Deep-sequoia: A multilayer French corpus

Deep-sequoia is a corpus of French sentences annotated in several projects with a set of annotation layers. It is freely available with the LGPL-LR License. The latest version is the 9.2 released in October 2020 (see the README file)

Annotation layers:

Layer Code Project Version Guidelines
Surface syntax S Sequoia (2012) See Sequoia included in deep syntax annotation guidelines (in French)
Deep syntax D Deep-sequoia (2014) Since 1.0 deep syntax annotation guidelines (in French)
MWEs and Named Entities P PARSEME-FR (2019) Since 9.0 PARSEME-FR guide
Supersense annotation F FRSEMCOR (2020) Since 9.1 FRSEMCOR guide (in French)


Initial version (constituency trees + surface dependencies)

Deep-sequoia: deep syntactic annotations

PARSEME-FR: MWE and named entities annotation

FRSEMCOR: Supersens annotation

, Proc. of LREC 2020, Marseille, France.