Methodology and Tooling for Automated Classification and Ranking of Software Bug Reports
Information Theories and Applications · 30.05.2026 · 33 (2) · 143-178
DOI: 10.54521/ijita33-02-p02 ↗Kurz gefasst
Fachartikel in der Zeitschrift Information Theories and Applications, online erschienen am 30.05.2026. Verfasst von Olena Chebanyuk und 1 weiteren Autor:innen.
Abstract (Originalsprache)
This paper presents a practical methodology for the automated classification and ranking of bug reports, designed to optimize defect management processes in environments with large volumes of unstructured data. The methodology integrates natural language preprocessing, pre-trained word embeddings (GloVe), and an ensemble machine learning architecture that combines a Bidirectional Long Short-Term Memory (Bi-LSTM) network for deep semantic analysis with Logistic Regression to ensure result interpretability. The proposed approach addresses the challenges of class imbalance and subjectivity in defect descriptions by ensuring decision-making transparency through a keyword impact explanation mechanism. The method achieves high performance in detecting critical errors (Recall for priority P1 ≈ 0.89) and effectively filters out improvement requests.
Zitieren
APA
Chebanyuk, O., & Dmytrenok, B. (2026). Methodology and Tooling for Automated Classification and Ranking of Software Bug Reports. Information Theories and Applications, 33(2), 143-178. https://doi.org/10.54521/ijita33-02-p02
Vancouver
Chebanyuk O, Dmytrenok B. Methodology and Tooling for Automated Classification and Ranking of Software Bug Reports. Information Theories and Applications. 2026;33(2):143-178. doi:10.54521/ijita33-02-p02
BibTeX
@article{Chebanyuk2026,
author = {Chebanyuk, Olena and Dmytrenok, Bohdan},
title = {Methodology and Tooling for Automated Classification and Ranking of Software Bug Reports},
journal = {Information Theories and Applications},
year = {2026},
volume = {33},
number = {2},
pages = {143--178},
doi = {10.54521/ijita33-02-p02}
}
EUVIMED