Re: [igraph] label propagation community

[Lin Freeman]

This confuses me.  The original method must at least begin at the first step by randomly choosing among the uniquely named adjacent and equally weighted nodes.  But your procedure would do that iff the node in question was adjacent to multiple nodes with equal weights.  Why not choose among adjacent nodes with a biased probability in which each adjacent node would be chosen proportional to its tie strength?

On Tue, Nov 9, 2010 at 12:01 PM, Tamas Nepusz <address@hidden> wrote:

Dear Prof. Freeman,

> The manual reports, "This version extends the original method by the ability to take edge weights into consideration . . ."  But it doesn't tell exactly how the weights are taken into account.  Where can I find that information?

Nowhere yet, but I'll add it to the documentation soon. Basically, when the method decides the new label of a node, it calculates the total weights of edges leading to other nodes of label 1, 2, ..., k (where k is the number of labels) and then chooses the one with the highest total weight. The original method does exactly the same thing while assuming that all edges have equal weight.

--
Tamas


_______________________________________________
igraph-help mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/igraph-help

-- 
Linton C. Freeman
Institute for Mathematical Behavioral Sciences and
Department of Sociology
School of Social Sciences  SSPA 2143
University of California
Irvine, CA 92697-5100

Office:             (949) 824-6698
Home (CA):      (949) 494-6139
          (FL):      (941) 778-1074
Secretary:        (949) 824-3663
FAX:                (949) 824-3733