CloneRipples: predicting change propagation between code clone instances by graph-based deep learning

Code clones are recognized as a code smell that may require additional effort for simultaneous changes of multiple clone instances during software maintenance. To alleviate quality threats caused by inconsistent changes in clone instances, it is essential to accurately and efficiently make the decis...

Full description

Saved in:
Bibliographic Details
Published in:Empirical software engineering : an international journal 2025-02, Vol.30 (1), p.14, Article 14
Main Authors: Wu, Yijian, Chen, Yuan, Peng, Xin, Hu, Bin, Wang, Xiaochen, Fu, Baiqiang, Zhao, Wenyun
Format: Article
Language:English
Subjects:
Artificial neural networks
Compilers
Computer Science
Datasets
Decisions
Deep learning
Interpreters
Predictive maintenance
Programming Languages
Propagation
Software Engineering/Programming and Operating Systems
Source code
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Code clones are recognized as a code smell that may require additional effort for simultaneous changes of multiple clone instances during software maintenance. To alleviate quality threats caused by inconsistent changes in clone instances, it is essential to accurately and efficiently make the decisions of change propagation between code clone instances. Our exploratory study has revealed that a clone class can have both propagation-required changes and propagation-free changes and thus fine-grained change propagation decisions are required. Based on the findings, we propose a graph-based deep learning approach to predict the change propagation requirements of clone instances. We design a deep learning model that employs a Relational Graph Convolutional Network (R-GCN) to predict the clone change propagation requirement. In order to evaluate our approach, we construct a dataset that includes 24,672 pairs of matched changes and 38,041 non-matched changes based on 51 open-source Java projects. Experiment results show that the approach achieves high precision (83.1%), recall (81.2%), and F1-score (82.1%). We implemented an IntelliJ IDEA tool called CloneRipples to assist developers to decide the necessity of change propagation between code clone instances seamlessly in development environment. Manual inspection identified the chances for purifying the dataset by rectifying the data labels of non-matched changes. Extended experiments for various data purification strategies reveal feasible ways to improve the prediction effectiveness and generality.
ISSN:1382-3256
1573-7616
DOI:10.1007/s10664-024-10567-0