Detect Fake Reviews Using Random Forest and Support Vector Machine

Abstract

With the rapid development of e-commerce, which makes it possible to buy and sell products and services online, customers are increasingly using these online shop sites to fulfill their needs. After purchase, customers write reviews about their personal experiences, feelings and emotions. Reviews of a product are the main source of information for customers to make decisions to buy or not a product. However, reviews that should be one piece of information that can be trusted by customers can actually be manipulated by the owner of the seller. Where sellers can spam reviews to increase their product ratings or bring down their competitors. Therefore this study discusses detecting fake reviews on productreviews on Tokopedia. Where the method used is the distribution post tagging feature to perform detection. By using the post tagging feature method the distribution got 856 fake reviews and 4478 genuine reviews. In the fake reviews, there were 628 reviews written with the aim of increasing product sales or brand names from store owners, while there were 228 reviews aimed at dropping their competitors or competitors. Furthermore, the classification is carried out using the random forest algorithm model and the support vector machine. By dividing the dataset for training data by 80% while 20% for data testing. Here it is known that the support vector machine gets much higher accuracy than the random forest. The support vector machine gets an accuracy of 98% while the random forest gets an accuracy of 60%