Development of a Fair and Efficient Online Exam Generation System for the 'Database Systems' Lecture
The outbreak of COVID-19 forced universities in Austria and all over the world to adapt their teaching of in-person courses to an online mode. This also includes holding the final exams online instead of in the premises of the university in a supervised fashion. Obviously, this leads to increasing incidents of academic dishonesty as cheating during an online exam is much more easier and tempting for students. Although, examiners try to minimize the cheating opportunities (i.e. by observing them with a webcam and a microphone), they cannot completely exclude cheating activities since proctoring procedures are often insufficient or restricted by law. The only way to reduce the opportunities to a minimum is by generating unique exams for every student, which of course involves new challenges like generating fair exams with the same level of difficulty. The goal of this master thesis is to develop such a fair generation system for the mandatory course 'Database Systems'. The generation system will be based on a question pool. Therefore, outstanding issues like: How many questions are needed for the question pool? or How long can a question be reused? need to be initially investigated. It is very likely, that the existing question pool must be extended.