Visualizing the workings of agent-based models: Diagrams as a tool for communication and knowledge acquisition

Abstract

One of the biggest challenges in developing agent-based models (ABMs) is the acquisition of domain knowledge. Interviews, or other types of face-to-face communication, are one of the ways of doing it but problems might arise if we deal with tacit knowledge and if the interviewer and the interviewee come from a very different cultural and educational background. This research proposes a visualization approach to be used in the interaction between domain and modeling experts that will facilitate communication in these kinds of situations. The approach is based on the findings of knowledge visualization and related fields, as well as on a firm understanding of ABMs. Three important aspects of ABMs that need to be visualized in order to facilitate understanding were identified. In a case study, performed in relation to a project where a spatial ABM was developed, three node-link diagrams were created according to the developed approach. They visualize the conceptual structure, the simulation process, and the data model of the ABM. These diagrams were positively received by the stakeholders of the project and they improved workflow and communication in the project.