Description of the identifiers

LGTagger is an open-source Part-of-speech tagger that also recognizes Multiword units.
It is based on Conditional Random Fields (CRF) and large-coverage lexical resources.
The lexical resources can be composed of morphosyntactic dictionaries (including simple and compound words) and strongly lexicalized local grammars.
It presently works for French.


Web page of the tool with the download link : LGTagger

ATILF-LFF is a generic transition system for the identification of Verbal Multiword Expressions (VMWEs).
This system is a data-driven system applicable to several languages.
It produces a robust, efficient performance and very competitive scores, with respect to the shared task results.
The system was developed and evaluated using the datasets of the PARSEME shared task on VMWE identification
and accommodates the variety of linguistic resources provided for each language, in terms of accompanying morphological and syntactic information.


Link to the Github page : ATILF-LLF

Veyn is a system for automatic identification of multiword expressions in running text
submitted to the PARSEME shared task 2018. The model is first trained on a MWE-annotated corpus,
and then can be applied to any new text to identify MWEs that are similar to those in the training corpus.
Veyn is based on a sequence tagger using recurrent neural networks. As input features it takes the lemmas and POS tags of words.
We represent the output MWEs using a variant of the begin-inside-outside encoding scheme combined with the MWE category tag.


Link to the Github page: Veyn

This system participated in edition 1.1 of the PARSEME shared task on automatic identification of verbal multiword expressions (VMWEs).
Our system focuses on the task of VMWE variant identification by using morphosyntactic information in the training data to predict if candidates extracted from the test corpus could be idiomatic, thanks to a naive Bayes classifier.
Candidate identification also includes their categorization according to the PARSEME guidelines


Link to the Gitlab page: VarIDE