Software vulnerability intelligent detection method and system based on code attribute graph and medium

Abstract

The invention discloses a software vulnerability intelligent detection method and system based on a code attribute graph and a medium. The method comprises the steps: generating the code attribute graph for a detected code sample, and extracting an abstract syntax tree sequence, a control flow graph sequence and a program dependence graph sequence; converting the obtained various sequences into numerical type vectors through a pre-trained vector mapper, extracting feature vectors through corresponding pre-trained feature extractors, and combining the extracted three feature vectors into feature vectors of a code sample; utilizing a pre-trained vulnerability detector to obtain the probability that the code sample is a vulnerability code according to the feature direction of the code sample, and if the probability exceeds a preset threshold value, judging that the code sample is the vulnerability code. According to the processing mode, more grammars, semantics and other information of the source codes can be reserved, the vulnerability detection effect of the cross-software source code mixing-oriented real data set can be improved, and the false alarm rate and the missing report rate are reduced.

Description

technical field [0001] The invention relates to the technical fields of computer program detection and software vulnerability detection and detection, in particular to a software vulnerability intelligent detection method, system and medium based on a code attribute graph. Background technique [0002] Software vulnerability refers to the flaws existing in software during its life cycle (that is, the whole process of development, deployment, and execution). It causes security issues such as leakage of user privacy data, theft of digital assets, and alteration of system information, resulting in huge economic losses and reputational losses for enterprises and individuals. With the increasing complexity and scale of software systems, the frequency of vulnerable codes is also increasing. According to data released by the National Vulnerability Database (NVD), the number of vulnerable codes released in 2017 was 14,645, the number of vulnerable codes released in 2018 was 16,511,...

AI Technical Summary

Problems solved by technology

Dynamic testing methods are usually time-consuming and lack completeness, prone to false negatives

The hybrid method combines static and dynamic analysis techniques to overcome some shortcomings, but the hybrid method is difficult to fully realize automatic detection, and the detection effect on unknown vulnerabilities is relatively weak

[0004] At present, the scale of software is becoming larger and more complex, and the forms of vulnerabilities are becoming more diverse. Traditional software vulnerability detection methods have difficulties in dealing with complex and diverse vulnerabilities.

Images