Outiller l'occitan: nouvelles ressources et lemmatisation

Aleksandra Miletic Haddad

Outiller l'occitan: nouvelles ressources et lemmatisation

Translated title of the contribution: New resources and lemmatization experiments for Occitan

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

Abstract

This paper presents recent contributions to the creation of NLP tools and resources for Occitan. Several existing resources were modified or adapted, in particular a rule-based tokenizer, a morphosyntactic lexicon and a treebank. These resources were used to train and evaluate neural lemmatization models. As part of these experiments, a large corpus based on Wikipedia (2 million tokens) was POS-tagged and lemmatized. This new resource is shared through Zenodo.

Translated title of the contribution	New resources and lemmatization experiments for Occitan
Original language	French
Title of host publication	Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN) : Volume 1: travaux de recherche originaux - articles longs
Editors	Christophe Servan, Anne Vilnat
Number of pages	15
Place of Publication	Paris
Publisher	Association pour le Traitement Automatique des Langues
Publication date	Jun 2023
Pages	217-231
Publication status	Published - Jun 2023
MoE publication type	D3 Professional conference proceedings
Event	Conférence sur le Traitement Automatique des Langues Naturelles - Paris, France Duration: 5 Jun 2023 → 9 Jun 2023 Conference number: 30

Fields of Science

6121 Languages
Occitan
lemmatisation
low-resourced languages

Access to Document

proceedings_TALN_longs_224Final published version, 517 KBLicence: CC BY

Cite this

@inproceedings{dd9f050d266b4ec2bc2324b365a90930,

title = "Outiller l'occitan: nouvelles ressources et lemmatisation",

abstract = "This paper presents recent contributions to the creation of NLP tools and resources for Occitan. Several existing resources were modified or adapted, in particular a rule-based tokenizer, a morphosyntactic lexicon and a treebank. These resources were used to train and evaluate neural lemmatization models. As part of these experiments, a large corpus based on Wikipedia (2 million tokens) was POS-tagged and lemmatized. This new resource is shared through Zenodo.",

keywords = "6121 Spr{\aa}kvetenskaper, Occitan, lemmatisation, low-resourced languages",

author = "{Miletic Haddad}, Aleksandra",

year = "2023",

month = jun,

language = "franska",

pages = "217--231",

editor = "Christophe Servan and Anne Vilnat",

booktitle = "Actes de la 30e Conf{\'e}rence sur le Traitement Automatique des Langues Naturelles (TALN)",

publisher = "Association pour le Traitement Automatique des Langues",

address = "Frankrike",

note = "Conf{\'e}rence sur le Traitement Automatique des Langues Naturelles ; Conference date: 05-06-2023 Through 09-06-2023",

}

Miletic Haddad, A 2023, Outiller l'occitan: nouvelles ressources et lemmatisation. in C Servan & A Vilnat (eds), Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN): Volume 1: travaux de recherche originaux - articles longs. Association pour le Traitement Automatique des Langues, Paris, pp. 217-231, Conférence sur le Traitement Automatique des Langues Naturelles, Paris, France, 05/06/2023.

Outiller l'occitan: nouvelles ressources et lemmatisation. / Miletic Haddad, Aleksandra.
Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN): Volume 1: travaux de recherche originaux - articles longs. ed. / Christophe Servan; Anne Vilnat. Paris: Association pour le Traitement Automatique des Langues, 2023. p. 217-231.

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

TY - GEN

T1 - Outiller l'occitan

T2 - Conférence sur le Traitement Automatique des Langues Naturelles

AU - Miletic Haddad, Aleksandra

N1 - Conference code: 30

PY - 2023/6

Y1 - 2023/6

N2 - This paper presents recent contributions to the creation of NLP tools and resources for Occitan. Several existing resources were modified or adapted, in particular a rule-based tokenizer, a morphosyntactic lexicon and a treebank. These resources were used to train and evaluate neural lemmatization models. As part of these experiments, a large corpus based on Wikipedia (2 million tokens) was POS-tagged and lemmatized. This new resource is shared through Zenodo.

AB - This paper presents recent contributions to the creation of NLP tools and resources for Occitan. Several existing resources were modified or adapted, in particular a rule-based tokenizer, a morphosyntactic lexicon and a treebank. These resources were used to train and evaluate neural lemmatization models. As part of these experiments, a large corpus based on Wikipedia (2 million tokens) was POS-tagged and lemmatized. This new resource is shared through Zenodo.

KW - 6121 Språkvetenskaper

KW - Occitan

KW - lemmatisation

KW - low-resourced languages

UR - https://coria-taln-2023.sciencesconf.org/data/proceedings_TALN_longs.pdf

M3 - Konferensbidrag

SP - 217

EP - 231

BT - Actes de la 30e Conférence sur le Traitement Automatique des Langues Naturelles (TALN)

A2 - Servan, Christophe

A2 - Vilnat, Anne

PB - Association pour le Traitement Automatique des Langues

CY - Paris

Y2 - 5 June 2023 through 9 June 2023

ER -

Outiller l'occitan: nouvelles ressources et lemmatisation

Abstract

Fields of Science

Access to Document

Other files and links

CorCoDial: Corpus-based computational dialectology: exploiting machine translation techniques to extract, visualize and interpret dialectal patterns

OcWikiAnnot: Annotated Wikipedia Corpus of Occitan

Cite this

Outiller l'occitan: nouvelles ressources et lemmatisation

Abstract

Fields of Science

Access to Document

Other files and links

Projects

CorCoDial: Corpus-based computational dialectology: exploiting machine translation techniques to extract, visualize and interpret dialectal patterns

Datasets

OcWikiAnnot: Annotated Wikipedia Corpus of Occitan

Cite this