def decode_all(tag_sequences):
# decode from all sequences, each sequence with a unique id
ents = []
for sent_id, tags in enumerate(tag_sequences):
for ent in decode_from_bioes(tags):
ent['sent_id'] = sent_id
ents += [ent]
return ents
def build_ents(self):
""" Build the list of entities by iterating over all tokens. Return all entities as a list.
Note that unlike other attributes, since NER requires raw text, the actual tagging are always
performed at and attached to the `Token`s, instead of `Word`s.
"""
self.ents = []
tags = [w.ner for w in self.tokens]
decoded = decode_from_bioes(tags)
for e in decoded:
ent_tokens = self.tokens[e['start']:e['end']+1]
self.ents.append(Span(tokens=ent_tokens, type=e['type'], doc=self.doc, sent=self))
return self.ents