We have the cyber space occupied with most of the opinions comments and reviews. We also see the use of opinions in decision making process of many organizations. Not only organizations use these reviews but also users use them to a great extent. So using this opportunity many groups try to game this system by providing fake reviews. These reviews enhance or demote the emotions of the products they are acting upon. Many of the organizations pay such groups to promote their product and acquire most of the market share. For a genuine user experience these fake reviews should be detected and deleted. Work had been performed on detecting individual fake reviews and individual fake reviewers; but a fake reviewer group is much more damaging as they take the total control of the product sentiments. This project presents a way to detect these fake reviewer groups. This uses indicators and models to calculate the spamicity of the group. This system deals with detecting fake reviewers group rather than individual fake reviewers.
Random forest-SVM-naïve bayes