3 ms·
It's way easier to compute betweenness centrality? Eigenvector centrality requires inverting a potentially giant link matrix -- a notoriously expensive operati
by frisco 17y ago
It's way easier to compute betweenness centrality? Eigenvector centrality requires inverting a potentially giant link matrix -- a notoriously expensive operation, which says nothing of the additional code complexity needed to do it properly.
- brendano 17y agobetweenness centrality also has a nice interpretation that might make some sense for this application. or so.
- jey 17y agoYou can get the first eigenvector through power iteration, just take some random vector then repeatedly normalize it and multiply by the adjacency matrix. So no giant matrix inversion needed.
- frisco 17y agoAlso, I should point out that his reasoning for betweenness over eigvector centrality was given in a link in the article: http://www.stoweboyd.com/message/its-betweenness-that-matters-not-your-eigenvalue-the-dark-ma.html http://www.stoweboyd.com/message/its-betweenness-that-matter...