Combine ner.teach and ner.correct?

[yw2903]

Is there a recipe that could combine the function of ner.teach and ner.correct? Correcting the predicted entity while looping through and updating the model at the same time?

[ines (Ines Montani)]

Hi! There's no built-in recipe for this at the moment and we haven't run extensive experiments with an approach like this, but it's definitlely something you can implement. (The new dep.correct actually does something very similar.)

I've posted more details and ideas on this thread a while back:

Combining ner.teach with patterns file and manual correction of spans

Hi! What you describe is definitely possible – it might just take some experimentation to find the best configuration that produces the most useful results. For example, whether to upate the model with complete or incomplete annotations, how to incorporate patterns (if at all) etc. It probably makes sense to start off with a recipe like ner.correct that adds entities to the outgoing texts and lets you edit them manually, and then add the update functionality to it that updates the model in the …