Multivariate Graph Visualization
Collaboration network of researchers at ALCF since 2014. Data courtesy Argonne National Laboratory.
Visualizing large real-world networks, such as social networks andscientific collaboration networks, is challenging not only becausethey contain large numbers of nodes and links but also due to theirmultivariate nature. Applications that analyze such datasets tend tofocus on problems related to visualizing either multiple attributes onnodes or the topology of the network. Very few applications focuson both. This research explores a new approach to visualizing suchmultivariate networks using a sunburst chart to encode attributeson the nodes and a combination of a treemap layout and a suitablegraph layout to control the topology. We show the results of thisapproach by creating a collaboration network using a dataset thatcomprises references to all research papers published by users of theArgonne Leadership Computing Facility in the last three years. Thegoal of this visualization is to show a holistic view of the scholarlywork from a research facility, which in turn helps to identify researchgroups and the researchers acting as bridges among them.
The ddiLab Graph Visualization Team:
This research was supported in part by the U.S. Department of Energy Office of Sciences Advanced Scientific Computing Researchprogram, under Contract DE-AC02-06CH11357.