Dynamic network visualization in 1.5D

Abstract

The dynamic network visualization has been a challenging topic due to the complexity introduced by the extra time dimension. Existing solutions to this problem are usually good for the overview and presentation, but not for the interactive analysis. We propose in this paper a new approach which only considers the dynamic network central to a focus node (aka dynamic ego network). The navigation of the entire network is achieved by switching the focus node with user interactions. With this approach, the complexity of the compressed dynamic network is greatly reduced without sacrificing the network and time affinity central to the focus node. As a result, we are able to present each dynamic ego network in a single static view, well supporting user analysis on temporal network patterns. We describe our general framework including the network data pre-processing, 1.5D network and trend visualization design, layout algorithms, as well as several customized interactions. In addition, we show that our approach can also be extended to visualize the event-based and multimodal dynamic networks. Finally, we demonstrate, through two practical case studies, the effectiveness of our solution in support of visual evidence and pattern discovery.