Data-Driven Feedback Generator for Online Programing Courses
Ke Wang, Benjamin Lin, Bjorn Rettig, Paul Pardi, and Rishabh Singh
Manually providing feedback for programming assignments is a tedious task in traditional classroom education. The challenge increases drastically in Massive open online courses (MOOCs), where the student-teacher ratio can reach thousands to one or even millions to one. Despite the necessity, the current automated feedback approaches suffer from significant weaknesses: inability to scale to larger programs, manual involvement of teacher effort, and lack of precision for pin-pointing errors. We present a technique to tackle these challenges by developing a data-driven automated grader, iGrader, capable of generating instant and precise feedback for programming assignments.