Advancing Novel Approaches to Improve the Practices and Processes of Software Development in Online Collaborative Coding Platforms

点击翻译按钮获取中文摘要

The vast majority (as much as 99%) of codebases used for software development contain open source code [BNPMedia, 2020]. Key benefits of Open Source Software (OSS) are the ability to access, modify, and use the source code; meaning that in theory any capable user is able to participate in development of source code. This has increased the overall innovation and creativity of software and plays a critical role in the business models for high-tech firms and other industries, but is not without its drawbacks. As OSS is accessible to all means, it can be used and misused. OSS practitioners can expect security risks, bug proneness, and lack of user support and expertise. Notably, my collaborators and I have recently shown that previous unconsidered bug origins have a significant, negative impact on bug prediction and localization and that the evaluation of OSS contributions might not be "free" of sytemic bias. My long-term objective is to develop techniques to ensure that high-quality, effective, efficient, and unbiased OSS systems are developed for diverse users without wasting resources or increasing costs. Within the scope of this Discovery Grant (DG) research program, progress towards this goal will be reached through three sub-objectives (SO): (SO1) Studying novel techniques to predict and locate intrinsic and extrinsic bugs during OSS development; (SO2) Improving OSS development practices by understanding and characterizing extrinsic bugs; (SO3) Devising an OSS framework for developing unbiased and fair OSS systems. These three sub-objectives will be led by 3 PhD, 3 MSc and 1 Undergrad HQP under my supervision. I will actively recruit HQP from historically marginalized groups and consider HQP applications from a holistic perspective. The sub-objectives will be accomplished using mixed-methods research approaches. At a high level, I aim to employ quantitative statistical analyses and a variety of deep learning-based, machine learning-based, and natural language processing techniques on OSS development observational data. I will use mining software repository approaches to gather the OSS development data from OSS repositories hosted in online collaborative coding platforms. I will also use qualitative research methods, e.g., interviews and surveys, to gauge the effectiveness of the techniques and tools developed during the program. In addition, I will use longitudinal field studies to understand how effective the proposed tools/techniques are to help software engineers improve software quality and adjust the behavior of the tools accordingly. The outcome of this DG is particularly beneficial to a multi-cultural society like Canada that is increasingly looking to play a considerable role in technology development as canadians can function in the forefront of the OSS development community. The novelty for the National Science and Engineering Community lies in the techniques and tools developed that will help improve quality of OSS systems.