Visualizing Student Game Design Project Similarities

Abstract

High dimensional cosine calculation is a tool that is often used to discover the similarity between two vectors in semantic space. This research uses vector similarities to create a novel way of visually representing the submitted work of a whole classroom of students over the course of a semester. Using a high dimensional cosine calculation, every student assignment submission is compared to one another in the Educational Game Design Class, an undergraduate/graduate programming class taught in Spring 2009 at the University of Colorado Boulder. This is accomplished by first converting every student submission into a representative vector based on submission project code. Through creating a visualization of these similarity scores, called a ‘Similarity Matrix’, interesting patterns begin to emerge indicating notable phenomena such as class ‘watershed moments’ and relative in-class effectiveness of presented programming concepts.