ABSTRACT
In the past few decades E-learning in higher education is increased and played a vital role in pandemics like COVID-19. Particularly, online examinations are conducted on e-learning platforms which leads to many security and cheating issues. For this reason, numerous research is available proposed methodologies and techniques for seamless execution of online examination. However, it is hard to find any study that provides the latest systematic literature review of anti-cheat or cheating prediction techniques and the approaches in the literature. We have analyzed 2223 studies. However, after applying inclusion and exclusion criteria 23 studies relevant studies are finalized. The review revealed that there are three types of proctoring, fully live online, recorded & reviewed and fully automated. This study provides a comparative analysis of online examination techniques & tools performed on 23 studies from the last five years 2017 to 2021. Furthermore, in this time duration five leading cheating prevention features are identified.14 important techniques which are mostly used in this time duration are found in which best frequent approach used in literature is NLP and 10 data sets including both public and private are identified. Proceeding toward the proposed solution, a total of 20 tools for the anti-cheat examinations are found. Almost 23 leading existing tools were found in the literature. To narrow down the criteria for adoption factor is analyzed and studies of the online anti-cheat examination solution adoption in different countries are also investigated. Finally, the overall cost of the e-learning infrastructure, specifically the conduction of examinations is determined by comparing the key factors of the global adoption with major online exam features. © 2022 IEEE.
ABSTRACT
Since COVID 19, there have been significant advancements in the field of teaching and learning. Academic institutions are going digital to provide their students more resources. Due to technology, students now have more alternatives to study and improve skills at their own pace. In terms of assessments, there has been a shift toward online tests. The absence of a physical invigilator is perhaps the most significant impediment in online mode. Henceforth, online proctoring services are becoming more popular, and AI-powered proctoring solutions are becoming demanding. In this project, we describe a strategy for avoiding the physical presence of a proctor during the test by developing a multi-modal system. We captured video using a webcam along active window capture. The face of the test taker is identified and analyzed to forecast his emotions. To identify his head pose, his feature points are identified. Furthermore, aspects including a phone, a book, or the presence of another person are detected. This combination of models creates an intelligent rule-based inference system which is capable of determining if any malpractice took place during the examination. © 2022 IEEE.