NER training data creation

, ,
1

I know the question I am about to ask might be trivial or idiotic but I am a bit confused about it.
What happens when I add a text to the training data but with no entities as input. Will that entry be ignored during training or does that add any value regarding the model understand which are non entities? Or will it throw a error?

2

Hello @arda,

Thank you for your question.
Adding examples without any labeled entity to the training data is indeed useful for the model to learn correct entities, see the infobox "When should I press accept, reject or ignore?" in our prodigy docs. Therefore, no error is thrown for examples with no entities, and they aren't ignored either.

1 Like
3

Thanks @Jette16 for the prompt reply. Appreciate it.

1 Like