[igraph] Community detection algorithm for (almost) full weighted graph

[capitano . nemo]

Hello,

I have a series of graphs created from dense matrices such as

r1 <- c(NA, 0.2, 0.3, 0)
r2 <- c(NA, NA, 0.4, 0.6)
r3 <- c(NA, NA, NA, 0.1)
r4 <- rep(NA, 4)
m <- rbind(r1,r2,r3,r4)
rownames(m) <- seq(1:4)
colnames(m) <- seq(1:4)
m
# 1   2   3   4
# 1 NA 0.2 0.3 0.0
# 2 NA  NA 0.4 0.6
# 3 NA  NA  NA 0.1
# 4 NA  NA  NA  NA

For this matrix I create a graph with

g <- graph.adjacency(m, mode="undirected", weighted=TRUE, diag=FALSE)

My real graphs are not extremely large (the largest has 11249 nodes) but 
they are pretty dense: as in the example above nodes have relations with 
almost every other nodes, but without any multiedge. Yet each edge is 
weighted. The "weight" attribute of the edge is the closeness of the 
pair of nodes. That is, for example an edge with a "weight" of 0.5 
indicates that the endpoints are closer than a endpoints with an edge of 
0.3. I am interested in drawing communities that reflect the "closeness" 
among nodes.

Then I want to apply a community detection algorithm which will evaluate 
graph structure mainly on edges' weights.

I tested

fastgreedy.community(g)

on my small graph and it works pretty well in dividing nodes according 
to edges' weight.

But I wonder if I should use other algos that better capture the 
weighted structure of my graph.

Thanks