Eigenvector centrality in weighted graphs

Mar 13, 2010 at 3:06 AM

First of all , congrats to the team for this fantastic package !

I would like to compute eigenvector centralities on w weighted graphs ( weight number of emails between 2 correspondents) and I'm unclear what edge weight if any Eigenvector autofill takes into account. seems that it's unweighted ? Am I right ? Any workaround ? Thanks



Mar 13, 2010 at 6:10 PM


You are correct: NodeXL's eigenvector centrality calculations do not take edge weights into account.  I don't know of any workaround for this, but I've put in a feature request for it.

-- Tony

Mar 15, 2010 at 6:47 AM

Thanks Tony for a quick answer. If the Eigenvector package exists, it must be quite easy ( at least for symmetric matrices ) but would need to clarify what column is used as weight in NodeXL. I'm working in Telecom, and thereare very strong contrasts between a 1 minute conversation and a 1 hour one... so weighting wth duration ( for eigenvectors, but also for degree  I guess ) is critical.

Congrates again