A framework for the automatic detection and evaluation of sentiment in text

Abstract

The aim of this project is to develop a framework for sentiment analysis. Sentiment Analysis is a fast growing area in NLP with a big number of possible applications amongst which are product review mining, trend analysis of blogs and question answer systems for non-factual questions. The framework is designed to be extensible and as much as possible independent of any current sentiment analysis technique. Modules are developed that perform specific Sentiment Analysis tasks that can be plugged in the framework. Some modules that can be plugged into this framework are a module to determine the polarity of opinions and a module to determine the strength of opinions. Apart from developing the framework novel and alternative methods and techniques for some of the tasks involved in the Sentiment Analysis are investigated.