SQL statement auditing method based on graph attention neural network

The invention discloses an SQL statement auditing method based on a graph attention neural network, and the method specifically comprises the following steps: receiving a structured query language SQL statement auditing request, and obtaining a to-be-audited SQL statement; converting the SQL statement into a sequence which can be processed by a GAT (Graph Attention Neural Network); coding the sequence, and abstracting each node of the sequence into a vector representation; based on learning of a graph attention neural network GAT, generating a syntax tree structure with a structured expression; and formulating an SQL auditing rule, performing comprehensive traversal according to the SQL auditing rule syntax tree structure, performing rule matching on each node, identifying potential problems, and completing SQL auditing. By generating the syntax tree structure, the grammar and semantic structures of the SQL statement are deeply understood, and the accuracy of checking the SQL statement is improved; the method