- French FrameNet aka ASFALDA
- Discourse and dialogue related resources
- Discourse multilingual segmentation: Discut (2021),
ToNy (2019, winner of Disrpt 2019 evaluation campaign)
French pretrained model - The discovery dataset, a discourse marker prediction dataset used to fine-tune pretrained language models for semantic/pragmatic tasks, cf Mining Discourse Markers for Unsupervised Sentence Representation Learning (NAACL 2019)
- Lexical item pairs in context, a dataset used for the 2014 ACL paper "Predicting the relevance of distributional semantic similarity with contextual information"
- A corpus of discourse annotated french texts from the Annodis project, along with some dedicated tools.
- Discourse multilingual segmentation: Discut (2021),
ToNy (2019, winner of Disrpt 2019 evaluation campaign)
- Temporal annotation of texts
- the Apetite package, developped with Pascal Denis and Xavier Tannier, provides a set of tools to handle ISO-TimeML annotations, predict temporal structures from timex/event mark-ups, and different ways of evaluating the results. It is licensed under the Cecill, a GPL-like license.
- Experiments on evaluation measures: this archive documents a set of experiments around the evaluation of temporal structures, with reproducing instructions and precomputed results that can be visualized with the Apetite package, and are reported in this JAIR article