PageRank

PageRank is a way to measure the "authority" of vertices in a graph [1]. A simplistic alternative to PageRank is to measure the in-degrees at each vertex. The algorithm is describe as follows:

1. Initialize the vertices with a starting PageRank of 1/N, where N is the number of vertices in the graph.

2. Loop:

   • For each vertex, transmit a PageRank of 1/M along each outbound edge, where M is the outdegree of the vertex.
   • At each vertex receiving incoming PageRanks from adjacent vertices, sum these up and make that the new PageRank for the vertex.
   • If PageRanks haven't significantly changed across the graph since the previous iteration, then exit.

   Note: example taken from [2]

Invoking PageRank in GraphX

Object-Oriented and Object-Based are the two ways to invoke PageRank on GraphX [1].

References

1. Malak, M., & East, R. (2016). Spark GraphX in Action. Manning.

2. Cormen, T., Leiserson, C., Rivest, R., & Stein, C. (2009). Introduction to Algorithms. The MIT Press.