An Integrated Approach to Sentiment Analysis using Machine Learning Algorithms

Abstract

Sentiment Analysis (SA) is nothing but mining the emotion from many sources. Some of them include texts, audio, video, etc. Every individual has their own opinion, hence own reviews and ratings. Based on these reviews if we classify the sentiment of the opinion of the public, the profit or loss calculations on the product or application is directly found. Our algorithm takes Naïve Bayes (NB) as a foundation of classification of textual data taken from the public and categorizes the tweets accordingly. To this, we are adding an organic emotion factor called Average Impact Factor (AMF). In the market, there were several algorithms which can be used in mining the sentiment from the given textual data. But this sentiment has flaws as it cannot detect the true emotion from the text or it overfits the opinion of the public. Based on this idea, we integrated the AMF on the public tweets and reviews to evaluate the true sentiment and to improve the time factor too of the opinion mining. We used data of tweets related to Demonetization that happened in India, 2016. When compared to NB Classifier and Support Vector Machine (SVM) algorithms, there is an improvement in time constraint and accuracy too in our Integrated Sentimental Analysis (ISA) classifier.