Corpus dell'Italiano annotato con ca. 3000 polirematiche verbali per PARSEME shared task on automatic identification of verbal MWEs

Italian corpus annotated with verbal multiword expressions for the PARSEME shared task on automatic identification of verbal MWEs. The corpus was developed using the Paisà corpus as a basis and it consists of :
- A training corpus manually annotated according to common guidelines (http://parsemefr.lif.univ-mrs.fr/guidelines-hypertext/?page=080_Annotation_management/020_Annotation_platform_FLAT): 15721 sentences from it_blog, it_wikinews and it_wikipedia.
- An evaluation corpus annotated according to the same guidelines (http://parsemefr.lif.univ-mrs.fr/guidelines-hypertext/?page=080_Annotation_management/020_Annotation_platform_FLAT): 1279 sentences from it_blog files.
Extra corpus data
The train and test PARSEME-TSV file is perfectly aligned to a CoNLL-U file.
Lemmas (CoNLL-U): Available (automatically annotated).
POS-tags (CoNLL-U): Available (automatically annotated). The tagset is the ISST-TANL Tagsets.
Morphological features (CoNLL-U): Available (automatically annotated).
Dependency relations (CoNLL-U):Available (automatically annotated). The inventory is Universal Dependency Relations.
No-space information (PARSEME-TSV): Available (automatically annotated).
Tokenisation: The tokenization follows the original tokenization of the PAISÀ corpus
Annotation: VMWEs in this language are annotated for the following categories: LVC, ID, IReflV, VPC, OTH.