Search This Blog

Sunday, November 29, 2009

Sentiment analysis - Text mining

A new feature is added to TexLexAn. The version 0.34 ( texlexan CLI 0.34 + texlexan GUI 0.30 ) is able to estimate the sentiment polarity and to evaluate the intensity of the sentiment in the text.

The sentiment analysis is based on a knowledge base and a simple lexical analyser. The analyser returns a global rating of the sentiment in the text and extracts each sentence expressing a sentiment.

This program runs under Linux and FreeBSD and is available on sourceforge:
http://sourceforge.net/projects/texlexan/files/

In present time, only English and French languages are implemented. There are two only dictionaries: keyworder.en.dicE and keyworder.fr.dicE in the package (pack 1.42); they are very incomplete and contain only 80 words expressing a sentiment. I will explain in another post how to complete these dictionaries.