From Reactive to Proactive: Graph Convolutional Networks for Future Software Module Coupling Prediction

Abstract

Software module dependency prediction is a critical task in modern software engineering for preventing future connectivity issues and improving system sustainability. This study proposes a Graph Convolutional Network (GCN) based framework to predict potential inter-module dependencies using comprehensive software metrics. Experiments were conducted on the complete NASA JM1 dataset (10,885 modules), selected for its scale and extensive use in software engineering research. All 21 software metrics were utilized without dimensionality reduction. A K-Nearest Neighbors (KNN) graph modeling approach (k=8) with a cosine similarity threshold of 0.2 captured structural relationships, producing 85,002 training edges across 15 connected components. The proposed three-layer residual GCN architecture (21→128→128→64) integrates ReLU activation, 30% dropout, and residual skip connections, along with a link-prediction-oriented data partitioning strategy. The model achieved strong performance with 97.58% AUC, 92.12% F1-score, and 99.99% recall. In addition to predictive performance, the framework demonstrated high computational efficiency, requiring an average of 0.165 seconds per training epoch and completing training in 33.1 seconds. These results indicate that the model is suitable for scalable deployment and real-time DevOps integration. By enabling proactive dependency forecasting, the proposed approach supports early identification of design risks and improves software quality management in large-scale development environments.

Keywords

• graph convolutional networks
• software dependency prediction
• machine learning
• software metrics
• NASA JM1 dataset

Reaktiften Proaktife: Gelecekteki Yazılım Modül Bağlantı Tahmini için Graf Evrişimsel Ağları

Abstract

Yazılım modülleri arasındaki bağımlılıkların tahmin edilmesi, gelecekte ortaya çıkabilecek bağlantı sorunlarının önlenmesi ve sistem sürdürülebilirliğinin artırılması açısından modern yazılım mühendisliğinde kritik bir konudur. Bu çalışmada, kapsamlı yazılım metriklerini kullanarak potansiyel modüller arası bağımlılıkları tahmin etmek amacıyla Grafik Evrişimsel Ağ (GCN) tabanlı bir çerçeve önerilmektedir. Deneyler, ölçeği ve yazılım mühendisliği araştırmalarında yaygın kullanımı nedeniyle seçilen NASA JM1 veri kümesinin tamamı (10.885 modül) üzerinde gerçekleştirilmiştir. Tüm 21 yazılım metriği boyut indirgeme uygulanmadan kullanılmıştır. k=8 parametresi ve 0.2 kosinüs benzerlik eşiği ile oluşturulan K-En Yakın Komşu (KNN) tabanlı grafik modelleme yöntemi, yapısal ilişkileri başarıyla yakalayarak 15 bağlantılı bileşen üzerinde 85.002 eğitim kenarı üretmiştir. Önerilen üç katmanlı artık bağlantılı GCN mimarisi (21→128→128→64), ReLU aktivasyonu, %30 dropout ve artık (residual) atlama bağlantıları ile birlikte bağlantı tahmini odaklı bir veri bölme stratejisi içermektedir. Model, %97.58 AUC, %92.12 F1-skoru ve %99.99 recall değerleri ile güçlü bir performans sergilemiştir. Ayrıca model, ortalama 0.165 saniyelik epoch süresi ve toplam 33.1 saniyelik eğitim süresi ile yüksek hesaplama verimliliği göstermiştir. Bu sonuçlar, önerilen yaklaşımın ölçeklenebilir dağıtım ve gerçek zamanlı DevOps entegrasyonu için uygun olduğunu göstermektedir. Potansiyel bağımlılıkların proaktif olarak tahmin edilmesini sağlayan bu yaklaşım, tasarım aşamasındaki risklerin erken tespitine katkı sağlayarak büyük ölçekli yazılım geliştirme ortamlarında kalite yönetimini desteklemektedir.

Keywords

• Graf evrişimsel ağları
• yazılım bağımlılık tahmini
• makine öğrenmesi
• yazılım metrikleri
• NASA JM1 veri seti

