Networks analysts often need to compare nodes in different parts of a network. When zoomed to fit a computer screen,
the detailed structure and node labels of even a moderately-sized network (say, with 500 nodes) can become invisible
or difficult to read. Still, the coarse network structure typically remains visible, and helps orient an analyst’s zooming,
scrolling, and panning operations. These operations are very useful when studying details and reading node labels, but in
the process of zooming in on one network region, an analyst may lose track of details elsewhere. To address such problems,
we present in this paper multi-focus and multi-window techniques that improve interactive exploration of networks. Based
on an analyst’s selection of focus nodes, our techniques partition and selectively zoom in on network details, including
node labels, close to the focus nodes. Detailed data associated with the zoomed-in nodes can thus be more easily accessed
and inspected. The approach enables a user to simultaneously focus on and analyze multiple node neighborhoods while
keeping the full network structure in view. We demonstrate our technique by showing how it supports interactive debugging
of a Bayesian network model of an electrical power system. In addition, we show that it can simplify visual analysis of an
electrical power network as well as a medical Bayesian network.
Information-rich data sets bring several challenges in the areas of visualization and analysis, even when associated with node-link network visualizations. This paper presents an integration of multi-focus and multi-level techniques that enable interactive, multi-step comparisons in node-link networks. We describe NetEx, a visualization tool that enables users to simultaneously explore different parts of a network and its thematic data, such as time series or conditional probability tables. NetEx, implemented as a Cytoscape plug-in, has been applied to the analysis of electrical power networks, Bayesian networks, and the Enron e-mail repository. In this paper we briefly discuss visualization and analysis of the Enron social network, but focus on data from an electrical power network. Specifically, we demonstrate how NetEx supports the analytical task of electrical power system fault diagnosis. Results from a user study with 25 subjects suggest that NetEx enables more accurate isolation of complex faults compared to an especially designed software tool.