def simple_subjects_and_objects(self, verb):
"""
Extract all simple subjects and objects for a given verb.
Uses Textacy get_objects_of_verb and get_subjects_of_verb methods
param: verb A Spacy Token
return: A list of verb subjects and objects (Spacy Tokens or Spans)
"""
verb_objects = get_objects_of_verb(verb)
verb_subjects = get_subjects_of_verb(verb)
verb_objects.extend(verb_subjects)
return verb_objects
def subject_verb_object_triples(doc):
"""
Extract an ordered sequence of subject-verb-object (SVO) triples from a
spacy-parsed doc. Note that this only works for SVO languages.
Args:
doc (``spacy.Doc`` or ``spacy.Span``): either a spacy document
or a sentence thereof
Yields:
(``spacy.Span``, ``spacy.Span``, ``spacy.Span``): the next 3-tuple from ``doc``
representing a (subject, verb, object) triple, in order of apperance
# TODO: What to do about questions, where it may be VSO instead of SVO?
# TODO: What about non-adjacent verb negations?
# TODO: What about object (noun) negations?
"""
try:
sents = doc.sents
except AttributeError:
sents = [doc]
for sent in sents:
start_i = sent[0].i
verbs = get_main_verbs_of_sent(sent)
for verb in verbs:
subjs = get_subjects_of_verb(verb)
if not subjs:
continue
objs = get_objects_of_verb(verb)
if not objs:
continue
# add adjacent auxiliaries to verbs, for context
# and add compounds to compound nouns
verb_span = get_span_for_verb_auxiliaries(verb)
verb = sent[verb_span[0] - start_i: verb_span[1] - start_i + 1]
for subj in subjs:
subj = sent[get_span_for_compound_noun(subj)[0] - start_i: subj.i - start_i + 1]
for obj in objs:
if obj.pos == NOUN:
span = get_span_for_compound_noun(obj)
elif obj.pos == VERB:
span = get_span_for_verb_auxiliaries(obj)
else:
span = (obj.i, obj.i)
obj = sent[span[0] - start_i: span[1] - start_i + 1]
yield (subj, verb, obj)
def verb_relevance(self, verb, article):
"""
Checks a verb for relevance by:
1. Comparing to structure term lemmas
2. Comparing to person term lemmas
3. Looking for special cases such as 'leave homeless'
"""
# case for eviction first because we have 'forced eviction' case which would be picked by the 'elif' below
if 'eviction' in [obj.lemma_ for obj in get_objects_of_verb(verb)]:
verb_objects = get_objects_of_verb(verb)
for verb_object in verb_objects:
if verb_object.text == 'eviction' or verb_object.text == 'evictions':
return self.structure_unit_lemmas + self.person_unit_lemmas, Fact(
verb, article[verb.i:verb_object.i + 1],
verb.lemma_ + " " + "eviction", "term")
elif verb.lemma_ in self.joint_term_lemmas:
return self.structure_unit_lemmas + self.person_unit_lemmas, Fact(
verb, verb, verb.lemma_, "term")
elif verb.lemma_ in self.structure_term_lemmas:
return self.structure_unit_lemmas, Fact(verb, verb, verb.lemma_,
"term")
elif verb.lemma_ in self.person_term_lemmas:
return self.person_unit_lemmas, Fact(verb, verb, verb.lemma_,
"term")
elif verb.lemma_ in ('leave', 'render', 'become'):
children = verb.children
obj_predicate = None
for child in children:
if child.dep_ in ('oprd', 'dobj', 'acomp'):
obj_predicate = child
if obj_predicate:
if obj_predicate.lemma_ in self.structure_term_lemmas:
return self.structure_unit_lemmas, Fact(
verb, article[verb.i:obj_predicate.i + 1],
'leave ' + obj_predicate.lemma_, "term")
elif obj_predicate.lemma_ in self.person_term_lemmas:
return self.person_unit_lemmas, Fact(
verb, article[verb.i:obj_predicate.i + 1],
'leave ' + obj_predicate.lemma_, "term")
elif verb.lemma_ == 'affect' and self.article_relevance(article):
return self.structure_unit_lemmas + self.person_unit_lemmas, Fact(
verb, verb, verb.lemma_, "term")
elif verb.lemma_ in ('fear', 'assume'):
verb_objects = get_objects_of_verb(verb)
if verb_objects:
verb_object = verb_objects[0]
if verb_object.lemma_ in self.person_term_lemmas:
return self.person_unit_lemmas, Fact(
verb, article[verb.i:verb_object.i + 1],
verb.lemma_ + " " + verb_object.text, "term")
elif verb_object.lemma_ in self.structure_term_lemmas:
return self.structure_unit_lemmas, Fact(
verb, article[verb.i:verb_object.i + 1],
verb.lemma_ + " " + verb_object.text, "term")
elif verb.lemma_ == 'claim':
verb_objects = get_objects_of_verb(verb)
for verb_object in verb_objects:
if verb_object.text == 'lives':
return self.person_unit_lemmas, Fact(
verb, article[verb.i:verb_object.i + 1],
verb.lemma_ + " " + "lives", "term")
return None, None
def subject_verb_object_triples(doc):
"""
Extract an ordered sequence of subject-verb-object (SVO) triples from a
spacy-parsed doc. Note that this only works for SVO languages.
Args:
doc (``textacy.Doc`` or ``spacy.Doc`` or ``spacy.Span``)
Yields:
Tuple[``spacy.Span``, ``spacy.Span``, ``spacy.Span``]: the next 3-tuple
of spans from ``doc`` representing a (subject, verb, object) triple,
in order of appearance
"""
# TODO: Rewrite rules based on http://www.anthology.aclweb.org/W/W12/W12-0702.pdf
# TODO: Think about relative clauses (that-of) e.g., products that include
# TODO: What to do about questions, where it may be VSO instead of SVO?
# TODO: What about non-adjacent verb negations?
# TODO: What about object (noun) negations?
if isinstance(doc, SpacySpan):
sents = [doc]
else: # textacy.Doc or spacy.Doc
sents = doc.sents
for sent in sents:
start_i = sent[0].i
verbs_init = get_main_verbs_of_sent(sent)
list_candidates = []
verb_tmp_token = None
for verb_init in verbs_init:
if (verb_init['token'] != verb_tmp_token):
verb_tmp_token = verb_init['token']
subjs = get_subjects_of_verb(verb_init['token'], sent)
if not subjs:
continue
verbs = get_span_for_verb_auxiliaries(verb_init['token'],
start_i, sent)
list_candidates.append((subjs, verbs))
for subjs, verbs in list_candidates:
for verb in verbs:
objs = get_objects_of_verb(verb['token'])
if not objs:
continue
# add adjacent auxiliaries to verbs, for context
# and add compounds to compound nouns
for subj in subjs:
subj_type = subj.ent_type_
subj = sent[get_span_for_compound_noun(subj)[0] -
start_i:subj.i - start_i + 1]
for obj in objs:
obj_type = obj.ent_type_
if obj.pos != VERB: #obj.pos == NOUN or obj.pos == PROPN:
span = get_span_for_compound_noun(obj)
elif obj.pos == VERB:
#span = get_span_for_verb_auxiliaries(obj, start_i, sent)
span = (obj.i, obj.i)
else:
span = (obj.i, obj.i)
obj = sent[span[0] - start_i:span[1] - start_i + 1]
score = subj.similarity(obj) + obj.similarity(subj)
yield (subj, verb, obj, score, subj_type, obj_type)