Evaluating user cognition of network diagrams

Abstract

Edges crossing and nodes overlapping have a significant effect on the users’ recognition and comprehension of network diagrams. In this study, we propose a visual evaluation method for users’ cognition of network diagrams. First, this method carries out a set of cognitive experiments to collect the user’s cognitive performance that affects the variables, including accuracy and response time. The user’s pupil diameter is measured through an eye tracker to reflect their cognitive load. Second, the significance test points out the visual features as independent variables and then establishes an evaluation regression model. The experimental results show that the number of edges, edge length, node visual interference, and edge occlusion contribute to the evaluation models of response time, and edge occlusion and the number of node connections contribute to the accuracy model. Finally, these evaluation models demonstrate good predictability when assessing users’ cognition of network diagrams and provide practical recommendations for their use.