==============
SentiSense Tools

This document (May, 2012) is included in the SentiSense Tools folder which is available for research purposes in the url
http://nil.fdi.ucm.es/index.php?q=node/456

==============
Citation Info

The SentiSense Tools were presented in the paper:

Carrillo de Albornoz J., Plaza L., Gervs P. 2012. SentiSense: An easily scalable concept-based affective lexicon for Sentiment Analysis. The 8th International Conference on Language Resources and Evaluation (LREC 2012)

@inproceedings { 481,
	title = {SentiSense: An easily scalable concept-based affective lexicon for Sentiment Analysis},
	booktitle = {The 8th International Conference on Language Resources and Evaluation (LREC 2012). To appear},
	year = {2012},
	author = {Carrillo de Albornoz, Jorge and Plaza, Laura and Gervs, Pablo}
}

Please see the paper for further details.

==============
Contact

If you have any problem, comment or suggestion, please write us:
Jorge Carrillo de Albornoz: jcalbornoz@lsi.uned.es
Laura Plaza: lplazam@fdi.ucm.es

==============
Content:

* SentiSente_Tagger v0.2
* SentiSense_Visualizer v0.2
* Perl
* README.txt

==============
Manual

A complete manual about the tools and the SentiSense Concept will be available soon.

==================
SentiSense Tagger
==================

In order to allow users to easily expand the SentiSense Affective Lexicon, we have implemented this tagging software. SentiSense Tagger allows annotators to select a data set from which to collect the vocabulary to be labeled, provided that the data set conforms to the format of SentiSense Tagger. This tool is intended for final users to expand SentiSense Affective Lexicon and adapt it to their application domains.

Dependencies
======================
The SentiSense Tagger is a multi-platform which need the following software installed in your system in order to run:

* Java jre 1.5 o higher.
* WordNet 2.1 and the WNHOME environment variable set to the location where you unpacked your WordNet distribution (it defaults to "/usr/local/WordNet-2.1" on Unix and "C:\textbackslash Program Files\textbackslash WordNet\textbackslash 2.1" on a Windows). SentiSense i is supposed to work with all WordNet versions always the appropriate WordNet Sense-Relate perl package and dependencies are properly installed in the system, but only tested with 2.1.
* Perl and the PERLHOME environment variable set to the location where you instaled your Perl distribution (it defaults to "/usr/bin/perl" on Unix and "C:\Perl" on a Windows without the last slash). Also you should include the "bin" directory of the Perl distribution in the path variable.
* WordNet::SenseRelate perl package. Notice that this package need a specific WordNet::Similarity perl package depending on the WordNet version used. Here you can find all the information about the dependencies: http://wn-similarity.sourceforge.net/. As the installation of the this package, and its dependencies, is quite complicate, we provide a Perl installation already with the needed perl packages. This Perl distribution is for Windows. Just copy the Perl folder to you system and add the PERLHOME environment variable and add the "bin" folder to your path variable.

Running SentiSense Tagger
=========================

Just double click on the executable jar. Additionaly, you can use the "java  -Xmx 1500m -jar SentiSense_Tagger.jar" command to increase the java heap and avoid java.lang.OutOfMemoryError: Java heap space exception in very large corpora.

======================
SentiSense Visualizer
======================
SentiSense Visualizer shows, for each synset in the lexicon, its synset identifier and the words or terms that compose the synset. When a synset is selected, its emotional category is shown. Moreover, the application permits to change the emotional category associated to a given synset. The bottom right corner also displays statistics of the number of synset/word within the lexicon and their distribution among different emotional categories and parts of speech.

Dependencies
======================
The SentiSense Visualizer is a multi-platform which need the following software installed in your system in order to run:

* Java jre 1.5 o higher.
* WordNet 2.1 and the WNHOME environment variable set to the location where you unpacked your WordNet distribution (it defaults to "/usr/local/WordNet-2.1" on Unix and "C:\textbackslash Program Files\textbackslash WordNet\textbackslash 2.1" on a Windows). SentiSense i is supposed to work with all WordNet versions always the appropriate WordNet Sense-Relate perl package and dependencies are properly installed in the system, but only tested with 2.1.

Running SentiSense Visualizer
=========================

Just double click on the executable jar. 

======================