Fine-grained mixed semantic vulnerability detection method and system

The invention discloses a fine-grained mixed semantic vulnerability detection method and system, and belongs to the technical field of network security. Comprising the steps that sequence code representation is input into a pre-training language model for processing, and a global semantic feature vector and an attention score embedding matrix are obtained; inputting the sequence code representation into a preset multi-scale fusion convolutional neural network for processing to obtain a local feature vector; inputting the graph code representation into a graph convolutional neural network with a residual structure for processing to obtain a graph embedding vector; the global semantic feature vector, the local feature vector and the graph embedding vector are fused and then input into a trained vulnerability detection model for processing, and a vulnerability detection result is obtained; and performing fine-grained detection on the sequence code representation according to a vulnerability detection result and