A Software Defect Prediction Method Based on Convolutional Neural Network

Abstract

A method for predicting software defects based on a convolutional neural network disclosed by the present invention comprises the following steps: analyzing the source codes of each file in a software project to form a set of AST Token vectors; establishing a mapping between integers and Tokens, and converting the AST Token vectors to Convert it into a numerical vector; use SMOTE technology to deal with the problem of unbalanced classification of the numerical vector set data; build a convolutional neural network on the basis of the numerical vector set, and extract the feature vector that can express the semantics of the code; use the convolutional neural network to learn The feature is merged with the traditional manual static feature; the data set with the merged feature is input into the support vector machine classifier, and the software defect prediction model is trained. The invention can be directly applied to the defect prediction task of actual software, can capture the semantic features of the source code, solve the problem of lack of semantic feature analysis in the traditional method, and further improve the accuracy of the defect prediction model.

Description

technical field [0001] The invention relates to the field of software analysis and defect prediction in software engineering, in particular to a software defect prediction method based on a convolutional neural network. Background technique [0002] Potential and unknown defects in software will seriously affect the quality of software, so software analysis and defect prediction techniques play an important role in software quality assurance tasks. If software defects can be found early, it will help the software team understand the quality status of the current project, and then allocate testing resources reasonably. However, it is impractical to manually review all code units in a project. Therefore, more and more software engineering researchers and practitioners have begun to pay close attention to software defect prediction technology based on machine learning, and try to use a variety of machine learning methods to Detect potentially defective modules and files in sof...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and provide a software defect prediction method based on convolutional neural network, which combines the technology of neural network automatic feature generation to solve the problem of traditional manual features in existing software defect prediction methods. The problem of missing semantic features

Images