Fraud detection in E-commerce using natural language processing

Abstract

Electronic commerce sometimes referred to as e-commerce is a type of business that enables both businesses and private individuals to purchase and sell products and services online. E-commerce in Bangladesh is thriving from the last decade, especially during the coronavirus pandemic with the growth of online sales. Digital commerce is currently struggling to regain trust after allegations of annexation and fraud surfaced against a few firms in recent months. Over 11.48% clients of the internet business area were beguiled last year from di↵erent web based business and Facebook trade (business) sites. Fake reviews are one of the most prominent fraudulent activities in this field. When we try to buy anything online or book any hotel from an app or a ride from any ride sharing app we heavily rely on the reviews of past customers.It makes the decision making process easier. This is why, with the ongoing development of e-commerce platforms online reviews are seen as essential to upholding a company’s reputation. Generally a positive feedback from a customer gathers the attraction of many searching for the same product. For this reason, many e-commerce sites are generating fake reviews to attract more customers towards them. Detecting fake reviews is an ongoing research area. As all the reviews are not trustworthy and honest, it is crucial for us to develop techniques for detecting fake reviews. We are proposing a machine learning approach to generate and detect fake reviews.We used Natural Language Processing(NLP) to extract meaningful features from a text for detecting fraud reviews. Therefore, in this study, we present a comprehensive and e↵ective framework that enhances the e