Accepting and rejecting in ner.teach recipe


I have a long text (like a paragraph),and would like to use ner.teach to assist me.

If there are two spans of tokens that has the same entity label, but the model only picks up one span, should I reject?


If you're collecting binary annotations, you're only ever seeing one suggestion at a time. The suggestions are based on the score of that particular entity. So you're only giving feedback about the highlighted span, and you'd never be seeing more than one. So the span you see is correct, you should accept it. If the span is incorrect (completely or partially), you should reject it (also see this docs section on partial suggestions).