VERITY: DEVELOPMENT OF AN INTEGRATED MULTI-ROLE ONLINE EXAMINATION PLATFORM WITH AI-BASED PROCTORING, PLAGIARISM DETECTION, AND LLM-CONTENT ANALYSIS

Abstract

The rapid growth of online examinations has highlighted a rather disjointed problem in academic integrity enforcement: proctoring systems, plagiarism detectors, and large language model (LLM) content detectors are each deployed as stand-alone tools, leaving institutional reviewers to reconcile evidence across disconnected dashboards. This paper presents Verity, an integrated multi-role online examination platform that combines AI-based proctoring, multi-layer plagiarism detection, and LLM-based content analysis in a single architecture. The system has three role-separated user interfaces, student, proctor, and administrator, tied to a common Unified Student-Examination Record (USER) data model. Verity was developed using the Design Science Research methodology and assessed through a 28-participant System Usability Scale (SUS) study and a three-expert heuristic review. The findings reveal an average SUS score of 80.1 (Good) with no participant scoring below 70 and no major usability violations remaining to be fixed. This research offers a replicable architecture and a prototype that leads the way in the design of trustworthy and scalable online assessments.