Jaccard Example for Apache Fluo

Incrementally computes the Jaccard for a bipartite graph. All of the code refers to two node types, a P-node and a S-node. This example computes the Jaccard for all P-nodes. It assume that edges are only between P-nodes and S-nodes.

Commands

The following command will initialize Fluo with the Jaccard example application.

bin/init.sh

The following command will load edges from files after initialization.

bin/load.sh <remainder> <modulus> <file{ file}>

The modulus and remainder are there to support loading a subset of the edges, which is useful for testing this example. During load each edge is hashed and if hash(edge) % modulus == remainer is true then the edge is loaded.

To see how this is useful for testing, consider the following commands.

#loads all edges in files edges1.txt and edges2.text
bin/load.sh 1 edges1.txt edges2.txt

All these two commands load the edges in very differents temporal orders, the end result when the two commands are finished should be the same.

#loads 1/3 of the edges in files edges1.txt and edges2.txt and wait between
bin/load.sh 3 edges1.txt edges2.txt

After loading data, the following command can be used to query a pnode.

$FLUO_HOME/bin/fluo exec jaccard fj.cmd.Query <pnode>