==============================================================================================================================================
				SEW: Automatic Construction and Evaluation of a Large Semantically Enriched Wikipedia

					Alessandro Raganato, Claudio Delli Bovi and Roberto Navigli
==============================================================================================================================================

This package contains SEW, a sense-annotated corpus automatically built from Wikipedia. Two versions of SEW are available:

- "complete": The complete version comprises annotations gathered from all the heuristics (including those with overlapping mentions)
- "conservative": The conservative version includes only one sense annotation per tagged mention and no overlap.

For more information please refer to Section 3 (Building a Semantically Enriched Wikipedia) of the reference paper.
The sense inventory of SEW is BabelNet (http://babelnet.org), the largest multilingual encyclopedic dictionary and semantic network.

NOTE:
- Each Wikipedia article is stored in an individual XML file named with the corresponding page title.
- Text enclosed within tags has been escaped using XML entities. In order to retrieve the actual human-readable character we highly suggest to unescape it.
  (e.g. in Java, you can use the StringEscapeUtils class provided by the apache commons-lang API).

Please find below more details on the XML format of the sense-annotated files: 

==============================================================================================================================================
FORMAT OF THE XML FILES
==============================================================================================================================================

Each file contains a "wikiArticle" tag, with the attributes "language" (ISO code of the language) and "title" (page title in Wikipedia).
The wikiArticle tag contains two main tags:

  - "text": The plain text of the Wikipedia article
  - "annotations": A list of "annotation" tags 


An individual "annotation" tag refers to a sense annotation provided by SEW. Each annotation includes the following attributes:

  - "babelNetID":	The unique sense identifier as provided by the sense inventory (BabelNet);
  - "mention":	The surface form of the mention as it appears in the plain text of the article;
  - "anchorStart":	The token-based starting index (inclusive) of the annotation;
  - "anchorEnd":	The token-based ending index (exclusive) of the annotation;
  - "type":	Sense annotation type symbol (see Table 1 of the reference paper).


There are 8 different annotation types:

 - HL: Original Hyperlink
 - SP: Surface Mention Propagation
 - LP: Lemmatized Mention Propagation
 - PP: Person Mention Propagation
 - WIL: Wikipedia Inlink Propagation
 - BIL: BabelNet Inlink Propagation
 - CP: Category Propagation
 - MP: Monosemous Content Word
 
Please refer to the paper (Section 4: Propagation Heuristics) for a more detailed description.

A sample XML file can be found at: http://lcl.uniroma1.it/sew/sample.

==============================================================================================================================================
REFERENCE PAPER
==============================================================================================================================================

When using SEW, please refer to the following paper:

Alessandro Raganato, Claudio Delli Bovi and Roberto Navigli.
Automatic Construction and Evaluation of a Large Semantically Enriched Wikipedia.
Proceedings of 25th International Joint Conference on Artificial Intelligence (IJCAI-16), New York City, New York, USA, 9-15 July 2016.


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

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


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

Except for the original Wikilinks (HL), all the annotations in SEW are licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 License.

Original Wikilinks (HL) are licensed under the Creative Commons Attribution-ShareAlike 3.0  Unported License (CC-BY-SA).