RAPID PSO BASED FEATURES SELECTION FOR CLASSIFICATION

Abstract

Sentiment analysis is one of the most promising research areas in the field of data analysis. In sentiment analysis we do classification of the data samples into positive, negative or neutral classes and bring out proper conclusions from the data. From movie reviews to the twitter data and many others we can impose sentiment analysis. There are many classification models that can be used for sentiment analysis. Some of these models are effective in doing the analysis and some are not. Though the effectiveness and efficiency of the analysis depend on many factors one of the major factors is the underlying feature selection process of the classification models. The features are the attributes of the data samples. Instead of classify the data samples based on the complete set of attributes one can build a model which uses the subset of attribute. Our proposed framework integrates a metaheuristic optimization technique which is a variation of Particle Swarm Optimization which we have named as RAPID-PSO with the multiple classification models which extract a subset of features from the complete set of features. Based on this sub set of features the framework farther classifies the samples. The result shows that our proposed frame work outperforms the other frameworks that use other classical optimization techniques such as –Grid search, Gradient Descend, Classical-PSO, Multiple-PSO, IPSO in terms of effectiveness and efficiency.