Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201064,
        author = {sharan and CH. Chandu and K. Vikas and M. Satish and S. Pavithraa},
        title = {BUG PRIORITY PREDICTION USING DEEP ENSEMBLE},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {3006-3015},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201064},
        abstract = {Bug priority prediction is a critical component in contemporary software engineering, as it directly influences issue resolution timelines and resource management efficiency. In large-scale development environments, assigning priorities to reported bugs manually becomes increasingly challenging due to the growing volume of reports and the subjective nature of human judgment. This often results in inconsistencies, delays, and potential misclassification of critical issues. Conventional methods based on statistical analysis and traditional machine learning algorithms typically depend on manually engineered features, which limits their ability to capture the deeper semantic meaning embedded within textual bug descriptions.
To overcome these challenges, this study introduces a deep ensemble-based framework for automated bug priority prediction. The proposed approach integrates multiple deep learning architectures to enhance predictive capability and robustness. Initially, advanced text preprocessing and feature representation techniques are applied to convert unstructured bug reports into structured numerical forms suitable for model training. The framework employs base learners such as Convolutional Neural Networks and Long Short-Term Memory networks to extract both contextual and sequential information from the data. While convolutional models capture localized patterns, recurrent models effectively learn long-term dependencies within textual sequences.
To further improve performance, the outputs of these individual models are combined using ensemble strategies such as stacking or voting. This integration enhances generalization ability and reduces the limitations of relying on a single model. The effectiveness of the proposed system is validated using real-world bug tracking datasets, where it demonstrates notable improvements in performance metrics, including accuracy, precision, recall, and F1-score, when compared to standalone models and conventional approaches.
In addition, the framework addresses challenges such as class imbalance, ensuring consistent prediction performance across different priority categories. The experimental results confirm that deep ensemble learning offers a scalable and reliable solution for automating bug triaging processes. By enabling accurate and efficient prioritization of issues, the proposed method contributes to improved software quality and accelerated development cycles. Future extensions of this work may involve incorporating transformer-based architectures, enabling real-time deployment, and integrating the system with continuous integration and DevOps pipelines to further enhance automation in software maintenance workflows.},
        keywords = {Bug Priority Prediction, Deep Learning, Ensemble Learning, Software Engineering, Classification, Bug Reports.},
        month = {May},
        }