Global information perception graph neural network system for code vulnerability detection

The invention provides a global information perception graph neural network system for code vulnerability detection. The global information perception graph neural network system comprises a relation code representation module and a global information perception module, the relation code representation module adds edge type information in an information aggregation process of the graph convolutional neural network, enriches node feature representation by using the edge type information, and enhances node features by using an attention mechanism; the global information perception module extracts global features and local features in the code attribute graph by using large kernel convolution and small kernel convolution in the graph convolutional neural network, and learns more abstract and advanced graph representation for code vulnerability classification. The method has the advantages that the defect that a traditional graph neural network is difficult to effectively capture large graph representation can be