EduArchive AI: AI-Powered Educational Archive & Exam Analytics System Using Natural Language Processing

About the Project

Universities keep past examination papers as static PDF files that can only be found by course code, title and year. Students cannot search questions by topic, conventional OCR corrupts mathematical notation, and lecturers still check exam quality by hand. EduArchive AI is a web and mobile system that turns past papers into a searchable, analysable question bank. Lecturers upload papers as PDF, Word, PowerPoint or phone-camera scans; a multimodal large language model extracts every question with its mathematics preserved as LaTeX and tags its Bloom's taxonomy level and difficulty. Students can search by meaning rather than exact keywords and ask an AI chat assistant that answers only from real past-year questions and shows its sources. The core innovation is the Examination Quality Index (EQI), which scores each paper out of 100 on six indicators: Bloom's level, syllabus coverage, question diversity, difficulty mix, originality against earlier papers, and alignment with Course Learning Outcomes. The difficulty mix is plotted on a 2-simplex triangle, and a revision advisor tells lecturers exactly which question to change and how much the score would improve. The system is built with React, Node.js, MongoDB, Python, XGBoost and React Native, using free and open-source tools.