Using association statistics to rank risk of Android application

• Published in: 2015 IEEE International Conference on Computer and Communications (ICCC) pp. 6 - 10
• Main Authors: Chenkai Guo, Jing Xu, Lei Liu, Sihan Xu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2015
• Subjects: Android application; Androids; association statistics; Classification algorithms; Feature extraction; Malware; permissions; risk evaluation; Risk management; Smart phones
• ISBN: 9781467381253; 146738125X
• DOI: 10.1109/CompComm.2015.7387530

With the development of Android system and intelligent mobile technology, Android applications have been widely used in recent years. However, the trust on these applications may be used by attackers who design malware to achieve illegal income. Therefore, to evaluate the security risk of applications is necessary. In this paper, an evaluation approach based on association statistics of application permissions is represented. By analyzing the frequency of permission combinations, the malicious features of Android applications is estimated. Meanwhile, the redundant frequency is wiped off by computing the frequency discount, and then the risk seeds are correspondingly achieved. The final risk score is calculated by the seeds from malicious and benign aspects. To verify the effectiveness of our approach, we took 1260 malware as "malicious" datasets and 10,247 valid apps collected from Android Market as "benign" datasets, and sufficient experiments show that our approach has better results compared with other traditional methods.