LUD: An Automatic Scoring and Feedback System for Programming Assignments

Abstract

The increase in usage of online learning systems, has caused a renewed interest in using computers to provide more support to the student's learning experiences, allow instructors to focus on activities that require human intervention, and enable courses with a large number of students to help them achieve the learning objectives. One experience that has been particularly complex to emulate in online environments is assignment grading. The grading cycle tends to be a weak point in the experience. Among its drawbacks, two main issues are potential turnaround time and that assignments tend to provide only one opportunity to show understanding of the course content. In the context of programming assignments, this experience is particularly problematic as the programming cycle tends to be an iterative process. This paper discusses the initial implementation of an automatic scoring and feedback system for programming assignments. The system includes feedback on syntax, semantics, and code structure. We explain the architecture of the system and the results of an experiment run with 20 students that shows the effects of the system. In the experiment, we observed that the feedback system improves student performance in the assignment as measured by the grade assigned to them.