Reducing the Gap Between the Conceptual Models of Students and Experts Using Graph-Based Adaptive Instructional Systems

Abstract

Decision-making in complex environments requires a detailed understanding of causality to avoid unintended consequences and consider multiple scenarios. Concept mapping is of the key tools to support such decision-making activities. A system is abstracted as a ‘map’ or graph, in which relevant factors are represented as nodes and connected through causal links. Although studies have shown that concept maps help students recall the evidence and apply the knowledge gained to new cases, barriers such as the difficulty of assessment have prevented a wide adoption of concept maps by instructors. Previous works have partly addressed this concern by using graph algorithms to automatically assess or score a student’s map, but instructors are often more interested in correcting their students’ misconceptions than in only counting mistakes. In this paper, we design and implement a system to automatically guide students in modifying their conceptual models such that they get increasingly similar to the models produced by experts on the same problem. Although a simple automatic feedback may tell students to add/remove concepts or links based on whether they appear in the expert’s map, our approach combines expertise from systems science and participatory modeling to help students better understand why low-level changes are motivated by higher-level structures such as loops and alternative paths.