Review-based sentiment prediction of rating using natural language processing sentence-level sentiment analysis with bag-of-words approach

Abstract

User’s opinion plays a vital role in the global world of Internet as they are given freedom to express their feedback. A lot of hidden information and feeling of the user are expressed in the words that he/she used. Extracting this hidden information can help the service industry to better serve the need as per the user acceptance and to outfit the competition in the market. Methodology of this paper is intended to define the rating of the review given by the user. Method of bag-of-words approach taking the sentiment score and magnitude of the sentence using natural language processing is applied. The scale of one to five is considered in this experimental classification on the data set of hotel reviews. The results have shown that around 60% of the ratings can be predicted and 40% unpredicted with the given review. The Data set used in this experimental analysis is the Datafiniti’s hotel reviews.