Source Code Vulnerability Detection Method Based on Refactored Code Attribute Map

With the development of information technology, security incidents induced by code vulnerabilities occur frequently, and static code vulnerability detection methods based on deep learning suffer from feature redundancy and limited ability to detect complex vulnerabilities. In this paper, we propose a reconstructed code attribute graph-based vulnerability detection method for (Restructured Code Property Graph, RCPG).RCPG is constructed by a cross-function code property graph construction algorithm and code slicing technique, which enhances the node feature extraction and improves the model’s ability to capture code changes. Experimental results demonstrate that the method significantly outperforms the state-of-the-art approaches in terms of accuracy rate and F1 value.