SENTIMENT ANALYSIS OF EMOTICON BASED NEURO FUZZY SYSTEM

Abstract

Virtual societal space has become an integral part of human life, in which opinion generated by others plays a significant role in one's actions. Tweets as a point of the source have emerged as a way of expressing users' views on a particular subject. Historically sentiment analysis of tweets focused on polarized views, but in the real world, the opinion on a topic usually quite diverse. Therefore, using the classical classification methods cannot adequately fit the human sentiment, and we require a robust way that can match the broad scale of opinion-making. Present work describes an approach that can fit the opinion-making scale of human sentiment by using a fuzzy neuro system to train the model for opinionmaking. Emoticon used in tweets provides an additional layer with classical textual analysis to incorporate complex emotions that can be easily conveyed by emoticons. The scored sentiment is then classified using various machine learning algorithms to measure its accuracy.