Hi,
I am performing community detection on citation network graphs (~20k nodes). It seems like all (most?) community detection algorithms are based on modularity which according to this paper (
http://dl.acm.org/citation.cfm?id=2350193) is a bad idea. They propose conductance (or e.g. triangle participation ratio) as a metric to optimize for communities. In particular I am interested in a score for maximum community saliency (or e.g. minimum conductance cut).
Does iGraph have such capabilities? I could find anything about conductance in the docs.
I believe the Stanford SNAP library has similar functionality (C++) but I would prefer staying with Python if possible.
Any comments and ideas are very welcome!
Thanks,
Tim