References

1. Abbas S, Aftab S, Khan M. A, Ghazal TM, Al Hamadi H, and Yeun CY. Data and Ensemble Machine Learning Fusion Based Intelligent Software Defect Prediction System. Computers, Materials and Continua 2023; 75(3): 6083–6100.
2. Zhou C, He P, Zeng C, and Ma J. Software defect prediction with semantic and structural information of codes based on Graph Neural Networks. Inf Softw Technol 2022; 152: 107057.
3. Aftab S. et al. A Cloud-Based Software Defect Prediction System Using Data and Decision-Level Machine Learning Fusion 2023; 11(3): 632.
4. Ali M et al. Software Defect Prediction Using an Intelligent Ensemble-Based Model. IEEE Access 2025; 12: 20376-20395.
5. Daza A et al. Industrial applications of artificial intelligence in software defects prediction: Systematic review, challenges, and future works. Computers and Electrical Engineering 2025; 124:110411.
6. Aydın Ö, Samlı R. A comparative analysis of machine learning algorithms for software defect prediction using NASA datasets. Journal of Intelligent & Fuzzy Systems 2021; 40(2), 2087–2099.
7. Albattah A, Yavari R, Alsulami M, Algarni A, Bamasag O, Alhakami, H. Software bug prediction using machine learning and input metrics. Sensors 2021; 21(10), 3411.
8. Zhang X, Lin Z, Jiang S, Yu L. A data-driven framework for software defect prediction using feature selection and class imbalance techniques. Applied Sciences 2025; 15(1), 87.

9. Ahmed D, Singh R, Bedi P. Deep learning-based defect prediction with hybrid oversampling: A study on NASA datasets. Soft Computing 2023; 27(14), 1–19.
10. Dey S, Majumdar S, Dutta P. Multi-objective evolutionary fuzzy ensemble for software defect prediction. Information Sciences 2024; 658, 119–140.
11. Siddika A, Rahman M, Chowdhury M. Hybrid ensemble learning approach for software defect prediction. Journal of Systems and Software 2022; 190, 111321.
12. Abdelaziz M, Ramadan R, El-Fakharany M. Temporal convolutional networks with metaheuristic optimization for software defect prediction. Journal of King Saud University – Computer and Information Sciences 2024; 36(4), 823–834.
13. Giray G, Tosun A. Deep learning techniques in software defect prediction: A systematic literature review. ACM Transactions on Software Engineering and Methodology 2022; 31(4), 1–37.
14. Güven A. Makine Öğrenmesi Yöntemleri ile Yazılım Hata Tahmini. İstanbul Üniversitesi-Cerrahpaşa, Lisansüstü Eğitim Enstitüsü 2021.
15. Zeng C, Zhou CY, Lv SK, He P, and Huang J. GCN2defect: Graph Convolutional Networks for SMOTETomek-based Software Defect Prediction. in Proc. ISSRE, 2021, 69–79.
16. Gou X, Zhang A, Wang C, Liu Y, Zhao X and Yang S. Software Fault Localization Based on Network Spectrum and Graph Neural Network. IEEE Transactions on Reliability 2024; 73(4): 1819-1833.
17. Shen H, Ju X, Chen X and Yang G. EDP-BGCNN: Effective Defect Prediction via BERT-based Graph Convolutional Neural Network. 2023 IEEE 47th Annual Computers, Software, and Applications Conference (COMPSAC), Torino, Italy, 2023, 850-859.
18. Umesh N, Manjula D. HAG-SDP: A hierarchical attention-based graph neural network for software defect prediction. Expert Systems with Applications 2025; 244, 122946.
19. Sikic L, Kurdija AS, Vladimir K, and Silic M. Graph Neural Network for Source Code Defect Prediction. IEEE Access 2022; 10: 10402–10415
20. Cui M, Long S, Jiang Y, and Na X. Research of Software Defect Prediction Model Based on Complex Network and Graph Neural Network. Entropy 2022; 24(10):1373.
21. Bass L and Kazman R. Software Architecture in Practice. 2003.
22. Kruchten P, Nord RL, Ozkaya I. Technical Debt: From Metaphor to Theory and Practice. IEEE Softw 2012; 29(6): 18–21.
23. Shepperd S, Song Q, Sun Z, and Mair C. Data Quality: Some Comments on the NASA Software Defect Datasets. IEEE Trans Softw Eng 2013; 39(9): 1208–1215.
24. Wang Y, Zhan J, Chen L, Yu H. A comparative study of Graph Convolutional Networks and Graph Attention Networks for software module representation learning. Journal of Systems and Software 2021; 176, 110957.
25. García S, Luengo J, and Herrera F. Data Preprocessing in Data Mining. 2015; 72.
26. Batista GEAPA and Monard MC. An analysis of four missing data treatment methods for supervised learning. Appl Artif Intell 2003; 17(5–6): 519–533.
27. Ioffe S and Szegedy C. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. 2015.
28. Śliwerski J, Zimmermann T, and Zeller A. When Do Changes Induce Fixes?. 2005.
29. Zimmermann T, Weißgerber P, Diehl S, and Zeller A. Mining Version Histories to Guide Software Changes. 2004.
30. He K, Zhang X, Ren S, and Sun J. Deep Residual Learning for Image Recognition. in Proc. CVPR, 2016, pp. 770–778.
31. Bishop CM. Pattern Recognition and Machine Learning. Springer, 2006.