Graphs exhibit clustering behavior, and identification of communities is an important task in social networks. A node's clustering coefficient is the number of triadic closures (closed triples) in the node's neighborhood. This is an expression of transitivity. Nodes with higher transitivity exhibit higher subdensity, and if completely closed, form cliques that can be identified as communities. In this recipe, we will look at clustering and community detection in social networks.
You will again need NetworkX and, for the first time in this chapter, the python-louvain library.
These steps will guide you through the detection of communities within social networks:
- Let's actually get into some clustering. The
python-louvainlibrary uses NetworkX to perform community detection with thelouvainmethod. Here is a simple example of cluster partitioning on a small, built-in social network:
G = nx.karate_club_graph() ...