Publication

Abstract : Massive open online courses or MOOCs have gained significant prominence as an e-learning platform. With large participation in MOOCs, an instructor is challenged with a gigantic task of assessing submissions for correctness and authenticity. For offering reputed and high-quality course completion certificates, a plagiarism detector is a must. In this paper we propose a four-phase plagiarism detection system that operates on large-scale submissions. This system exploits the dual but independent concepts of geodata and fast string-matching algorithms on top of big data platforms to solve this problem to a large extent. Given the context of these learning systems, and very high volume student participation, we are able to apply several optimizations in checking for plagiarism. This system is intended to benefit the course instructors by encouraging them to post advanced levels of courseware and more complicated assignments without bothering much about evaluation-related issues such as plagiarism.