Automatic Scoring System of Fill-in-the-Blank Tests to Measure Programming Skills

Abstract

Programming learners in an educational institute are evaluated with test results to measure their skills. The tests should be able to measure the skills of learners accurately. We pay attention to a fill-in-the-blank test, which reduces a burden on both teachers and learners because they focus on specific code fragments in programs. However, there are problems that its blanks have various correct answers. Since it takes times to evaluate tests, learners cannot receive appropriate educational effect. In this paper, we propose an automatic scoring system for fill-in-the-blank tests promoting to reduce burden on both teachers and learners. The system evaluates answers filling blanks from behavior of the program when they are applied to corresponding parts of model programs, to score blanks that have a lot of correct alternatives. We conduct an experiment to verify whether our automatic scoring system for the fill-in-the-blank test promotes learners to achieve higher understanding for the programming. The experimental results show that the system covers various correct answers and gives learners chances to improve their understanding for programming skills