Project outcomes

The goal of our project is to develop linguistic resources (lexicons, corpora, annotation guidelines) and software (parsers, MWE identifiers and linkers). Some of them are currently under development and will be published here when they are ready.

Tools which annotate multiword expressions automatically in running text, developed within the project or in close collaboration with PARSEME-FR project members.

• ATILF-LLF system ("Transition" in shared task 2017)
• VarIDE
• Veyn
• LGTagger
• mwetoolkit
• Seen2Seen et Seen2Unseen (cf. PARSEME shared task 1.2)

Some of these tools can be tested online on the PARSEME-FR demonstrator

• Demonstrator of MWE identifiers and a corpus-lexicon browser
• On-line PARSEME-FR corpus browser on Grew-match
• PARSEME shared task utilities

The datasets of the PARSEME shared task contain 18-20 languages, including French, and can be downloaded from:

• Corpus edition 1.0 (2017) on LINDAT/CLARIN. The French dataset was used in the PARSEME Shared Task on identification of verbal multiword expressions (edition 1.0, 2017). You can also download the French data only from the ORTOLANG platform.
• Corpus edition 1.1 (2018) on LINDAT/CLARIN. The French dataset was used in the PARSEME Shared Task on identification of verbal multiword expressions (edition 1.1, 2018).
• Corpus edition 1.2 (2020) on gitlab (temporary). The French dataset was used in the PARSEME Shared Task on identification of verbal multiword expressions (edition 1.2, 2020).
• Annotation guidelines

• Released on LINDAT/CLARIN as part of the Deep sequoia corpus
• Annotation guidelines

• French corpus annotated both for MWEs and for coreference - data, annotation tools, manual validation results, statistics and reports.

• 4618 sentences with positive and negative examples of 90 selected verbal MWEs in French. The sentences stem from Wikipedia and webcrawling and were taken from the CoNLL shared task corpus.
• Paper describing the construction of this dataset

• Corpus in Basque, German, Greek, Polish and Portuguese
• Paper describing the construction and study of the corpus

• FrenchTAG metagrammar with encoded verbal MWEs
• Paper describing the encoding process and the evaluation

• For project members: PARSEME-FR GitLab