Collaborative Research: SHF: Medium: Bug Report Management 2.0

University Of Texas At Dallas, Richardson TX

Investigators

Abstract

Software systems often suffer from defects that lead to unexpected results. End users report these unexpected results via issue-reporting systems so that software engineers can identify and fix the related defects to improve the quality of the system. When reporting, users can describe the software problems using natural language or graphical information such as screenshots and videos. Unfortunately, everyday end users are rarely, if ever, trained in reporting software issues. In consequence, they often submit reports that are incomplete or hard to understand, resulting in excessive effort spent addressing the problems, or even the inability for the underlying defects to be identified and fixed. In addition, existing issue-reporting systems are unable to enforce quality standards for reports and fail to provide feedback to the reporters when they submit substandard information. This project will develop a novel-issue reporting system that will allow users to describe software problems interactively, through a dialogue with an automated software agent, rather than writing reports passively, with no feedback and quality assessment. The software agent will automatically convert the conversations into high-quality issue reports, which will be transmitted to the software engineers. The proposed system will allow software engineers to manage and fix defects faster, leading to higher-quality software systems. The project will also produce and disseminate educational material on best practices in reporting software problems. These materials are intended to be integrated into existing computer-literacy courses at all levels of education. In addition, the project will focus on recruiting and retaining computer science students from traditionally underrepresented categories. The project is centered on three specific goals. First, it will develop novel techniques for the automated analysis and quality assessment of defect reports. This component will adapt and build upon techniques for automated discourse analysis, dynamic program analysis, and computer vision. Second, it will improve the quality of issue reports through interactive mechanisms. This proactive reporting solution will be developed through cross-cutting research on empirical software engineering, human-computer interaction, automated text analysis, and advanced machine learning. This new dialogue-based reporting is expected to become the standard method by which many kinds of software issues will be reported. Finally, the project will develop more efficient and effective techniques for automated defect reproduction and duplicate detection, leveraging the high-quality reports created via the interactive reporting system. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.