TY  - CONF
T1  - Cluster-based Visualization of Dynamic Graphs
A1  - Held, Pascal
A1  - Hempel, Julia
A1  - Kruse, Rudolf
ED  - Hoffmann, Frank
ED  - Hüllermeier, Eyke
TI  - Proceedings. 23. Workshop Computational Intelligence, Dortmund, 5. - 6. Dezember 2013
T3  - Schriftenreihe des Instituts für Angewandte Informatik - Automatisierungstechnik, Karlsruher Institut für Technologie
Y1  - 2013
VL  - 46
SP  - 21
EP  - 37
PB  - KIT Scientific Publishing
AD  - Karlsruhe
SN  - 978-3-7315-0126-8
SN  - 1614-5267
UR  - http://digbib.ubka.uni-karlsruhe.de/volltexte/1000036887
M2  - doi: 10.5445/KSP/1000036887
N2  - Graph visualizations are applied for describing relations between objects in many application fields, e.g., in social network analysis and software visualization. Several clustering strategies can be used to identify groups of objects automatically. On the one hand, visualizing these clusters is useful to analyze and evaluate clustering algorithms. On the other hand, cluster visualization allows a fast estimation of similarity between objects and provides orientation in the graph. Because objects, relations and clusters might change over time, dynamic graph drawing received significant interest in the last decades. Several algorithms have been proposed enhancing well-known static layout algorithms. However the dynamic drawing of clusters in graphs is less considered. In this work, we propose three layout algorithms for dynamic clustered graphs. While two approaches are based on enhancing a force-directed layout, the third one uses a divide-and-conquer approach. The approaches are evaluated and compared based on different metrics. The results suggest that the divide-and-conquer approach is best suited for the dynamic drawing of clustered graphs since it separates the clusters well and stabilizes the layout.
ER  -