Entities, Dates, and Languages: Zero-Shot on Historical Texts with T0
Francesco de Toni
(1)
,
Christopher Akiki
(2)
,
Javier de La Rosa
(3)
,
Clémentine Fourrier
(4)
,
Enrique Manjavacas
(5)
,
Stefan Schweter
(6)
,
Daniel van Strien
(7)
Clémentine Fourrier
- Fonction : Auteur
- PersonId : 182193
- IdHAL : clefourrier
- ORCID : 0000-0001-6500-6030
Résumé
In this work, we explore whether the recently demonstrated zero-shot abilities of the T0 model extend to Named Entity Recognition for out-of-distribution languages and time periods. Using a historical newspaper corpus in 3 languages as test-bed, we use prompts to extract possible named entities. Our results show that a naive approach for prompt-based zero-shot multilingual Named Entity Recognition is error-prone, but highlights the potential of such an approach for historical languages lacking labeled datasets. Moreover, we also find that T0-like models can be probed to predict the publication date and language of a document, which could be very relevant for the study of historical texts.
Domaines
Linguistique Intelligence artificielle [cs.AI] Informatique et langage [cs.CL] Base de données [cs.DB]Format du dépôt | Notice |
---|---|
Type de dépôt | Communication dans un congrès |
Titre |
en
Entities, Dates, and Languages: Zero-Shot on Historical Texts with T0
|
Résumé |
en
In this work, we explore whether the recently demonstrated zero-shot abilities of the T0 model extend to Named Entity Recognition for out-of-distribution languages and time periods. Using a historical newspaper corpus in 3 languages as test-bed, we use prompts to extract possible named entities. Our results show that a naive approach for prompt-based zero-shot multilingual Named Entity Recognition is error-prone, but highlights the potential of such an approach for historical languages lacking labeled datasets. Moreover, we also find that T0-like models can be probed to predict the publication date and language of a document, which could be very relevant for the study of historical texts.
|
Auteur(s) |
Francesco de Toni
1
, Christopher Akiki
2
, Javier de La Rosa
3
, Clémentine Fourrier
4
, Enrique Manjavacas
5
, Stefan Schweter
6
, Daniel van Strien
7
1
UWA -
The University of Western Australia
( 4177 )
- 35 Stirling Highway
Perth WA 6009 Australia
- Australie
2
Leipzig University / Universität Leipzig
( 562203 )
- Augustusplatz 10, 04109 Leipzig, Allemagne
- Allemagne
3
National Library of Norway
( 1093991 )
- Norvège
4
ALMAnaCH -
Automatic Language Modelling and ANAlysis & Computational Humanities
( 482775 )
- France
5
Universiteit Leiden = Leiden University
( 121723 )
- Leiden University | 2300 RA Leiden
The Netherlands
- Pays-Bas
6
BSB -
Bayerische Staatsbibliothek
( 564889 )
- Bibliothèque de l'État de Bavière à Münich
- Allemagne
7
British Library
( 354279 )
- 96 Euston Road
London
NW1 2DB
- Royaume-Uni
|
Pays |
Irlande
|
Langue du document |
Anglais
|
Vulgarisation |
Non
|
Comité de lecture |
Oui
|
Invité |
Non
|
Audience |
Internationale
|
Actes |
Oui
|
Date début congrès |
2022-05-27
|
Ville |
Dublin
|
Titre du congrès |
BigScience 2022 - International Workshop on Challenges & Perspectives in Creating Large Language Models 2022
|
Domaine(s) |
|
arXiv Id | 2204.05211 |
Loading...