Sentiment Analysis of Twitter Data Using Multi Class Semantic Approach

Growth in the area of opinion mining and sentiment analysis has been rapid and aims to explore the opinions or text present on different platforms of social media through machine-learning techniques with sentiment, subjectivity analysis or polarity calculations. Social media is a popular network through which user can share their reviews about various topics, news, products etc. People use internet to access or update reviews so it is necessary to express opinion. Sentiment analysis is to classify these reviews based on its opinion as either positive or negative category. First we have preprocessed the dataset to convert unstructured reviews into structured form. Then we have used lexicon based approach to convert structured review into numerical score value. In lexicon based approach we have preprocessed dataset using feature selection and semantic analysis. Stop word removal, stemming, and calculating sentiment score with help of Twitter dataset have been done in preprocessing part. Then we have applied classification algorithm to classify opinion as either positive or negative. Support vector machine algorithm is used to classify reviews where Multi class kernel SVM is modified by its hyper parameters and compared with existing Naivesbayes algorithm. So optimized SVM gives good result than SVM and naïve bayes. At last we have compared performance of all classifier with respect to accuracy.