AdEye: Recognize Advertising Android Apps

In this paper, we introduce our system named AdEye which is used to recognize potential advertising libraries in Android system based on machine learning algorithms. Advertising is the driving force of many websites and service providers on the Internet. With the increasing number of smartphones, there is a fertile market for personalized and localized advertising. In order to fast accurately identify adware, the first step we should do is to research adware. We collected about 40,000 apps from Google Play and the other three markets in China. We repeatedly experiment and analyzed, we designed and implemented AdEye which is high scalability and accuracy ad recognized system. AdEye only needs to use the static method to extract some of the information in AndroidManifest.xml file, which is much faster than a dynamic method and only needs an average of 0.58 seconds to recognize an Android application. Through experimental testing, we find that the accuracy rate of AdEye up to 98.5% and the performance is better than other detection software.