Fake Review Prevention Using Classification and Authentication Techniques

Abstract

Day by day people shopping via E-commerce sites is burgeoning. Decision of placing orders relies on product and service reviews provided by the customers. The importance of the reviews has increased tremendously because they provide information about the quality of product and service. This stimulates sellers to exploit these reviews to increase their sales by deceiving the customers with false information. Thus, detection and prevention of fake reviews becomes pivotal. This paper focuses on detecting and preventing fake reviews using classification and authentication techniques. Review content and Reviewer behavior-based features were used to train different machine learning algorithms such as SVM, Random forest, and Decision tree; among these, Random forest classification algorithm had the highest accuracy of fake review detection with 73.33%. Prevention of fake reviews is achieved by making sure the right person gets to write the review by sending the review writing link to the registered email-id. This research is also concerned over preventing bots to write review by examining the keyboard and mouse activities of the machine. Captcha authentication method has been adopted to prevent bots from writing reviews.