Using singular value decomposition for graph clustering

Question

I have a fairly large graph (50-60 vertices) with directed, weighted edges, and I am attempting to cluster the vertices. Prior to this, I have only worked with undirected graphs having symmetric adjacency matrices, and I am struggling to figure out an effective means of clustering for the asymmetric case. It seems like the "SingularValueDecomposition" command may be helpful, but I have no experience with this process, and I'm struggling to find an adequate tutorial online. If anyone knows of such a tutorial (or has suggestions for another means of clustering), I would appreciate the help.

Answer 1

Here's a tutorial on using the SVD to do clustering...

http://meyer.math.ncsu.edu/Meyer/Talks/ClusteringMonopoli_9_11_07.pdf