Software defect prediction method based on heterogeneous graph neural network

Abstract

The invention discloses a software defect prediction method based on a heterogeneous graph neural network, and belongs to the field of software defect testing. The method comprises the steps that firstly, existing software warehouses and software defect log data are converted into corresponding code graphs and defect chains through corresponding analysis; according to the content of the defect description information, performing cross-domain association on nodes in the code graph and nodes in the defect graph through a prefix tree with a state machine; after representation vectors are generated for the code nodes and the defect nodes respectively, the representation vectors are sent into a heterogeneous graph neural network for multi-level attention aggregation, information transmission of content and semantic paths is obtained, and the code nodes and the defect nodes which are connected are obtained; decoding the connection nodes by using a knowledge graph representation learning method, and normalizing decoding scores to obtain whether the code nodes have defects or not; according to the method, software defect prediction is carried out in a new mode, and the accuracy of a software defect prediction tool is improved.

Description

technical field [0001] The invention belongs to the field of software defect testing, in particular to a software defect prediction method based on a heterogeneous graph neural network. Background technique [0002] In the process of software development, software defects will inevitably be introduced, and software defects have an important impact on software quality. [0003] With the rise of mobile Internet, cloud computing, blockchain and artificial intelligence, the scope of software quality assurance has also expanded from the traditional software industry to the emerging software field, and higher requirements have been put forward for software quality assurance. Diversified software applications have covered all walks of life. With the continuous iteration and updating of software technologies, the complexity of the software is also increasing. At this time, the reliability and stability of the software system is an issue that cannot be ignored. It is closely related...

AI Technical Summary

Problems solved by technology

In practical applications, software defect prediction technology driven by machine learning faces the problems of cold start and lack of labeled data, which limits the practical application value of software defect prediction technology based on artificial design features or machine learning methods.

At the same time, cross-project software defect prediction is a relatively difficult point: there are large differences between different projects, and software development backgrounds such as development requirements and application scenarios, actual development content, and development management processes cannot be directly migrated

Images