A Machine Learning Based Sentiment Analysis by Selecting Features for Predicting Customer Reviews
R. Nagamanjula, A. Pethalakshmi · 2018
Nowadays people can express their opinions and views publicly which can be favour and/or against any service, issue, product, event, or policy. With the rapid advancement of internet, people can share their feedback on the web in huge numbers. This large number of reviews for individuals can be crucial to improve their services and products, called Opinion Mining. It is also known as sentiment analysis which ultimate goal is to differentiate the emotions expressed within the reviews. In this paper, we analysed these reviews and classifying them into positive or negative opinions. This paper presents a novel classification method called Support Vector Machine (SVM) in order to improve the accuracy by forming two classes i.e. positive and negative. Initially, words are collected from social site Amazoon.in. (Reviews about electronic products, especially for mobile brands) and pre-processed using wordnet tool. To classify the review/comment, we selected some of the features which are evaluated from user review comments. The Information gain with Fast Correlation based Filter (FCBF) considered for feature selection and then the SVM classifier is find the classes. An intensive experimental study shows the efficiency of these enhancements and shows better performance in terms of precision, recall and f-measure.