Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A System for Interactive Visual Analysis of Large Graphs Using Motifs in Graph Editing and Aggregation

 
: Landesberger, Tatiana von; Görner, Melanie; Rehner, Robert; Schreck, Tobias

Magnor, M.A.:
Vision, Modeling, and Visualization Workshop 2009. Proceedings : November 16 - 18, 2009 Braunschweig, Germany
Magdeburg: Otto-von-Guericke-Universität Magdeburg, 2009
ISBN: 978-3-9804874-8-1
S.331-339
Workshop Vision, Modeling, and Visualization (VMV) <14, 2009, Braunschweig>
Englisch
Konferenzbeitrag
Fraunhofer IGD ()
interactive data analysis; visual analytics; graph visualization

Abstract
Network analysis is an important task in a wide variety of application domains including analysis of social, financial, or transportation networks, to name a few. The appropriate visualization of graphs may reveal useful insight into relationships between network entities and subnetworks. However, often further algorithmic analysis of network structures is needed.
In this paper, we propose a system for effective visual analysis of graphs which supports multiple analytic tasks. Our system enhances any graph layout algorithm by an analysis stage which detects predefined or arbitrarily specified subgraph structures (motifs). These motifs in turn are used to filter or aggregate the given network, which is particularly useful for search and analysis of interesting structures in large graphs. Our approach is fully interactive and can be iteratively refined, supporting analysis of graph structures at multiple levels of abstraction. Furthermore, our system supports the analysis of data- or user-driven graph dynamics by showing the implications of graph changes on the identified subgraph structures. The interactive facilities may be flexibly combined for gaining deep insight into the network structures for a wide range of analysis tasks. While we focus on directed, weighted graphs, the proposed tools can be easily extended to undirected and unweighted graphs. The usefulness of our approach is demonstrated by application on a phone call data set.

: http://publica.fraunhofer.de/dokumente/N-113128.html