About
COMPGame
A flipped-learning platform for COMP3423 Human–Computer Interaction at PolyU: learn each concept by playing with it first, then test yourself, with an AI tutor available throughout.
Why it exists
Traditional teaching tests you on material you have already been lectured at. COMPGame inverts that: you meet a concept inside an interactive activity, build an intuition for it, and only then check what you understood. Whether that ordering actually helps is not assumed here — it is measured, topic by topic, and reported honestly either way.
What is in it
Thirteen topics from the COMP3423 syllabus, released in lecture order: Fitts' Law, Gestalt principles, Hick's Law, Miller's 7±2, the principle of consistency, Weber's Law, Norman's action cycle, mental models, problem solving, visual perception, language and ambiguity, ergonomics, and experiment design. Each has an Understanding activity and an Assessment.
How it is built
The interface is Next.js, React, TypeScript and Tailwind CSS. The tutor is a retrieval-augmented pipeline — FastAPI and LangChain over a ChromaDB vector store built from the course lecture slides, answering through a locally hosted Ollama model. Nothing a student writes leaves the course machine.
Questions or problems
If you cannot sign in, or something here is not working, tell your course team — they can check your student ID against the class list. Questions about the study itself, including how to withdraw, are answered on the consent page you saw when you first signed in.