Pivotal Sentiment Tree Classifier

Abstract Sentiment Analysis also known as Opinion Mining plays a vital role in social media analytics call center data etc. There are many existing algorithms and methods to approachSentiment Analysis. Though these algorithms produce reasonable results they fail to give near optimal close to 100 results. This paper aims to obtain maximum accuracy in Sentiment Analysis in-comparisonwiththe other existing algorithms and approaches.We devised a new algorithm called Sentiment pivotal tree to achieveresults with maximum accuracy by taking into account on few key factors like identifying the expectations of the customers along with the inclusion of neutral words for analysis.In order to attribute the above factor we introduce a new term called Expectivealong with the existing termsPositive NegativeandNeutral.