Search for: All records

Students in introductory computer science courses often need individualized help when they get stuck solving programming problems. But providing such help can be time-consuming and thought-intensive, and therefore difficult to scale as Computer Science classes grow larger in size. Automatically generated fixes with explanations have the potential to integrate into a variety of mechanisms for providing help to students who are stuck on a programming problem. In this paper, we present a data-driven algorithm for generating explainable fixes to student code. We evaluate a Python implementation of the algorithm by comparing its output at different stages of the algorithm to state-of-the-art systems with similar goals. Our algorithm outperforms existing systems that can analyze and fix beginner-written Python code. Further, fixes it generates conform very well to corrections written by human experts for an existing benchmark of code correction quality. 

Undergraduate Teaching Assistants(TAs) in Computer Science courses are often the first and only point of contact when a student gets stuck on a programming problem. But these TAs are often relative beginners themselves, both in programming and in teaching. In this paper, we examine the impact of availability of corrected code on TAs’ ability to find, fix, and address bugs in student code. We found that seeing a corrected version of the student code helps TAs debug code 29% faster, and write more accurate and complete student-facing explanations of the bugs (30% more likely to correctly address a given bug). We also observed that TAs do not generally struggle with the conceptual understanding of the underlying material. Rather, their difficulties seem more related to issues with working memory, attention, and overall high cognitive load.