Evaluating the Performance of a Novel Rating System Based on Text Reviews

Abstract

Now a day’s sentiment analysis has become one of the fascinating domains in which each and every online shopping sites use this technique for gathering the product ratings or reviews based on various customers’ sentiment. As we all know that in current days there were a lot of web sites evolved for providing feedback for the online products purchase. But there is no single accurate website which can provide the exact rating from the text reviews. For any web sites almost all the reviews are facing with overloading problem with continuous duplicate reviews that were posted by the same user. In this proposed paper, we mainly designed a novel sentiment-based rating prediction method (NRPS) to improve prediction accuracy in recommender systems. As an extension we also included a concept like benchmark for the rated products based on text reviews. By conducting various experiments on our proposed model by taking a sample local website by taking some products into assumption, our simulation results show our NRPS greatly helps to improve the recommendation performance.