University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Graduation Talks

title

Network Visualization: Dynamics, Modes, and Collections

speaker

Martin Mader, University Konstanz
Konstanz, Germany

date & place

Wednesday, 13.01.2010, 17:00 h
Room C252

abstract

As complex networks receive increasing interest by network analysts, there is a strong need for appropriate network visualizations that allow for visual analysis and exploration. This thesis is to present systematic approaches to obtain meaningful network representations for different categories of complex network visualization tasks. We focus on the problem of finding an appropriate graph layout, though the actual graphical representation of network objects may also play a decisive role in certain applications at hand. Many graph drawing algorithms have been developed to handle a variety of generic graph types and representation styles, e.g., planar or layered graphs, or orthogonal drawings, respectively. However, most real-world networks come along with additional semantic information, posing extra constraints on the layout and, thus, requiring adaption of the layout algorithm used.
One class of complex network visualization tasks is to obtain graph layouts for a dynamic network comprising a sequence of graphs. There have been proposed some methods to solve the problem of finding coherent representations of successive networks by adapting layout algorithms to respect constraints on changes in the network representations. We want to systematically compare the corresponding fundamental strategies to trade off between displaying structural properties and complying with additional constraints, and want to come up with new ones. Other classes of complex networks, where concepts similar to those used for dynamic network visualization may be adapted for specific visualization tasks, are multi-modal networks and network collections.
We have good reason to believe that an energy-based graph layout method called stress-minimization is an appropriate tool to handle most cases. The methods' flexibility enables us to systematically engineer its specifics to meet the requirements posed by the different layout tasks.