYEKHLEF, Charlotte S. ALEXANDER, Jayla C. GRANT, Restrepo David AMARILES, Rajaa El HAMDANI, Sébastien MEEÙS, Aurore TROUSSEL, Michal ARASZKIEWICZ, Kevin D. ASHLEY (840 United States of America), Alexandra ASHLEY, Karl BRANTING, Mattia FALDUTI, Matthias GRABMAIR, Jakub HARAŠTA (203 Czech Republic, belonging to the institution), Tereza NOVOTNÁ (203 Czech Republic, belonging to the institution), Elizabeth TIPPETT and Shiwanni JOHNSON

Edition

1. vyd. New York, Eighteenth International Conference on Artificial Intelligence and Law: Proceedings of the Conference, p. 129-138, 10 pp. 2021

Publisher

ACM

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

50501 Law

Country of publisher

United States of America

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

electronic version available online

References:

Text příspěvku Sborník

RIV identification code

RIV/00216224:14220/21:00121820

Organization unit

Faculty of Law

ISBN

978-1-4503-8526-8

DOI

http://dx.doi.org/10.1145/3462757.3466149

Keywords in English

multi-lingual sentence embeddings; transfer learning; domain adaptation; adjudicatory decisions; document segmentation; annotation

Abstract

V originále

In this paper, we examine the use of multi-lingual sentence embeddings to transfer predictive models for functional segmentation of adjudicatory decisions across jurisdictions, legal systems (common and civil law), languages, and domains (i.e. contexts). Mechanisms for utilizing linguistic resources outside of their original context have significant potential benefits in AI & Law because differences between legal systems, languages, or traditions often block wider adoption of research outcomes. We analyze the use of LanguageAgnostic Sentence Representations in sequence labeling models using Gated Recurrent Units (GRUs) that are transferable across languages. To investigate transfer between different contexts we developed an annotation scheme for functional segmentation of adjudicatory decisions. We found that models generalize beyond the contexts on which they were trained (e.g., a model trained on administrative decisions from the US can be applied to criminal law decisions from Italy). Further, we found that training the models on multiple contexts increases robustness and improves overall performance when evaluating on previously unseen contexts. Finally, we found that pooling the training data from all the contexts enhances the models’ in-context performance.

Links

CZ.02.2.69/0.0/0.0/19_073/0016943, interní kód MU
(CEP code: EF19_073/0016943)

Name: Interní grantová agentura Masarykovy univerzity (Acronym: IGA MU)

Investor: Ministry of Education, Youth and Sports of the CR, Priority axis 2: Development of universities and human resources for research and development

Files attached

Savelka_et_al..pdf

Request the author's version of the file

Citovat

ŠAVELKA, Jaromír, Hannes WESTERMANN, Karim BENYEKHLEF, Charlotte S. ALEXANDER, Jayla C. GRANT, Restrepo David AMARILES, Rajaa El HAMDANI, Sébastien MEEÙS, Aurore TROUSSEL, Michal ARASZKIEWICZ, Kevin D. ASHLEY, Alexandra ASHLEY, Karl BRANTING, Mattia FALDUTI, Matthias GRABMAIR, Jakub HARAŠTA, Tereza NOVOTNÁ, Elizabeth TIPPETT and Shiwanni JOHNSON. Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains. Online. In Eighteenth International Conference on Artificial Intelligence and Law: Proceedings of the Conference. 1st ed. New York: ACM, 2021, p. 129-138. ISBN 978-1-4503-8526-8. Available from: https://dx.doi.org/10.1145/3462757.3466149.

@inproceedings{1777697,
   author = {Šavelka, Jaromír and Westermann, Hannes and Benyekhlef, Karim and Alexander, Charlotte S. and Grant, Jayla C. and Amariles, Restrepo David and Hamdani, Rajaa El and Meeùs, Sébastien and Troussel, Aurore and Araszkiewicz, Michal and Ashley, Kevin D. and Ashley, Alexandra and Branting, Karl and Falduti, Mattia and Grabmair, Matthias and Harašta, Jakub and Novotná, Tereza and Tippett, Elizabeth and Johnson, Shiwanni},
   address = {New York},
   booktitle = {Eighteenth International Conference on Artificial Intelligence and Law: Proceedings of the Conference},
   doi = {http://dx.doi.org/10.1145/3462757.3466149},
   edition = {1},
   keywords = {multi-lingual sentence embeddings; transfer learning; domain adaptation; adjudicatory decisions; document segmentation; annotation},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {New York},
   isbn = {978-1-4503-8526-8},
   pages = {129-138},
   publisher = {ACM},
   title = {Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains},
   url = {https://dl.acm.org/doi/10.1145/3462757.3466149},
   year = {2021}
}

TY  - JOUR
ID  - 1777697
AU  - Šavelka, Jaromír - Westermann, Hannes - Benyekhlef, Karim - Alexander, Charlotte S. - Grant, Jayla C. - Amariles, Restrepo David - Hamdani, Rajaa El - Meeùs, Sébastien - Troussel, Aurore - Araszkiewicz, Michal - Ashley, Kevin D. - Ashley, Alexandra - Branting, Karl - Falduti, Mattia - Grabmair, Matthias - Harašta, Jakub - Novotná, Tereza - Tippett, Elizabeth - Johnson, Shiwanni
PY  - 2021
TI  - Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains
PB  - ACM
CY  - New York
SN  - 9781450385268
KW  - multi-lingual sentence embeddings
KW  - transfer learning
KW  - domain adaptation
KW  - adjudicatory decisions
KW  - document segmentation
KW  - annotation
UR  - https://dl.acm.org/doi/10.1145/3462757.3466149
N2  - In this paper, we examine the use of multi-lingual sentence embeddings to transfer predictive models for functional segmentation of adjudicatory decisions across jurisdictions, legal systems (common and civil law), languages, and domains (i.e. contexts). Mechanisms for utilizing linguistic resources outside of their original context have significant potential benefits in AI & Law because differences between legal systems, languages, or traditions often block wider adoption of research outcomes. We analyze the use of LanguageAgnostic Sentence Representations in sequence labeling models using Gated Recurrent Units (GRUs) that are transferable across languages. To investigate transfer between different contexts we developed an annotation scheme for functional segmentation of adjudicatory decisions. We found that models generalize beyond the contexts on which they were trained (e.g., a model trained on administrative decisions from the US can be applied to criminal law decisions from Italy). Further, we found that training the models on multiple contexts increases robustness and improves overall performance when evaluating on previously unseen contexts. Finally, we found that pooling the training data from all the contexts enhances the models’ in-context performance.
ER  -

ŠAVELKA, Jaromír, Hannes WESTERMANN, Karim BENYEKHLEF, Charlotte S. ALEXANDER, Jayla C. GRANT, Restrepo David AMARILES, Rajaa El HAMDANI, Sébastien MEEÙS, Aurore TROUSSEL, Michal ARASZKIEWICZ, Kevin D. ASHLEY, Alexandra ASHLEY, Karl BRANTING, Mattia FALDUTI, Matthias GRABMAIR, Jakub HARAŠTA, Tereza NOVOTNÁ, Elizabeth TIPPETT and Shiwanni JOHNSON. Lex Rosetta: Transfer of Predictive Models Across Languages, Jurisdictions, and Legal Domains. Online. In \textit{Eighteenth International Conference on Artificial Intelligence and Law: Proceedings of the Conference}. 1st ed. New York: ACM, 2021, p.~129-138. ISBN~978-1-4503-8526-8. Available from: https://dx.doi.org/10.1145/3462757.3466149.

Detailed Information on Publication Record