Unique responses about answercoalesce HOT 4 CLOSED

the Famotidine notebook under documentation/experiments , using covidkop now produces this reproducibly.

from answercoalesce.

I now have a test that reproduces this behavior. Another odd effect is that the number of multiples varies from run to run.

This seems to be related to answers in which there are multiple properties connecting nodes.

If A is connected to B through 2 different predicates, then what does a coalescent opportunity look like? Do I need to find another answer that has the same two predicates between A and B' ? Or is one of the two good enough?

I think that the easier / POC version is to make things match 100% but we're going to need to move away from that on the sooner side.

from answercoalesce.

Also, looks like sometimes the hash is a predicate for edges, and sometimes it's a kg_id ?

from answercoalesce.

Fixed. The problem was in how we handle node that have multiple edges between them. We were arbitrarily selecting one. There was another problem also in looping both inside and outside of a function (!?).

from answercoalesce.