Analyzing and Securing Social Networks

487Attacks on Social Media andData Analytics Solutions

of detecting malware by using data mining (Kolter and Maloof, 2004; Masud et al., 2008a; Schultz

et al., 2001) involves classifying each executable as either benign or malicious. Most past works

have approached the problem as a static data classication problem, where the classication model

is trained with xed training data. However, the escalating rate of malware evolution and innovation

is not well suited to static training. Detection of continuously evolving malware is better treated as a

data stream classication problem. In this paradigm, the data stream is a sequence of executables ...