p.s. I think the measure of edge betweenness was first defined by anthonisse
in 1971. He called it the "rush". It was then implemented in the Gradap
social network analysis package. steve.
----- Original Message -----
From: "Ed Vielmetti" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Friday, February 22, 2002 11:46 PM
Subject: link centrality (by analogy with other centrality metrics)
> I'm experimenting with locating clusters in a network and
> analyzing a structure for fragility by removing links between
> pairs of nodes that each have high betweenness. The algorithm
> is pretty easy; look at the list of nodes ranked by betweenness,
> determine if the top two nodes have a link, if so snip it; if
> not, search iteratively through the list for the pair of links
> with the highest combination.
> Is there any standard metric that computes link centrality
> directly? (A reference would be good.) Intuitively it would
> seem to be the "path that's in the most geodesics in the
> network", and so it should be just about as hard to compute as
> node centrality. Google on "link centrality" turns up nothing.
> The motivation for thinking of links as items in their own right
> is simply that from telecommunications networks, where the
> circuit between two devices is the source of at least as much
> cause for analysis as the endpoints.
> Edward Vielmetti
> Ann Arbor, MI
> [log in to unmask]