ML + Web Application
ExamGuard AI
Behavioral risk classification for online exams
Overview
A CBT (computer-based testing) platform that tracks how students interact with the exam interface and produces an automated behavioral-risk classification for every completed session.
Problem
Online exams remove the physical invigilation of a paper exam hall, leaving lecturers with no automated signal about which sessions showed behavior worth a closer look.
Solution
ExamGuard AI tracks 12 behavioral metrics client-side during every exam attempt (tab switches, copy/paste, mouse movement, idle time, and more), then runs a trained Decision Tree classifier on submission to flag each session as Normal or Suspicious — a heuristic risk signal, not a verified cheating determination.
Key Features
- Role-separated CBT platform for admins, lecturers, and students
- Session-persistent question and option ordering per exam attempt
- Real-time behavioral tracking across 12 metrics
- Automated Decision Tree-based risk classification on every submission
- Lecturer results dashboard with per-student behavioral and prediction detail
- Excel and PDF result exports
Technology
Backend
Database
AI / ML
Frontend