Converting generated heterogeneities data for PSL training

[omarrana]

Generator generates heterogeneities files from seed.aml

Lets say we got doc1.aml and doc2.aml
Now we want to use this for our PSL model as training data.

We first need to run our PSL code which converts AML data into PSL data format.
But how should we compute similar.txt for training?

Should we run PSL rules for similar.txt for training as well? ( training and test would contradict as both uses same rule and accuracy level won't increase)

Or this should be done manually?