Abstract:

Because information system attack events caused by software misconducts or malicious damages occur frequently-software security requirements of information system are higher and higher. How to achieve the security analysis of software behaviors effectively has become a popular topic. This paper focuses on the software behaviors analysis technology based on big data samples. Considering the characteristics of massive, multi-dimensional, high-speed change and complex internal relations of big data samples, combining the static analysis based on hierarchical clustering algorithm with dynamic behaviors analysis based on SVM algorithm, this paper constructs a software behaviors analysis model based on machine learning algorithm. The model uses cloud centralized processing method, which can effectively save the resource consumption of the terminal, and realize the efficient and rapid detection of malicious software.

Key words: big data samples, software behavior, machine learning algorithm