Security importance assessment for system objects and malware detection

摘要: System objects play different roles in computer systems and exhibit levels of importance to system security. Assessing the helps us develop effective security protection methods. However, little work has focused on understanding assessing from a perspective. In this paper, we build dependency network access behaviors quantify system-wide Similar other networked systems, observe small-world effect power-law distributions for in- out-degree network. Exploring rich structures provides insights into We assess objects, with respect security, by centrality metrics propose an based model malware detection. evaluate various perspectives demonstrate their feasibility practicality. Furthermore, extensive experimental results real-world dataset that our is capable detecting 7257 samples 27,840 benign processes 93.92% true positive rate at 0.1% false rate.