gold.biluo_tags_from_offsets should throw E103 if two entity labels are overlapping

[RyanZHe]

Feature description

spacy.gold.biluo_tags_from_offsets should throw Error.E103 when the collection of entites passed in have overlapping tokens.

For example,

from spacy.gold import biluo_tags_from_offsets

doc = nlp(u"I like California Pizza Kitchen.")
entities = [(7, 17, "LOC"), (7, 31, "BRAND") ]
tags = biluo_tags_from_offsets(doc, entities) # should throw Error.E103

Right now tags is being assigned as ["O", "O", "U-LOC", "I-BRAND", "L-BRAND"], which violates the biluo tagging schema.

[pratapaprasanna]

I agree that an error has to be raised while forming tags only and not while giving input to gold corpus . It now throws error while feeding in data in goldcorpus. which I think a bad way.

Thanks

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.