Upload 8 files

.gitattributes

@@ -56,3 +56,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+corpus/train.conllu filter=lfs diff=lfs merge=lfs -text

corpus/README-1-shared-task.md

@@ -0,0 +1,43 @@
+README
+------
+Welcome to the data repository of the [PARSEME Shared Task on Verbal Multi-Word Expressions Identification](http://typo.uni-konstanz.de/parseme/index.php/2-general/142-parseme-shared-task-on-automatic-detection-of-verbal-mwes).
+
+This repository contains corpora in multiple languages.
+Corpora are human-annotated with occurrences of Verbal Multiword Expressions (VMWEs).
+
+You can download this repository as a [ZIP file](https://gitlab.com/parseme/sharedtask-data/repository/archive.zip?ref=master).
+
+
+Corpora
+-------
+Human annotations can be found for multiple language directories,
+such as `FR` for French, `DE` for German, `PL` for Polish, etc.
+Inside each language directory, you may find these files:
+
+* `README.md`: A description of the available data for that given language.
+* ~~`trial.parsemetsv`: Trial file with an example of the final data (released mid-December/2016).~~[^1]
+* `train.parsemetsv`: Training data (to be released early January/2017).
+* `test.blind.parsemetsv`: The blind test data (to be released by the end of January/2017).
+* `test.parsemetsv`: The gold test data (to be released after the shared task)
+* `stats.md`: Number of sentences, tokens and annotated VMWEs in train/test.
+
+The PARSEME-TSV files contain human annotation of VMWEs.
+Depending on the language, these files may also be paired with a CoNLL-U file
+for extra information (check the language-specific `README.md` files).
+
+Note: For some languages, the UPOSTAG and DEPREL fields of their
+CoNLL-U file may contain data that is not in the universal UD format. We thought it
+would be useful to provide this information nevertheless.
+
+
+Scripts
+-------
+The `bin` directory contains useful scripts.
+
+* `bin/evaluate.py`: script for evaluating machine predictions against human annotations.
+* `bin/checkParsemeTsvFormat.py`: script for checking if the prediction file is in the proper format.
+
+--------------------------------
+
+[^1]: Trial data was removed from the current release in order to avoid confusion.
+

corpus/README-2-french-data.md

@@ -0,0 +1,67 @@
+README
+------
+This is the README file for the French language.
+
+
+Corpora
+-------
+The verbal MWEs have been annotated:
+1. `sequoia`: The Sequoia Treebank
+2. `ud_1_4`: The Universal Dependencies 1.4 treebank for French
+
+
+Extra corpus data
+-----------------
+The train and test PARSEME-TSV file is perfectly aligned to a CoNLL-U file. 
+
+* Lemmas (CoNLL-U): Available
+* POS-tags (CoNLL-U): Available (manually annotated). The tagset for UD is [Universal POS-tags](http://universaldependencies.org/u/pos). The tagset for the Sequoia treebank is described in the [Sequoia documentation](http://passage.inria.fr/deepwiki/node/594#Categories_morpho-syntaxiques) (in French).
+* Morphological features (CoNLL-U): Available
+* Dependency relations (CoNLL-U): Available (manually annotated). The inventory is [Universal Dependency Relations](http://universaldependencies.org/u/dep) for UD, and [Sequoia surface dependencies](http://passage.inria.fr/deepwiki/node/594#Fonctions_grammaticales) for the Sequoia treebank. 
+* No-space information (PARSEME-TSV): Available for both sources (automatically annotated).
+
+
+Tokenization
+------------
+* sequoia corpus: the original tokenization of the sequoia corpus has been converted into that of the UD 1.4 French data.
+
+So for both sources we have the UD 1.4 tokenization, in which the following contractions appear as multi-word tokens (e.g. 1-2 au), split into words:
+E.g. : Au soleil
+```
+1-2 Au
+1 à
+2 le
+3 soleil
+```
+
+The list of contractions is:
+```
+au
+auquel
+aux
+auxquelles
+auxquels
+des
+desquelles
+du
+duquel
+```
+
+Note that the only ambiguous case are "des" / "du". Depending on the context, these tokens are either a plain determiner, or are split into preposition "de" + determiner "le" / "les".
+
+Annotation
+----------
+VMWEs in this language are annotated for the following categories: ID, LVC, IReflV, OTH.
+
+Authors
+----------
+The VMWEs annotations were performed by Marie Candito, Mathieu Constant, Caroline Pasquer, Yannick Parmentier, Carlos Ramisch, Jean-Yves Antoine.
+
+Licence
+----------
+The VMEs annotations are distributed under the terms of the [CC-BY v4 license](https://creativecommons.org/licenses/by/4.0/). As far as the CONLL-U files are concerned, the UD part of the corpus is under [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) and the Sequoia part is under [LGPL-LR](http://infolingu.univ-mlv.fr/DonneesLinguistiques/Lexiques-Grammaires/lgpllr.html). UD sentences can be identified by their `sentid` prefixed with `fr-ud`.
+ 
+Contact
+----------
+`marie.candito@gmail.com`
+

corpus/stats.md

@@ -0,0 +1,21 @@
+Statistics FR
+=============
+
+### train.parsemetsv
+* Sentences: 17880
+* Tokens: 450221
+* Total VMWEs: 4462
+  * `ID`: 1786
+  * `IReflV`: 1313
+  * `LVC`: 1362
+  * `OTH`: 1
+
+### test.parsemetsv
+* Sentences: 1667
+* Tokens: 35784
+* Total VMWEs: 500
+  * `ID`: 119
+  * `IReflV`: 105
+  * `LVC`: 271
+  * `OTH`: 5
+

corpus/train.conllu

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b6562fe305399c1c772f92d378196ca64d45edc55b1cf09ec37f9c136e4e4a1
+size 26010247