Generating and Evaluating Collective Concept Maps

Abstract

Concept maps are used in education to illustrate ideas and relationships among them. Instructors employ such maps to evaluate a student’s knowledge on a subject. Collective concept maps have been recently proposed as a tool to graphically summarize a group’s rather than an individual’s understanding on a topic. In this paper, we present a methodology that automatically generates collective concept maps, which relies on grouping similar ideas into node-clusters. We present a novel clustering algorithm that is shown to produce more informational maps compared to Markov clustering. We evaluate the collective map framework by applying it to sets of a total of 56 individual maps created by teachers (grades 2-12) and students (grades 6-11) during a week-long cybersecurity camp. Finally, we discuss how collective concept maps can support longitudinal research studies on program and student outcomes by providing a novel format for knowledge exchange. We have made our tool implementation publicly available.