Vulnerability Mining Method Based on Program Feature Graph

In order to better mine network security vulnerabilities, this paper proposes a program feature graph (PFG) that integrates multi-dimensional code information, integrating the program's parse tree, control flow, data flow, and program dependencies into a graph representation, and Modeled as a multi-relationship diagram. Then a vulnerability mining model BRGCN based on an improved graph convolutional neural network is constructed to mine potential vulnerabilities in the target program by learning the program feature graph. Experiments show that the method proposed in this article can effectively improve the accuracy of vulnerability mining.