Language Independent Multi-Class Sentiment Analysis

One of the common approaches for putting user review is social media. In the last year, several attempts have been dedicated to automatically mining view and feelings from natural language in social media posts, news and business product reviews. It is necessary to evaluate the user's emotion by studying the user's input text and feelings with the help of textual communication techniques. Till now the most of the work has been done on identifying the polarity of sentiment is classified into positive, negative , neutral only. Our system provides an emotional determination strategy from the text and emojis entered by the user with direct and indirect tweets. It classifies emotions more than three classes has been identified like happy, sad, angry, sarcasm, neutral using sentiword emoji's based dictionary The proposed method utilizes techniques of extraction of features and extraction of emotions, to determine emotion of the user. Some associated work is reviewed to design the system, which gives the concept of prior work done. It also deals with special words, emoji, direct and indirect emotions expressed in different languages. The classifier which is used is hybrid to avoid drawback of SVM, Naïve bayes and Maximum entropy.