Authorship is continuous: detecting plagiarism in student code assignments with version control
In academia, the conflict between plagiarist and instructor is reaching a crescendo. The increasing availability of source material available from the Internet, either for free or for hire, is met by the increasing sophistication of tools that help with the detection of these activities. These plagiarism detection tools work by comparing sets of documents to identify similar passages of text, but this approach has some inherent limitations. Chief among these is the single source problem. If a student hires someone to write a term paper for him or her, clearly plagiarism has taken place. However, there is no second copy of this paper that can serve as evidence that the work is not the student’s own. Without this second copy, these plagiarism detection tools are unable to differentiate between a plagiarized paper and legitimate work by other students. By reframing the task of “preventing plagiarism” into “ensuring authorship,” we have developed an approach that is theoretically able to detect when single source plagiarism has taken place in programming assignments involving source code. Requiring students to provide copies of their code in progress, and analyzing changes in subsequent versions can prove when a student performed his or her own work. An implementation of this approach was developed and tested on three courses taught by the Department of Computer Science at the Florida Institute of Technology. Results of these trials, along with additional benefits to the use of such a system are discussed.