A novel method to find important apps base on the analysis of components relationship

Abstract

With the mobile Internet rapidly developing and the number of mobile applications increasing sharply, the security of the mobile apps has been paid more and more attention in recent years. Many analysis methods for single app have been used in detecting the vulnerability and malicious code. Since mobile apps always related to each other by invoking components, some researchers began to focus on the analysis for multi-applications. But facing with millions of mobile applications, with limited resources, how to improve the ability of security analysis and protection is a difficult problem. For this purpose, we introduce a novel method to mine the correlation among a large number of applications, and finding the nodes that are in the critical position in the process of invoking components. In the proposed method, we first extract the important information from apps and build a database of components. Then, we try to analysis the potential relationship of apps based on the process of invoking components. Moreover, we proposed a novel metric of importance, which can help to find the apps which play important roles in the app-network. We did some experiments to evaluate the proposed method, the experiments show that, we can assess the influence of apps, and figure out the priority of targets during massive application analysis, whether for purpose of detection or protection.