International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 08 Issue: 01 | Jan 2021 www.irjet.net p-ISSN: 2395-0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 384 Scrutiny of Fraudulent Product Reviews and Approach to Filtrate the Aforementioned Chinmay Gadkari 1 , Saad Ahmed 2 , Dhanashree Darekar 3 , Devavrat Agnihotri 4 , Reena Pagare 5 1-4 Students, Dept. of Computer Science & Engineering, School of Engineering, MIT-ADT University, Pune, India 5 Professor, Dept. of Computer Science & Engineering, School of Engineering, MIT-ADT University, Pune, India ---------------------------------------------------------------------***---------------------------------------------------------------------- Abstract - The branding and promotion of product reviews is critical for online stores. They help build trust and loyalty, and typically establish the difference between your products and others. Since consumer interest could disrupt, fake reviews are a big challenge to websites and product search engines. While fake evaluations are good to everyone in the long run, the rise in this doubtful tactic is very fascinating. The fact that huge amounts of time, energy and resources are being poured into fake review spamming is an indicator of just how valuable it can be to have a decent number of reviews. In this manuscript, a conventional method is proposed for elimination of Fake Reviews on ecommerce sites. The solution is to use the reviewers' behavior features with review parameters such as the username, IP address and duplicate reviews to delete fake users. Key Words: Fake Review, LCS, IP Address, Duplicate, Rating, Bias 1. INTRODUCTION Over the last decade, online reviews have been becoming an ever-more-common part of consumers’ purchasing decisions. But, with reviews now being a huge part of online search results, a Fake Review can potentially disrupt the Customer interest. Online Product reviews are a preeminent for people to make decision buying products online. Availability of many similar products makes it difficult for a person to find out which one is the best for the buck, so relying on reviews is a must. As anyone can write a review and get away with it, an increase in fake and spam reviews has been seen, fabricated to look original in order to manipulate the market. Fake Reviews has caught a lot of attention lately. Specifically, the reviews that have been written either to popularize or benefit a brand or a product (therefore expressing a positive sentiment for a product) are called positive deceptive review spams. Whereas, the reviews that intend to malign or defame a competing product expressing a negative sentiment towards the product, are called negative deceptive review spams. The main contributions of this paper are as follows: 1) Supervised method based on Longest Common Substring(LCS) algorithm in order to remove duplicate or near-duplicate reviews, i.e., fake reviews. The model calculates the likeness of a review that can be generated from another one. 2) Review relevancy is also checked if the review is related to the product/brand is not. 3) Finally, user data and review data like account used and IP address is used to detect Fake ones. The remaining portion of the Paper is divided as follows. Section 2 for relevant previous works. Section 3 highlights on the Survey conducted by BrightLocal of Online Reviews Statistics. Section 4 classifies the use of Longest Common Subsubsequence Algorithm(LCS) for Similar Reviews, Later in Section it shows how use of Account data, Unique review ID, IP address can help detect fake Reviews. Section 5 concludes the work done and suggests direction for possible future work and improvements. 2. RELEVANT PRECEDENTS There should be a set of meaningful, relevant previous work to allow any product to eventually be implemented and executed. The useful assessment of previous works that used comparable technology layer to bring similar benefits to the end consumer enables us to create more polished system. There are some works available today on the internet which provide good functionalities to detect spam reviews. Below presented are some of the relevant precedents: [1] This Publication suggests the use of Weka Tool for Text Classification and is Completely based on Sentiment Analysis for Detection of Fake Reviews. It also shows us the difference between popular text classifying techniques, so as per the findings Support Vector Machine(SVM) is most accurate at 81.75% but it takes double the time of Naïve Bayes(NB) to build the model and NB is at 81.45% accuracy. Also, detection processes for fake positive reviews and fake negative reviews depend on the best and more accurate method. [2] An automated method is proposed to highlight review spam in product websites using review text based as well as reviewer-based methods. Supervised and unsupervised methods are applied