A Novel Smart CBT Model for Detecting Impersonators using Machine Learning Technique

The computer-based testing (CBT) platforms for conducting mass-driven examinations over computer networks in order to eliminate certain challenges such as delay in marking, misplacement of scripts, impersonation, monitoring and so on associated with the conventional Pen and Paper Type (PPT) of examination has also been seriously bedeviled with the same issue of impersonation commonly associated with the PPT system. The existing CBT systems relies solely on the CCTV system for monitoring people passively and the human invigilators (Proctors) for going round the examination halls in order to physically confirm the students face against their pictures on their respective system dashboard which takes so many time and effort just to screen people against impersonating and yet impersonation is on the increase with CBT system. The proposed Smart CBT model integrates an intelligent agent assessor to the existing CBT model using K-Nearest Neighbor (KNN) machine learning technique for detecting a likely case of impersonation threat considering the considering the level of accuracy and response time in answering the questions the agent delivers to the students shortly before the actual examination can commence. A total of 3,083 dataset was gathered, and 80% (2,466) of the dataset was used for training the model, while 20% (617) dataset was used in testing the model to enable the model detect unseen data correctly. Results revealed that 99.99% accuracy rate, precision, recall and f-score were obtained. The propose Smart CBT model is recommended for all tertiary institutions and commercial CBT software product adoption.