[igraph] Another question about overlapping communities

[Frederik Lyngsaa Lang]

Hi all,


I have been looking at finding a function for overlapping community detection in igraph. I started out using the clique percolation method (CPM) as suggested on the igraph wiki but since it ran too slowly for large networks I switched to Cfinder. Now, however, I experience the issue that CPM works better for affiliation networks with lots of cliques but I have a directed network. Has anyone altered some of the already existing algorithms in igraph for community detection based on modularity to detect overlapping communities such as CONGA or similar? And what about the label propagation algorithm implemented by Peter McMahan here. Is it any different from the one already implemented in igraph?

I would be very interested to know about possibilities for looking at overlaps in sparse and directed networks.


Kind regards,


Frederik