Rating Based Sentimental prediction: A Six Gram Statistical Mean-Median Approach

Abstract

Revolution in internet has opened a new platform for expressing the user opinions on various aspects openly in various Business domains.  The expressed views include very useful information with which analysis of the user expectation and requirements can be done.  Analyzing the user views and its prediction will help the business people to enhance their services to the customers as per user requirement.  In this paper a novel approach is proposed based on the word corpus used by the customer to express their feelings. Six-Grams model is used as the base for predicting the user rating depending on the classified feedback given by the user. Statistical methods of mean and median are applied by using supervised learning approach for classifying the feedback. Process of classification is done on a rating scale of 1 to 5. The results of the system have shown a better normalized performance over the other statistical methods.  This approach is   experimented with the hotel reviews from the data set available in datafiniti’s  hotel reviews.