Back to search

Schulich School of Engineering

Advancing Bug Report Management: Graph-Based Modeling, Large Language Models, and Quality Improvement

Abstract

dc:description.abstract

Software maintenance is a critical, cost-intensive phase of software development, heavily dependent on effective bug report management. Bug reports are vital artifacts in this process, enabling developers to identify, prioritize, and resolve software defects. However, managing these reports effectively poses two major challenges: (1) accurate prediction of bug severity and priority to optimize resource allocation and ensure timely resolution of critical defects, and (2) generating well-structured, high-quality bug reports. This thesis addresses both challenges through a different approach involving graph-based machine learning techniques and Instruction fine-tuning of the large language models. Firstly, we present a graph-based modeling technique for bug priority and severity prediction. By representing bug reports as nodes in a graph and using similarity be tween them as edges. Our Graph Neural Network (GNN) architecture consists of Graph Sample and AggreGatE (GraphSAGE) and Graph Attention Networks, which outperform fine-tuned Bidirectional Encoder Representations from Transformers (BERT) on predict ing severity and priority of bug reports. Secondly, we focus on improving the quality of bug reports through instruction fine-tuning. We fine-tuned open-source Large Language Models (LLMs) on structured bug reports, transforming informal bug descriptions into clear, structured reports adhering to standard templates. Our fine-tuned models achieved comparable performance to industry-grade LLMs, demonstrating practical effectiveness. To validate our approach, we developed a web-based tool called Bugwise, and conducted a human evaluation study involving both general users and experienced developers. AI generated reports were rated higher in clarity and completeness by experienced develop ers. To support future research, we present BugsRepo, a dataset of over 119,585 Mozilla bug reports, information on 19,351 contributors, and 12,614 well-structured bug reports, providing a valuable resource for the software engineering community. Overall, our re search demonstrates that graph-based models and instruction fine-tuning offer effective AI-driven solutions to key challenges in bug report management, ultimately enhancing the maintainability and reliability of software systems.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MSc)
Discipline thesis:degree_discipline
Engineering – Electrical & Computer
Grantor dc:publisher.institution
Schulich School of Engineering
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Acharya, Jagrit
Advisor dc:contributor.advisor
  • Gouri, Ginde
Committee members dc:contributor.committeemember
  • Tan, Benjamin
  • Novarun, Deb

Subjects

dc:subject × 4

Rights

dc:rights
Statement dc:rights
  • University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Citation
Acharya, J. (2025). Advancing bug report management: graph-based modeling, large language models, and quality improvement (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.

Chain of custody

source
Harvested from
University of Calgary
Base URL
ucalgary.scholaris.ca/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Acharya, Jagrit. Advancing Bug Report Management: Graph-Based Modeling, Large Language Models, and Quality Improvement. Schulich School of Engineering, 2025. https://hdl.handle.net/1880/122011