==============================================================================================================================================
				SEW-EMBED: Language-Independent Concept Representations from a Semantically Enriched Wikipedia

							Claudio Delli Bovi and Alessandro Raganato
==============================================================================================================================================

This package contains the two embedded augmentations of the vector representations constructed using SEW, the Semantically Enriched Wikipedia.
The sense inventory of SEW-EMBED is BabelNet (http://babelnet.org), the largest multilingual encyclopedic dictionary and semantic network.

Two versions of SEW-EMBED are available:

- W2V, based on the Google News pre-trained Word2Vec embeddings* as external embedded representation;
- NASARI, based on the NASARI embeddings for English** as external embedded representation.

Both versions follow the same file format. For more information please refer to the reference paper.

(*) 	https://code.google.com/archive/p/word2vec
(**) 	http://lcl.uniroma1.it/nasari

Please find below more details on the format: 

==============================================================================================================================================
FORMAT OF THE VECTOR REPRESENTATION FILES
==============================================================================================================================================

Each vector representation file is in space-separated format, with a single vector in each line.

The format is as follows:

	  SYNSET VALUE_1  VALUE_2  ...  VALUE_N

where SYNSET is the BabelNet synset being represented by the vector, and VALUE_1 ... VALUE_N constitute the numerical components of the vector.
Each vector has 400 dimensions (i.e. N=400) as both the external representations used are based on a 400-dimensional vector space.
When a SYNSET is not covered by SEW-EMBED is represented by an all-zero vector.
 
==============================================================================================================================================
REFERENCE PAPER
==============================================================================================================================================

When using these data, please refer to the following paper:

Claudio Delli Bovi and Alessandro Raganato.
Sew-Embed at SemEval-2017 Task 2: Language-Independent Concept Representations from a Semantically Enriched Wikipedia.
Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 252--257, Vancouver, Canada, 30 July-4 August 2017.


==============================================================================================================================================
CONTACT
==============================================================================================================================================
 
For any enquiry related to SEW, please contact:

- Claudio Delli Bovi (dellibovi [at] di.uniroma1 [dot] it)
- Alessandro Raganato (raganato [at] di.uniroma1 [dot] it) 

==============================================================================================================================================
LICENSES
==============================================================================================================================================

All vector representations constructed from SEW-EMBED are licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 License.