Lessons learnt in applying automated code plagiarism detection in an introductory programming module

Abstract

This paper investigates automated code plagiarism detection in the context of an introductory programming module. Three methods for detecting plagiarism are compared to determine whether these systems yield differing results. These methods are the use of MD5 hashes and the application of two plagiarism detection systems, namely MOSS and NED. The same set of solutions to the same problem was evaluated, using each of the three methods. This set was selected as a representative sample as it was characteristic of most other data sets submitted by students in the introductory programming module over the course of four years. The discrepancies in the results obtained by these detection techniques were used to devise guidelines for effectively detecting code plagiarism.