CAREER: Towards Understanding and Improving Crowd-based Software Video Tutorials

点击翻译按钮获取中文摘要

One of the most common and time-consuming activities performed by software programmers is searching for information to help them with their programming tasks. With the widespread use of the internet, programmers' information seeking is shifting more and more towards web resources, especially when knowledgeable colleagues or good internal documentation are not available. Information found in online resources has become indispensable for programmers, and online programming video tutorials are an important resource frequently consulted by both novice and experienced programmers. Therefore, it is of paramount importance to ensure that programmers can search, access, acquire, understand, and use this information efficiently and to develop tools and techniques that facilitate these actions. To achieve this one must first find answers to a few important questions. How do programmers use video tutorials to support their tasks? What information do they find useful? What tasks are improved? How one can better support programmers' information needs through leveraging video tutorials? The goal of this project is to address existing gaps and improve the body of knowledge about the topic coverage, usage, and quality of crowd-based programming video tutorials. The project will develop techniques to improve the way video tutorials are used and created by programmers. The outcomes of this project will transform the way programmers and computer science students acquire knowledge from video tutorials. The goal is to support better learning and instruction in software engineering settings and in the classroom. This can lead to time and effort savings that in turn can result in decreased software costs. The project will also organize a Scratch summer coding camp for middle schoolers which aims to broaden participation in the field of Computer Science. The project will focus in scope on publicly available video tutorials related to popular programming languages. The goals will be achieved through the application of novel techniques and tools that facilitate the extraction of correct source code, audio transcripts, and the topic coverage of programming video tutorials. A taxonomy will be developed to characterize the properties of quality programming video tutorials. Evaluation will be done using interviews, user studies, and surveys. A novel model of the quality of crowd-based programming video tutorials will be synthesized from user data and data automatically mined from video tutorials. The research will involve the application of techniques for information retrieval, program comprehension, and natural language processing.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.