Plagiarism on Mobile Programming Assessments: Automated Detection

In programming education, programming ethics are maintained by educating students about the matter and penalize those who breach the regulations. There are a number of plagiarism detectors available for programming. However, none of them are focused on mobile programming assessments though mobile programming becomes more popular nowadays. This paper presents a plagiarism detector for such assessments, focusing on Dart and Flutter. The detector tokenize student submissions, generalize identifier names and constants, and compare the submissions in a pairwise manner with MinHash. Although our detector cannot deal with some plagiarism disguises (program statement replacement and logic change), it can result in no less than 85% top-K precision on most occasions. It is quite efficient since it can perform 2400 comparisons in three seconds. The processing time might be longer for more submissions with larger amount of code.