AI-powered platform that builds knowledge graphs from lecture slides and generates adaptive formative assessments
Ruhma Hashmi · Advisor: Dr. Yuan An · College of Computing & Informatics, Drexel University
AILA automates the creation of high-quality multiple-choice questions (MCQs) from lecture slides by first building a knowledge graph (KG) of the content, then using that structured representation to prompt an LLM. The platform delivers questions through adaptive quizzes calibrated to individual student mastery.
Built as a full-stack web app: Next.js frontend, FastAPI backend, Google Gemini LLM, SQLite database.
| Paper | Venue | Links |
|---|---|---|
| Rate-Distortion Guided KG Construction from Lecture Notes Using Gromov-Wasserstein Optimal Transport | IEEE BigData 2025 | arXiv · PDF |
| Scaling Retrieval Practice with LLM: Improving MCQ Quality through Knowledge Graphs | ACM SIGCSE TS 2026 | DOI · PDF |
| Enhancing Student Learning with LLM-Generated Retrieval Practice Questions: An Empirical Study | arXiv 2025 (co-authored) | arXiv · PDF |
| Event | Year | File |
|---|---|---|
| Drexel STAR Scholars Showcase | 2025 | |
| Pennoni Honors College Student Showcase | 2026 |
cd aila_backend
pip install -r requirements.txt
uvicorn aila_backend.main:app --reloadcd aila_frontend
npm install
npm run devSet your Google Gemini API key via
.envin the backend. Frontend runs onlocalhost:3000, backend onlocalhost:8000.
aila-star/
├── aila_backend/ # FastAPI + SQLite (KG pipeline, MCQ gen, quiz logic)
├── aila_frontend/ # Next.js / React (dashboards, concept map, quiz UI)
└── docs/
├── papers/ # Published research PDFs
└── posters/ # Conference & showcase posters
Supported by Drexel STAR Scholars Program (2025) and UREP Mini-Grant ($1,000, 2026). Advised by Dr. Yuan An, College of Computing & Informatics, Drexel University.
Ruhma Hashmi · rh927@drexel.edu · linktr.ee/ruhmahashmi