Classify Authentic & Fraudulent Reviews Using Supervised Machine Learning

Abstract

Earlier, if the customer has to buy a product, then that customer has to visit the showroom. For the suggestions and reviews,he would take help from salespersons, friends or family who has already purchased the product. In recent times e-commerce has taken a leap with the availability of technology for everyone and online shopping has become lucrative with the offers provided by the e-commerce sites. Sales of products mostly tend to depend on reviews given by the customer who purchased products previously. Large part of purchasers read reviews of products given by customers or stores before making the decision of what, or from where to buy and whether to buy the product or not. So, in order to retain customers and increase sales with monitory gain, there has been a huge increase in deceptive opinion spam on online review websites.

Basically, fake review or fraudulent review is an untruthful review. Our work is aimed at identifying whether a review and ratings given are fake or truthful one by using supervised machine learning approaches. It analyses the existing feature extraction techniques. It also compares the performance of several experiments done on Amazon Review data set, which is extensive and reputable. We have compared the performance of machine learning classifiers like K-Nearest Neighbors (KNN), Naïve Bayes (NB), Random Forest, Support Vector Model (SVM),Decision tree, XGBoost and Logistic Regression.The results reveal that Logistic Regression outperforms the rest of classifiers in terms of accuracy. The results show that the system has better ability to detect a review as fake or original.

Keywords: Supervised Machine Learning, Fake Reviews, Fake Review Detection, Logistic Regression, Naive Bayes (NB), K-Nearest Neighbors (KNN), Random Forest, Decision tree, XGBoost, Support Vector Model (SVM)