Vulnerability identification method based on semantic learning of slice code dependency graph

The invention relates to the technical field of vulnerability detection, and particularly discloses a vulnerability identification method based on semantic learning of a slice code dependency graph, which comprises the following steps of: S1, inputting an executable binary program, and obtaining a pseudo code of each function in the binary program by utilizing a decompilation technology; s2, performing data dependency analysis and control dependency analysis on the dangerous library/API function calling execution process and the dangerous library/API function calling execution process in each pseudo code function, extracting fine-grained pseudo code snippets only related to dangerous library/API function calling, and constructing a slice code dependency graph according to the fine-grained pseudo code snippets; and the like. The method is suitable for vulnerability identification in a binary code scene, security vulnerability detection can be carried out on software under the limited condition that source code