User identification through keystroke dynamics as part of automated proctoring systems

Abstract

Subject. Popular online courses and testing programs integrate into correspondence education systems, which are more often than not based on automated proctoring. What makes the latter vulnerable is user identification. Objectives. We examine user identification methods through keystroke dynamics and devise a more accurate and effective technique for user identification through keystroke dynamics. Methods. The article sets out a three-tiered model for identifying users more accurately not only in automated proctoring environments, but also in critically sensitive locations. Results. We had an experiment, which showed a 97.5 percent accuracy of user identification. We significantly reduced illegitimate users at the statistical level of the three-tiered model. Conclusions and Relevance. Following the study, it is possible to develop a logic comparison method for higher accuracy. It will serve for creating a more refined model, which would accommodate for distinctions of each user and some deviations of users’ emotions. This would contribute to continuous user identification systems to monitor their emotional condition at critical locations.