GraphScape: integrated multivariate network visualization

@inproceedings{citeulike:2439621, abstract = {In this paper, we introduce a new method, GraphScape, to visualize multivariate networks, i.e., graphs with multivariate data associated with their nodes. GraphScape adopts a landscape metaphor with network structure displayed on a 2D plane and the surface height in the third dimension represents node attribute. More than one attribute can be visualized simultaneously by using multiple surfaces. In addition, GraphScape can be easily combined with existing methods to further increase the total number of attributes visualized. One of the major goals of GraphScape is to reveal multivariate graph clustering, which is based on both network structure and node attributes. This is achieved by a new layout algorithm and an innovative way of constructing attribute surface, which also allows visual clustering at different scales through interaction. A simplified attribute surface model is also proposed to reduce computation requirement when visualizing large networks. GraphScape is applied to networks of three different size (20, 100, and 1500) to demonstrate its effectiveness}, author = {Xu, K. and Cunningham, A. and Hong, S. H. and Thomas, B. H. }, booktitle = {Visualization, 2007. APVIS '07. 2007 6th International Asia-Pacific Symposium on}, citeulike-article-id = {2439621}, doi = {10.1109/APVIS.2007.329306}, journal = {Visualization, 2007. APVIS '07. 2007 6th International Asia-Pacific Symposium on}, keywords = {graph, multivariate, visualization}, pages = {33--40}, posted-at = {2008-02-28 03:35:13}, priority = {0}, title = {GraphScape: integrated multivariate network visualization}, url = {http://dx.doi.org/10.1109/APVIS.2007.329306}, year = {2007} }

TY - CONF ID - citeulike:2439621 L3 - citeulike-article-id:2439621 TI - GraphScape: integrated multivariate network visualization T2 - Visualization, 2007. APVIS '07. 2007 6th International Asia-Pacific Symposium on JF - Visualization, 2007. APVIS '07. 2007 6th International Asia-Pacific Symposium on SP - 33 EP - 40 N2 - In this paper, we introduce a new method, GraphScape, to visualize multivariate networks, i.e., graphs with multivariate data associated with their nodes. GraphScape adopts a landscape metaphor with network structure displayed on a 2D plane and the surface height in the third dimension represents node attribute. More than one attribute can be visualized simultaneously by using multiple surfaces. In addition, GraphScape can be easily combined with existing methods to further increase the total number of attributes visualized. One of the major goals of GraphScape is to reveal multivariate graph clustering, which is based on both network structure and node attributes. This is achieved by a new layout algorithm and an innovative way of constructing attribute surface, which also allows visual clustering at different scales through interaction. A simplified attribute surface model is also proposed to reduce computation requirement when visualizing large networks. GraphScape is applied to networks of three different size (20, 100, and 1500) to demonstrate its effectiveness KW - graph KW - multivariate KW - visualization AU - Xu, K AU - Cunningham, A AU - Hong, SH AU - Thomas, BH PY - 2007/// UR - http://dx.doi.org/10.1109/APVIS.2007.329306 ER -