Software defect prediction method and system based on network embedding

Abstract

The invention provides a software defect prediction method and a system based on network embedding. The method comprises the following steps: firstly, analyzing the inter-class dependency relationshipof source codes; establishing a directed software inter-class dependency network, performing network embedding on the established dependency network to obtain software structure feature data, and finally inputting the obtained software structure feature data into the software defect prediction model for evaluation to obtain software defect prediction in a software project. According to the method, potential structural features in software are automatically mined by using a network embedding method; therefore, the software defects can be more effectively predicted; the problem that structuralinformation in software is not utilized in an original defect prediction method is solved, the accuracy of a defect prediction model is improved, developers are helped to find possible defects in thesoftware in advance, test tasks are reasonably distributed, the test quantity is reduced, and the efficiency of the developers is improved.

Description

technical field [0001] The invention relates to the field of software testing and defect prediction in software engineering, in particular to a software defect prediction method and system based on network embedding. Background technique [0002] Software defect prediction is based on the past software data, and uses the model established to predict the defect of the new code. By predicting software defects, developers can find possible defects in software early, reduce the amount of testing, and improve testing efficiency. [0003] Today's research on software defect prediction technology can be divided into two stages. The first stage is the feature extraction stage, which makes the representation of defects more efficient by manually designing new features or combining features; the second stage is the classification of machine learning methods. stage, by using new machine learning algorithms to build more accurate models to provide better predictions. [0004] In the f...

AI Technical Summary

Problems solved by technology

Through network embedding, the directed graph nodes formed by the dependencies between software are projected into the low-dimensional vector space to realize the mining of software structural features, thus effectively solving the problem that traditional measurement features cannot capture software structural features

Images