Effectively managing and leveraging crowdsourced knowledge for software engineering

点击翻译按钮获取中文摘要

Knowledge-sharing platforms (e.g., Stack Overflow and GitHub) have changed how developers share and seek knowledge on the Internet. These platforms accumulate a massive amount of knowledge for Software Engineering (SE) that is contributed by a large and rapidly-changing group of participants (i.e., crowdsourced SE knowledge). For instance, Stack Overflow has accumulated a tremendous amount of SE knowledge (often including code examples, along with their textual description). A recent survey shows that 61% of developers seek help on Stack Overflow daily, which underlines the perceived value of such crowdsourced SE knowledge. As the amount of crowdsourced SE knowledge rapidly grows, managing and leveraging such a massive amount of knowledge is challenging. Challenge 1: The organization of knowledge is becoming challenging due to a continual inundation of generated information on knowledge-sharing platforms. Challenge 2: Maintaining the quality of such an enormous amount of knowledge requires an enormous amount of effort from the community. In fact, recent studies show that code examples on Stack Overflow could be toxic from various perspectives (e.g., being obsolete and insecure). Challenge 3: Different dimensions of knowledge are scattered across different sources (e.g., platforms), it becomes difficult for developers to effectively retrieve and leverage such knowledge. My long-term research goal is to improve the management and leveraging of big SE knowledge to support software development and maintenance. This proposed research program aims to address the above-mentioned challenges and has three short-term research objectives (ROs). RO1: Towards improving the organization of crowdsourced SE knowledge activities. My HQPs and I will conduct analysis to understand the characteristics of a good user-perceived organization for knowledge while proposing techniques to improve the organization of the knowledge. RO2: Ensuring the quality of crowdsourced SE knowledge. To maintain the quality of knowledge, typically for code examples in terms of obsolescence and security, I will propose automated techniques to detect obsolete and insecure code examples, and recommend solutions by leveraging software development history. RO3: Towards effectively leveraging the diverse SE knowledge. I will develop an infrastructure for gathering and linking SE knowledge from various sources and develop techniques for supporting multi-dimensional search for SE knowledge. The outcome of this research will provide solutions towards improving the management and leveraging of crowdsourced SE knowledge. It will benefit millions of learners in Canada and worldwide, and significantly improve software quality and reduce software development costs among Canadian companies. The proposed research will train ten highly qualified personnel (HQP) and prepare them with skills in software engineering and data science, the most in--demand skills in academic and industry settings.