Review Classification and False Feedback Detection using Machine Learning

Abstract

Today, a lot of people rely on content available regarding products or services in social media for their decisions for e.g. reviews and opinion before or during sale. This ability to share information with a great number of people with less effort allows spam account users and fake profiles to exploit the client’s trust on the same websites by the usage of spam messages to promote their own blogging activities and advertising material or even make way for phishing, scamming and so on. Because of this system,there is a great possibility that anybody can put up reviews or feedback which in turn provides a gap in security that spammer accounts can use to give out false feedback regarding enterprises and their services Users with fake profiles can easily use their accounts to give false reviews or give fake accounts of the product.The topic also covers the use of fake accounts that give out spamming messages that may cause loss to the victim.To overcome this problem  for detection of fake reviews and unwanted user accounts,the proposed solution uses 3 specific machine learning algorithms namely Naive Bayes , support vector machine(SVM)and Random forest algorithm to go through the data presented and identify whether the message is a fake review or sent from a fake user account.The system takes in data and uses feature engineering presented from the reviews or messages sent from the user account and through the use of different algorithms to identify and differentiate spam reviews from genuine ones .