An efficient sentiment analysis using novel bioinspired feature selection and fuzzy classification

Abstract

Sentiment analysis otherwise known as opinion mining is a result of the increased use of internet along with the sudden spurt of online review sites and social media Sentiment or opinion mainly depends upon what general public think or comment and this generally includes products services policies and even politics where they opine either positively or negatively and the opinion is shared by users of a particular product or service The amount of data generated is huge and analysis of the same falls under the category of big data Big data is now part of every sector and function of the global economy Selecting a sub set of variables from the input which delineate input data in such a way as to decrease the effect of noise or inappropriate variables and yet provide efficiently better prediction outcomes is the focus of feature selection In sentiment analysis extracting features is the most difficult task as it needs the usage of Natural Language Processing NLP methods for automated identification of features in the opinions being analyzed. As feature selection aids in saving classification expenses with respect to time as well as computation load it has gained a lot of prominence Here the Term Frequency Inverse Document Frequency TF IDF feature extraction method is used In the representation of the TF IDF the term frequency for all the words are normalized by using the inverse document frequency that will bring down the weight relating to the frequency of occurrence in any collection The central theme here is using the Information Gain IG based feature selection for opinion mining newline