Simulating Student Mistakes to Evaluate the Fairness of Automated Grading

摘要

The use of autograding to assess programming students may lead to unfairness if an autograder is incorrectly configured. Mutation analysis offers a potential solution to this problem. By simulating student coding mistakes, an automated technique can evaluate the fairness and completeness of an autograding configuration. In this paper, we introduce a set of mutation operators to be used in such a technique, derived from a mistake classification of real student solutions for two introductory programming tasks.