Code retrieval method and system based on graph neural network and computer readable storage medium

Abstract

The invention discloses a code retrieval method and system based on a graph neural network and a computer readable storage medium. The method comprises the steps: S1, obtaining code data, and carrying out code integrity preprocessing; s2, sequence information, control flow graph information and program dependency graph information of the preprocessed codes are extracted respectively; s3, constructing a code graph of the code sequence by using the sequence information of the code, the control flow graph information and the program dependency graph information; s4, constructing a code retrieval model based on the graph neural network, and training the code retrieval model based on the graph neural network by utilizing the code sequence-based code graph and natural language description; and S5, performing code retrieval by using the trained code retrieval model based on the graph neural network. According to the method, code semantic and structural feature extraction is more complete. Meanwhile, a single graph neural network structure is adopted, the training time of the model is shortened, parameter adjustment is reduced, and the retrieval accuracy is improved.

Description

technical field [0001] The present invention relates to the technical field of neural networks, and more specifically, to a code retrieval method, system and computer-readable storage medium based on a graph neural network. Background technique [0002] The traditional method of code retrieval through natural language usually uses the sequence information of the code or the graph structure information of the code alone, and does not fully obtain the semantic and structural information of the code. In other methods for code retrieval through natural language, most of them use sequential neural network and tree neural network, while ignoring the characteristics of code graph structure. [0003] In the prior art, the Chinese invention patent with publication number CN107015905A disclosed a method and device for querying source code on August 4, 2017. The method for querying source code is pre-established with functional units in the software model The corresponding relationshi...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a code retrieval method and system based on a graph neural network in order to overcome the defects of the existing code retrieval through natural language without considering the semantic and structural information of the complete code, incomplete feature extraction, and poor accuracy. and computer readable storage media

Images