Junto on Hadoop

[p.li...@gmail.com]

Hi All! Quick question about running Junto on large graphs:

What is the recommended way to create a "node factored" representation for
graphs that don't fit into main memory? It looks like EdgeFactored2NodeFactored
needs to read the entire graph in at once.

Also, can I do something like this: hdfs_input_pattern=input/* ? 
Where /input would contain the different parts of graph (in node factored representation).

Thanks!