/netclust

Efficient cluster detection tool for large graphs (networks)

Primary LanguageCGNU General Public License v3.0GPL-3.0

netclust/netindex

are fast and memory-efficient C programs developed to detect clusters of nodes in very large graphs (networks) with billions of edges using a scalable nearest-neighbor approach with (nearly) linear time and space complexity ~O(E), E number of edges (Kuzniar et al., 2010). While netindex pre-processes the input graph into an efficient binary format, netclust performs the actual cluster detection in the graph by taking a single user-defined cutoff value (similarity or distance).

Compile and run

cd src
make        # default make 'all' binaries or separately 'netindex' | 'netclust'
make clean  # remove binaries
netindex ../examples/graph.tsv          # index the input (weighted) graph
netclust ../examples/graph.tsv O1 S 500 # detect clusters in the graph