Code reviewer recommendation based on cross-project and technology experience

点击翻译按钮获取中文摘要

Code review is a manual examination of source code to find out simple errors, coding rules violations, or logical mistakes in the newly developed code. Peer code review is an informal but highly effective code review practice adopted by the software industries in recent years. VendAsta is one of the largest software companies based in Saskatchewan with 150+ employees and 50+ software products and services. The company uses GitHub, an online project hosting service, to store and manage all of its source code, and emphasizes peer code review for ensuring high quality in the products and services. One of the most challenging tasks involved with the review process is the identification of an expert reviewer. The knowledge as to who is an appropriate reviewer for any given code is tribal, which means it is not very consistent, and these identification tasks are almost impossible for new employees. Thus, developers at VendAsta often spend a significant amount of time in looking for appropriate code reviewers. Our research will address this non-trivial problem, and we will develop a tool that can automatically recommend a list of appropriate code reviewers for any given code review task exploiting both the cross-project experience in the usage of libraries and specialized technologies. The tool would automate the code reviewer selection process, and thus, would increase developers' productivity as well as software product quality in the long run, providing direct economic benefits to VendAsta and Canada overall.