Abstract: Now days, mobile App is a very popular and well known concept due to the rapid advancement in the mobile technology and mobile devices. Due to the large number of mobile Apps, ranking fraud is the key challenge in front of the mobile App market. Ranking fraud refers to fraudulent or vulnerable activities which have a purpose of bumping up the Apps in the popularity list. In fact, it becomes more and more frequent for App developers to use tricky means, like increasing their Apps’ sales or posting fake App ratings, to commit ranking fraud. While the importance and necessity of preventing ranking fraud has been widely recognized. After understanding the details of ranking fraud and the need of ranking fraud detection, the paper proposes a ranking fraud detection system for mobile Apps. The proposed system mines the active periods such as leading sessions of mobile apps to accurately locate the ranking fraud. These leading sessions can be useful for detecting the local anomaly instead of global anomaly of App rankings. Besides this, by modeling Apps ranking, rating and review behaviours using statistical hypotheses tests, we investigate three types of evidences, they are ranking based evidences, rating based evidences and review based evidences. Furthermore, we propose an aggregation method based on optimization to integrate all the evidences for fraud detection. Finally, the proposed system will be evaluated with real-world App data which is to be collected from the App Store for a long time period.
Keywords: Mobile Apps, ranking fraud detection, evidence aggregation, historical ranking records, rating and review.