def rule0_func(self, sent_data):
sent_id, sent_tokens, sent_text, noun_chunks, named_entities, ent1, ent2 = sent_data
ent1_token = ent_text_to_ent(ent1, sent_data)
ent2_token = ent_text_to_ent(ent2, sent_data)
if ent1_token.label_ in self.persons_ents and ent2_token.label_ in self.places_ents:
return True
return False
def remove_non_relevant_samples(data):
filtered_data = []
for sent_data, y in data:
sent_id, sent_tokens, sent_text, noun_chunks, named_entities, ent1, ent2 = sent_data
ent1_token = ent_text_to_ent(ent1, sent_data)
ent2_token = ent_text_to_ent(ent2, sent_data)
if ent1_token.label_ in person_ents and ent2_token.label_ in places_ents:
filtered_data.append((sent_data, y))
return filtered_data
def rule1_func(self, sent_data):
sent_id, sent_tokens, sent_text, noun_chunks, named_entities, ent1, ent2 = sent_data
ent1_token = ent_text_to_ent(ent1, sent_data)
ent2_token = ent_text_to_ent(ent2, sent_data)
if ent1_token.label_ in self.places_ents and ent2_token.label_ in self.persons_ents:
if "'s" in words_between_ents(sent_data) and ents_between(
ent1_token, ent2_token, sent_data) == 0:
return True
return False
def words_between_ents(sent_data):
sent_id, sent_tokens, sent_text, noun_chunks, named_entities, ent1, ent2 = sent_data
words_in_sent = [x.text for x in sent_tokens]
ent1_token = ent_text_to_ent(ent1, sent_data)
ent2_token = ent_text_to_ent(ent2, sent_data)
words_between = []
if ent1_token.start > ent2_token.start:
return words_between
for i in range(ent1_token.end, ent2_token.start):
words_between.append(words_in_sent[i])
return words_between