Predicting Object Communication Errors in Constructor Development
An important challenge in dynamic software development is to predict object formation run-time object communication errors in complex environments involving multiple and multi-level object inheritance. This paper proposes a technique for doing so. The technique is intended to stop or fix software bu...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11083620/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | An important challenge in dynamic software development is to predict object formation run-time object communication errors in complex environments involving multiple and multi-level object inheritance. This paper proposes a technique for doing so. The technique is intended to stop or fix software bugs, particularly in situations where it is believed that object communications would persist across different settings. The research addresses a critical gap in existing methodologies by integrating static and dynamic object-oriented metrics, providing a holistic approach to defect prediction. Additionally, the paper presents a software testing defect prediction model that categorizes problematic classes according to inheritance defects found in a particular class. Earlier researchers studied various methods for predicting and mitigating software defects in object-oriented programming. We propose a defect prediction model that categorizes problematic classes based on inheritance defects to overcome the gaps in existing methodologies by introducing them. We evaluated this object communication error prediction using a set of 150 common errors drawn primarily from real-world open-source repositories and enriched with synthesized cases reflecting rare but critical inheritance-related bugs, ensuring comprehensive and realistic error representation. The methodology employs classification techniques, including K-Nearest Neighbors (KNN) for k-fold cross-validation, Random Forest, Decision Trees, and Support Vector Machines (SVM), alongside object-oriented metrics such as inheritance, cohesion, and coupling. Key performance metrics precision (78%), F1 score (76.4%), recall (74.9%), and ROC AUC (89%), demonstrate the model’s superiority over prior approaches. These results underscore the practical applicability of the model in improving defect detection accuracy and reducing software failures. |
|---|---|
| ISSN: | 2169-3536 |