Comparative Study of Classifiers for Monitoring Fake Reviews of Online Products Using Opinion Mining

Abstract

With the increasing popularity of e-commerce, online dealers seek reviews or opinions from customers regarding the quality and service of their sold products. As the number of customer reviews grows rapidly, potential buyers face difficulties in reading and assessing them to make informed decisions. Unfortunately, some review websites include fake positive reviews, either added by the product companies themselves or submitted by users who have not made a purchase. This situation makes it challenging for users to distinguish genuine reviews from fake ones, leading to a misleading impression of products and potentially impacting online sales negatively. To address this issue, an essential system called “Fake Review Monitoring” is necessary for E-Commerce websites. In this study, we propose three classifiers: Random Forest, Naïve Bayes, and Support Vector Machine (SVM) to detect fake reviews. Through a comparative study of these classifiers, we measure their performances and determine the best classifier. The results demonstrate that the SVM classifier outperforms the other two, making it a promising choice for detecting fake reviews effectively.