Multivariate Network Visualization Techniques

A companion website for the STAR Report on Multivariate Network Visualization Techniques.

Quilts
Placeholder image

A quilt is a tabular layout optimized for layered networks. Quilts are similar to an adjacency matrix in that nodes are represented as either rows or columns, and edges are shown in the cells at the intersection of the source and target nodes.

Optimized for networks with several node or edge attributes. Also ideal for tasks on single nodes and neighbors.

Supports tasks on paths, clusters, and sub-networks.

Not ideal for networks with more than 100 nodes, or dense networks.

Examples Figures from the Literature

Placeholder image
Bezerianos et al. 2010
Placeholder image
Bezerianos et al. 2010

Technique Scores

Reccommended Usage

Quilts are well suited for layered networks or k-partite networks where all partitions have connections to at most two layers. For these kinds of networks, quilts require less screen-space than matrices and have similar favorable properties in terms of attributes (see Table 2). Links between nonconsecutive layers, however, can be problematic to integrate. Albeit the class of networks suitable for quilts is small, they support all relevant tasks on these well.

Example Papers

    Watson B., Brink D., Stallman M., Devajaran R., Rhyne T., Patel H., Visualizing very large layered graphs with quilts. (2008), pp. 8, .

    Bezerianos A., Dragicevic P., Fekete J., Bae J., Watson B., GeneaQuilts: A System for Exploring Large Genealogies. IEEE Transactions on Visualization and Computer Graphics (2010), vol. 16, no. 6, pp. 1073-1081, doi:10.1109/TVCG.2010.159.

    Bae J., Watson B., Developing and Evaluating Quilts for the Depiction of Large Layered Graphs. IEEE Transactions on Visualization and Computer Graphics (2011), vol. 17, no. 12, pp. 2268-2275, doi:10.1109/TVCG.2011.187.