Community, affiliation, and strengths

Let's now look at the network connections and others. These algorithms are applied widely for fraud detection and security applications. Triangular spamming is a well-known technique that can be detected using the triangle count and community algorithms. Another interesting application of the triangle count is to estimate and rank communities. The age of a community is related to the density of the triangles; new communities will have fewer triangles, and as the communities mature, triangles start to form. Another interesting application is the concept of a heavy hitter in a community, defined as any vertex that has more than sqrt(n) degrees. Finding heavy hitter triangles would be like finding influential people in a community. Connected communities and strongly connected communities expose the structure in an underlying graph, akin to the Panama papers. And all these are APIs in GraphX. No wonder GraphX is part of the processing stack for Linkedin...