UD_French-Sequoia is an automatic conversion of the SUD_French-Sequoia treebank, which comes from the former corpus French Sequoia corpus.
UD_French-Sequoia is an automatic conversion of the SUD_French-Sequoia treebank, which comes from the former corpus French Sequoia corpus. The French Sequoia corpus was converted in April 2023 to SUD with the Grew rewriting system described here. Since this date, the corpus is maintained in SUD and converted to UD.
The first version of the Sequoia Corpus was presented in (Candito & Seddah, 2012).
The whole corpus contains 70,546 tokens in 3,099 sentences.
In UD_French-Sequoia, data were randomly split into:
fr_sequoia-ud-test.conllu: 10,044 tokens in 456 sentencesfr_sequoia-ud-dev.conllu: 9,999 tokens in 412 sentencesfr_sequoia-ud-train.conllu: 50,503 tokens in 2,231 sentences
The original sentences of the corpus are taken from:
- French Europarl (
sent_idprefix:Europar.550) - Wikipédia Fr (
sent_idprefix:frwiki_50.1000) - Newspaper Est Républicain (
sent_idprefix:annodis.er) - European Medicines Agency (
sent_idprefix:emea-fr-devandemea-fr-test)
The conversion has been performed by Bruno Guillaume with the Graph Rewriting System described here developed by Bruno Guillaume and Guy Perrier.
The Sequoia Corpus was presented in (Candito & Seddah, 2012) and revised later, notably during the project of deep annotation described in (Candito & al. 2014) and (Perrier & al. 2014).
(Candito & Seddah, 2012) Marie Candito, Djamé Seddah. Le corpus Sequoia : annotation syntaxique et exploitation pour l'adaptation d'analyseur par pont lexical. TALN 2012 - 19e conférence sur le Traitement Automatique des Langues Naturelles, Jun 2012, Grenoble, France. 2012.
(Candito & al. 2014) Marie Candito, Guy Perrier, Bruno Guillaume, Corentin Ribeyre, Karën Fort, Djamé Seddah and Éric de la Clergerie. (2014) Deep Syntax Annotation of the Sequoia French Treebank. Proc. of LREC 2014, Reykjavík, Iceland.
(Perrier & al. 2014) Guy Perrier, Marie Candito, Bruno Guillaume, Corentin Ribeyre, Karën Fort and Djamé Seddah. (2014) Un schéma d'annotation en dépendances syntaxiques profondes pour le français. Proc. of TALN 2014, Marseille, France.
(Bonfante & al. 2018) Guillaume Bonfante, Bruno Guillaume, Guy Perrier. Application of Graph Rewriting to Natural Language Processing. ISTE Wiley, 1, pp.272, 2018, Logic, Linguistics and Computer Science Set.
- 2024-05-15 v2.14
- Add missing features (mainly Gender)
- More consistant annotations of idioms wrt other French treebanks
- 2023-05-15 v2.12
- The treebank is now maintained in SUD and converted to UD. See https://github.com/surfacesyntacticud/SUD_French-Sequoia
- 2020-11-14 v2.7
- No major change
- 2020-05-15 v2.6
- Fix some annotation errors found during PARSEME-Fr project
- Fix some inconsistent lemmas and morphological features
- 2019-11-15 v2.5
- Update the conversion process to improve consistency with other French treebanks:
- expletive annotation with relations
expl:subj,expl:compandexpl:pass aux->aux:tenseMWEPOS->EXTPOS
- expletive annotation with relations
- Update the conversion process to improve consistency with other French treebanks:
- 2019-05-15 v2.4
- Several corrections in the original corpus (often found during PARSEME-Fr annotation project)
- Update the conversion process (mainly to handle new validation rules)
- 2018-11-15 v2.3
- Change date annotation
- Improve NOUN/PROPN distinction
- Several corrections in the original corpus and in the conversion process
- 2018-04-15 v2.2
- Subtyping of the
oblrelation with 3 relationsobl:arg,obl:mod,obl:agent - Several corrections in the original corpus and in the conversion process
- Subtyping of the
- 2017-11-15 v2.1
- Manual corrections in the original Treebank
- Application of an updated conversion system, taking into account new decisions taken for harmonisation of several French Treebanks (causative, copules, auxiliaries)
- 2017-03-01 v2.0
- First release in UD