I am a CNRS researcher working in Getalp team (part of Laboratoire d’Informatique de Grenoble). Before that, I was a research scientist at Naver Labs Europe (2019) and a research associate at the University of Edinburgh (2018) in Shay Cohen’s research group. I obtained my PhD in 2017 at Paris Diderot University, which is now part of Université de Paris. Specifically, I worked in the Laboratoire de Linguistique Formelle (LLF), under the supervision of Benoît Crabbé and on the topic of natural language parsing.

My research focuses on Natural Language Processing. Some of my current research interests: syntactic/semantic parsing, structured prediction, data-efficient learning, multitask learning.

Contact: first.last@univ-grenoble-alpes.fr

Publications
Tools and data

Publications

2023:

• On Detecting Policy-Related Political Ads: An Exploratory Analysis of Meta Ads in 2022 French Election.
  Vera Sosnovik, Romaissa Kessi, Maximin Coavoux, Oana Goga.
  TheWebConf2023.
  [pdf] [bib] [preprint]

• BERT Is Not The Count: Learning to Match Mathematical Statements with Proofs.
  Weixian Li, Yftah Ziser, Maximin Coavoux and Shay B. Cohen.
  EACL 2023.
  [pdf] [bib] [preprint] [old preprint version]

2022:

• (In French) Apprentissage profond pour l’estimation du quotient ouvert à partir du signal électroglottographique.
  Minh-Châu Nguyên, Maximin Coavoux, Solange Rossato.
  LIFT 2022.
  [pdf] [bib]

• (In French) Extraction de Phrases Préfabriquées des Interactions à partir d’un corpus arboré du français parlé : une étude exploratoire.
  Marie-Sophie Pausé, Agnès Tutin, Olivier Kraif, Maximin Coavoux.
  CMLF 2022.
  [pdf] [bib]

• End-to-End Dependency Parsing of Spoken French.
  Adrien Pupier, Maximin Coavoux, Benjamin Lecouteux, Jérôme Goulian.
  Interspeech 2022.
  [pdf] [bib] [Adrien’s code]

2021:

• BERT-Proof Syntactic Structures: Investigating Errors in Discontinuous Constituency Parsing.
  Maximin Coavoux.
  Findings of ACL 2021.
  [pdf] [bib] [test suite] [parser’s code]

• (In French) Contribution d’informations syntaxiques aux capacités de généralisation compositionelle des modèles seq2seq convolutifs.
  Diana Nicoleta Popa, William N. Havard, Maximin Coavoux, Eric Gaussier and Laurent Besacier.
  TALN 2021 (short).
  [pdf] [bib]

• Self-Supervised and Controlled Multi-Document Opinion Summarization.
  Hady Elsahar, Maximin Coavoux, Matthias Gallé, Jos Rozen.
  EACL 2021 (long).
  [pdf] [bib] [preprint]

2020:

• (In French) FlauBERT : des modèles de langue contextualisés pré-entraînés pour le français.
  Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
  TALN 2020 (short).
  [pdf] [bib]

• (In French) Qu’apporte BERT à l’analyse syntaxique en constituants discontinus ? Une suite de tests pour évaluer les prédictions de structures syntaxiques discontinues en anglais.
  Maximin Coavoux
  TALN 2020 (short).
  [pdf] [bib]

• FlauBERT: Unsupervised Language Model Pre-training for French.
  Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
  LREC 2020.
  [pdf][bib] [pre-print] [code for FlauBERT] [code for constituency parsing experiments]

2019:

• Unsupervised Aspect-Based Multi-Document Abstractive Summarization.
  Maximin Coavoux, Hady Elsahar, Matthias Gallé.
  NewSum 2019 workshop (short).
  [pdf] [bib]

• Discontinuous Constituency Parsing with a Stack-free Transition System and a Dynamic Oracle.
  Maximin Coavoux, Shay B. Cohen.
  NAACL 2019 (long).
  [pdf] [bib] [code] [slides]

• Unlexicalized Transition-based Discontinuous Constituency Parsing.
  Maximin Coavoux, Benoît Crabbé, Shay B. Cohen.
  TACL 2019.
  [pdf] [html] [bib] [code]

2018:

• Privacy-preserving Neural Representation of Text.
  Maximin Coavoux, Shashi Narayan, Shay B. Cohen.
  EMNLP 2018 (long).
  [pdf] [bib] [code]

2017:

• Multilingual Lexicalized Constituency Parsing with Word-Level Auxiliary Tasks.
  Maximin Coavoux, Benoît Crabbé.
  EACL 2017 (short).
  [pdf] [bib] [code]
• Incremental Discontinuous Phrase Structure Parsing with the GAP Transition.
  Maximin Coavoux, Benoît Crabbé.
  EACL 2017 (long). Outstanding paper award.
  [pdf] [bib] [code] [slides]
• (In French) Représentation et analyse automatique des discontinuités syntaxiques dans les corpus arborés en constituants du français.
  Maximin Coavoux, Benoît Crabbé.
  TALN 2017 (long).
  [pdf] [bib] [code] [slides]
• Cross-lingual RST Discourse Parsing.
  Chloé Braud, Maximin Coavoux, Anders Søgaard.
  EACL 2017 (long).
  [pdf] [bib]

2016:

• (In French) Prédiction structurée pour l’analyse syntaxique en constituants par transitions : modèles denses et modèles creux.
  Maximin Coavoux, Benoît Crabbé.
  Traitement Automatique des Langues, ATALA, 2016, 57 (1).
  [pdf] [bib]
• Neural greedy constituent parsing with dynamic oracles.
  Maximin Coavoux, Benoît Crabbé.
  ACL 2016 (long).
  [pdf] [bib] [code] [slides]

2015:

• (In French) Comparaison d’architectures neuronales pour l’analyse syntaxique en constituants.
  Maximin Coavoux, Benoît Crabbé.
  TALN 2015 (long). Best paper award.
  [pdf] [bib]

Thesis:

• Discontinuous Constituency Parsing of Morphologically Rich Languages.
  Maximin Coavoux. PhD thesis. 2017 [pdf] [bib]

Tools and data

• Bert based discontinuous parser: a discontinuous parser for English and German, based on BERT and released with pretrained models.

• Discontinuous Constituency Treebanks for French
• Wikipedia parsed to constituency trees
• Mind the gap: discontinuous (and projective) constituency parser, released with pretrained models.