Effective assignment and assistance to software developers and reviewers
Zanjani, Motahareh Bahrami
MetadataShow full item record
The conducted research is within the realm of software maintenance and evolution. Human reliance and dominance are ubiquitous in sustaining a high-quality large software system. Automatically assigning the right solution providers to the maintenance task at hand is arguably as important as providing the right tool support for it, especially in the far too commonly found state of inadequate or obsolete documentation of large-scale software systems. Several maintenance tasks related to assignment and assistance to software developers and reviewers are addressed, and multiple solutions are presented. The key insight behind the presented solutions is the analysis and use of micro-levels of human-to-code and human-to-human interactions. The formulated methodology consists of the restrained use of machine learning techniques, lightweight source code analysis, and mathematical quantification of different markers of developer and reviewer expertise from these micro interactions. In this dissertation, we first present the automated solutions for Software Change Impact Analysis based on interaction and code review activities of developers. Then we provide explanation for two separate developer expertise models which use the micro-levels of human-to-code and human-to-human interactions from the previous code review and interaction activities of developers. Next, we present a reviewer expertise model based on code review activities of developers and show how this expertise model can be used for Code Reviewer Recommendation. At the end we examine the influential features that characterize the acceptance probability of a submitted patch (implemented code change) by developers. We present a predictive model that classifies whether a patch will be accepted or not as soon as it is submitted for code review in order to assist developers and reviewers in prioritizing and focusing their efforts. A rigorous empirical validation on large open source and commercial systems shows that the solutions based on the presented methodology outperform several existing solutions. The quantitative gains of our solutions across a spectrum of evaluation metrics along with their statistical significance are reported.
Thesis (Ph.D.)-- Wichita State University, College of Engineering, Dept. of Electrical Engineering and Computer Science