DataModel population efficiency

[kordjamshidi]

I am re-running SRL experiments and I am not sure why the population so extremely slow. I wonder, who is available to re-check the data model population code and see where the redundant searches are happening?

[kordjamshidi]

According to @bhargav this line causes the issue, and that is true for my experience too,

joinNodes.foreach(_.addFromChild(this, instance, train, populateEdge))

but needs investigation, since we even do not call it when runnig SRL.

[bhargav]

Also as an update to the cause of this, it started happening in SRL because the usage of JoinNodes in the DataModel was added in line 42 of SRLMultiGraphDataModel.scala

Older version of SRL did not have this usage of joinNodes thus no population was happening before.