Parsing NER outputs into a table

[Gadro]

Hello everyone!

I'm currently working through this tutorial (Timestamp: 4:38) and I'm having a difficult time understanding the portion of the workflows in which they parse the output of the NER model into a table. Specifically, what values are they using "Join" and "Unique" on? How do you imagine they collapse it back into the table shown in the tutorial?

Thanks in advance!

[BrianR · Accepted solution]

Hi @Gadro, came across your question, apologies a little delayed. I took a look at that video you linked...I also had to test out some different combinations. Here's how I configured the join - in short, we also want to join on JSON_Name2 to capture the situations where there are multiple possibilities both in the label and the entity...then we use the unique to filter out the results based on the unique fields RecordID, Entity, and Right_JSON_ValueString.