Evaluating overall quality of dynamic network visualizations

Visualizing dynamic networks is a challenging task. One of the challenges we face is how to maintain visual complexity and overall quality of visualizations at a reasonable and sustainable level so that the information about the network embedded in the visualization can be effectively comprehended by the viewer. Many techniques and algorithms have been proposed and developed to facilitate the discovery of changing patterns. Much research has also been done in investigating how visualization should be constructed to be effective. However, how to measure and compare the quality of visualizations of a changing network at different time points has not been well researched. In this paper, we report on a preliminary work towards this direction. In particular, we apply an existing multi-dimensional overall quality measure in a user study data of different networks and found that the measured quality is positively correlated with user task performance regardless of network size.