Re: [igraph] betweeness centrality

[Matthew Galati]

I guess you are right. I just took the formula from Freeman's classic
paper, and never realized that it only works for undirected graphs,
for directed graphs you need to halve that, because there are twice as
many shortest paths, right?

Yes. I think the normalization should be 1/((n-1)(n-2)). And, for undirected, if you double count (which you probably do), then you need the 2.