This package provides interfaces for personalized PageRank on Pregel/GraphX.
// Gets test graph data from `https://snap.stanford.edu/data/ca-CondMat.html`
$ wget https://snap.stanford.edu/data/ca-CondMat.txt.gz
// Launches a Spark v2.4.5 shell with this package
// Note that you can set true to `spark.graphx.pageRank.loggingEnabled` for checking the progress
$ git clone https://github.com/maropu/spark-graphx-pregel-personalized-pagerank.git
$ $SPARK_HOME/bin/spark-shell --conf spark.jars ./spark-graphx-pregel-personalized-pagerank/spark-graphx-pregel-personalized-pagerank_2.11_2.4.5-0.1.0-SNAPSHOT-with-dependencies.jar --conf spark.graphx.pageRank.loggingEnabled=true -v
...
scala> import org.apache.spark.graphx._
// Loads the downloaded graph data
scala> val graph = GraphLoader.edgeListFile(sc, "ca-CondMat.txt.gz")
// Computes personalized PageRank
scala> lib.PageRankEx.runParallelPersonalizedPageRankUntilConvergence(graph, sources = Seq(130, 81626), tol = 0.01)
Pregel finished iteration 0 (# of active messages: 1090)
Aggregated metrics: value:(1.8499999999999999,1.8499999999999996) delta:(1.8499999999999999,1.8499999999999996)
Pregel finished iteration 1 (# of active messages: 424)
Aggregated metrics: value:(2.5724999999999874,2.572500000000004) delta:(0.7832142857142885,0.7697222222222233)
Pregel finished iteration 2 (# of active messages: 64)
Aggregated metrics: value:(2.952715671543223,2.6335382923099413) delta:(0.43964207846964315,0.5921036916707042)
Pregel finished iteration 3 (# of active messages: 22)
Aggregated metrics: value:(3.145608510503435,2.633552646467126) delta:(0.2908540302238811,0.5914781586882217)
Pregel finished iteration 4 (# of active messages: 0)
Aggregated metrics: value:(3.1912734581542,2.633554118907456) delta:(0.2230614608527432,0.5914717973238139)
For API details, see PageRankEx.scala.
If you hit some bugs and requests, please leave some comments on Issues or Twitter(@maropu).