A NOVEL APPROACH ON SENTIMENT ANALYSIS FOR USER REVIEWS IN SOCIAL MEDIA

Abstract Sentiment analysis has been in the forefront in research in machine learning for a couple of decades. The need for sentiment classification arises from the online trading, where customer satisfaction is crucial. Yet, as there is no face-to-face interaction between producer and consumer, feedback is in the form of text reviews, star ratings, comments, discussions on the blogs, so they play an important part in product or service evaluation. Individual social roles in a social network have become more and more important in terms of personalized services. Existing word embedding learning algorithms uses only the contexts but ignore the sentiment of texts. It is problematic for sentiment analysis because the words with similar contexts but opposite sentiment polarity, such as good and bad, are mapped to neighboring word vectors. Apply sentiment embedding's to word-level sentiment analysis, sentence level sentiment classification, and building sentiment lexicons. This work provides the learning of natural language processing and k means. This is where opinion mining and sentiment analysis comes into picture. Sentiment analysis of online reviews and other user generated content is an important research problem for its wide range of applications. In this paper present a survey of different approaches for sentiment analysis and combining them to form a system with best features from several approaches between concept level and aspect level sentiment analysis. Thus further consider different techniques used to perform sentiment analysis and the applications of sentiment analysis in the stand alone systems. In this paper, it is proposed that learning sentiment-specific word embedding's dubbed sentiment embedding's for sentiment analysis and retain the effectiveness of word contexts and exploit sentiment of texts for learning more powerful continuous word representations. By capturing both context and sentiment level evidences, the nearest neighbors in the embedding space are not only semantically similar but also favor to have the same sentiment polarity, so that it is able to separate good and bad to opposite ends of the spectrum.