Login | Register

Extending the Reach of Fault Localization to Assist in Automated Debugging


Extending the Reach of Fault Localization to Assist in Automated Debugging

Chen, An Ran (2023) Extending the Reach of Fault Localization to Assist in Automated Debugging. PhD thesis, Concordia University.

[thumbnail of Chen_PhD_F2023.pdf]
Text (application/pdf)
Chen_PhD_F2023.pdf - Accepted Version
Available under License Spectrum Terms of Access.


Software debugging is one of the most time-consuming tasks in modern software maintenance. To assist developers with debugging, researchers have proposed fault localization techniques. These techniques aim to automate the process of locating faults in software, which can greatly reduce debugging time and assist developers in understanding the faults. Effective fault localization is also crucial for automated program repair techniques, as it helps identify potential faulty locations for patching.

Despite recent efforts to advance fault localization techniques, their effectiveness is still limited. With the increasing complexity of modern software, fault localization may not always provide direct identification of the root causes of faults. Further, there is a lack of studies on their application in modern software development. Most prior studies have evaluated these techniques in traditional software development settings, where only a single snapshot of the system is considered. However, modern software development often involves continuous and fine-grained changes to the system. This dissertation proposes a series of approaches to explore new automated debugging solutions that can enhance software quality assurance and reliability practices, with a specific focus on extending the reach of fault localization in modern software development.

The dissertation begins with an empirical study on user-reported logs in bug reports, revealing that re-constructed execution paths from these logs provide valuable debugging hints. To further assist developers in debugging, we propose using static analysis techniques for information-retrieval and path-guided fault localization. By leveraging execution paths from logs in bug reports, we can improve the effectiveness of fault localization techniques. Second, we investigate the characteristics of operational data in continuous integration that can help capture faults early in the testing phase. As there is currently no available continuous integration benchmark that incorporates continuous test execution and failure, we present T-Evos, a dataset that comprises various operational data in continuous integration settings. We propose automated fault localization techniques that integrate change information from continuous integration settings, and demonstrate that leveraging such fine-grained change information can significantly improve their effectiveness. Finally, the dissertation investigates the data cleanness in fault localization by examining developers' knowledge in fault-triggering tests. The study reveals a significant degradation in the performance of fault localization techniques when evaluated on faults without developer knowledge.

Through case studies and experiments, the proposed techniques in this dissertation significantly improve the effectiveness of fault localization and facilitate their adoption in modern software development. Additionally, this dissertation provides valuable insights into new debugging solutions for future research.

Divisions:Concordia University > Gina Cody School of Engineering and Computer Science > Computer Science and Software Engineering
Item Type:Thesis (PhD)
Authors:Chen, An Ran
Institution:Concordia University
Degree Name:Ph. D.
Program:Software Engineering
Date:23 June 2023
Thesis Supervisor(s):Chen, Tse-Hsun
ID Code:992561
Deposited By: AN RAN CHEN
Deposited On:17 Nov 2023 14:58
Last Modified:17 Nov 2023 14:58
All items in Spectrum are protected by copyright, with all rights reserved. The use of items is governed by Spectrum's terms of access.

Repository Staff Only: item control page

Downloads per month over past year

Research related to the current document (at the CORE website)
- Research related to the current document (at the CORE website)
Back to top Back to top