CatraDroid: A Call Trace Driven Detection of Malicious Behaiviors in Android Applications.

摘要: The explosive growth of Android malware has led to a strong interest in developing efficient and precise detection approach. Recent efforts have shown that machine learning-based classification is promising direction, the API-level features are extremely representative discriminate been drastically used different forms. In this work, we implement light-weight system, CatraDroid, recovers semantics at call graph level classify applications. CatraDroid leverages text mining technique capture list sensitive APIs from knowledge consisting exploits databases, code samples, configurations codebases. It builds complete for applications identifies traces entry methods API calls. Using as features, our approach can effectively benign Through evaluation, demonstrated outperforms state-of-art approach, with high-quality extracted by static analysis.