Android authorship attribution through string analysis

摘要: With the rising popularity of Android mobile devices, amount malicious applications targeting platform has been increasing tremendously. To mitigate risk apps, there is a need for an automated system to detect these applications. Current detection techniques rely on signatures well-documented malware, and hence may not be able new malware samples. Instead generating samples themselves, in this work, we propose develop lightweight that can generate writers by leveraging string components present their binaries. Using author signatures, effectively wide range existing, as well any new, generated particular authors. The proposed achieved 98%, 96%, 71% accuracy over datasets 1559 benign, 262 malicious, 96 obfuscated applications, respectively. string-based approach compared only 50% obtained with existing Ding Samadzadeh's system.