Social network visualization has the problem of being “globally sparse and locally dense”, making it a dilemma to satisfy both aspects. This paper proposed a “Node Link + Adjacency Matrix” framework for solving this problem.
For each dense subgraph, they use an adjacent matrix to represent the network while each node (visible along the row & column of the matrix) can be visually linked to the other node or subgraph. The whole picture consists of many subgraphs linked with sparse edges.
The visualization is editable in the following ways:
– Drag & drop of
** node to node (to form a 2×2 matrix),
** node to matrix (add 1 row and 1 column to an existing matrix), and
** matrix to matrix (merge two matrices).
– The label and size of matrix are resizable
– Cross-window drag & drop is enabled
One of the neat thing in this paper is the consideration of animating the transition between node link view and matrix view. One interesting thought is: nodes correspond to points in node-link diagrams but to line segments (rows and columns) in matrices and edges correspond to lines in node-link diagrams but to points in matrices. The question is likely to become transit between points and lines. The solution of this paper comes from the idea that we can also draw a node-link graph on a matrix (Figure 3). And such non-typical node-link serves as the intermediate view when animation happens.
Lastly, in a case study where the co-author relationship of InfoVis 2004 was visualized using NodeTrix.