Code similarity analysis technology based on graph neural networks

This paper focuses on the code similarity analysis technology based on graph neural networks. By representing code as graph structures such as program dependency graphs and control flow graphs, graph neural networks are utilized to learn the feature representations of these graphs. In-depth research is conducted on how to design graph neural network architectures that are suitable for code graph structures and how to integrate different types of graph information, aiming to capture the semantic and structural similarities of code more accurately and provide a more effective approach for code similarity analysis.