Summarization of User Reviews on E-Commerce Websites Using Hidden Markov Model

Abstract

E-commerce is one of the most important reasons for the growth and sustenance of the internet, with the number of regular users of these sites increasing day by day. The user review forums on e-commerce websites help to get an overview about products of interest. However these reviews are not very organized with respect to whether features of the product, or service provided by the dealers is under discussion. These product reviews can further be positive, negative or a combination of both. This paper discusses a solution to the problem through summarization of product reviews across popular e-commerce websites. The Hidden Markov Model is used to find the component-feature pair, and the SentiWordNet libraries for Weight and Polarity Assignment. After aggregation, the final output is a graph which lists the top six aspects of the product discussed in the review forums along with the weighted polarity representation.