Clustering by edge weight

Mar 17, 2014 at 4:38 PM
Hi everybody,

I wanted to analyze a facebook's fan page using the user to user graph and the post to post to graph. At first, I thought that the clusters created in the user to user graph would have their correspondence with the clusters created in the post to post graph. Obviously, I was wrong.

My goal is to observe, at fist glance, which subjects are discussed on each user cluster. But I didn't take into count that a single facebook profile can modify the whole post to post network.

It's not very intuitive the fact that the two most related posts (in terms of edge weight) appear in different clusters in the post to post network... so my question is:

Is there any way I can take the edge weight as a variable used to construct the clusters in the post to post graph?

Thanks in advance!