Uncertainty-Aware Visualization of Networks

I will discuss the need of taking uncertainty into account in visual analytics, and present a methodology for incorporating uncertainty. I will show, in the case of visualizing and interacting with complex networks, two aspects of uncertainty, namely the stability and sensitivity of centrality metrics, are appropriate to represent the relative importance of a node with respect to other nodes. These properties allows an analyst to visualize at a glance competitive or cooperative clusters, where small changes in parts of the clusters have a negative or positive influence on other clusters. In addition, this methodology provides more robust graph simplification, such that the analyst can obtain a simpler representation of a complex graph that preserves the centrality of its nodes.