A Framework for Automated Assignment Generation and Marking for Plagiarism Mitigation

Most academic institutions have policies around academic integrity, and students who are found to have violated the policies are disciplined. In spite of this, a number of students cheat. The most common and easily detectable form is plagiarism, where someone else's work is copied across and claimed as one's own. Experience suggests that about 30% of the class might be plagiarizing, though some research point to as much as 70% cheating in various forms. Dealing with plagiarism is a highly time-consuming affair. Prior research observed high value low frequent assignments as the most plagiarized as opposed to low value high frequent ones. It is therefore desirable to have low value high frequent assignments so as to reduce plagiarism incidents, thereby reducing the time spent on dealing with detected plagiarism cases. This paper discusses the implementation of an automated assignment generation and marking framework that is able to deliver high frequent assignments and automatically grade the submitted solutions. More importantly, the framework supports personalized assignments so that every student gets a different problem set to solve. This means that blindly copying answers from another student will not help gain any mark. The paper briefly shares some of the experience using the framework in engineering and science, where staff and students felt positively about the system and observed a huge reduction in plagiarism incidents. The reduction in the incidents resulted in saving a large amount of time that would have otherwise been spent on dealing with the incidents.