def get_verbs(story, span):
verbs = []
for token in span:
if NLPUtility.is_verb(token) and str.lower(token.text) != 'can':
verbs.append(token)
return MinerUtility.get_span(story, verbs)
def get_verbs(story, span): verbs = [] for token in span: if NLPUtility.is_verb(token): verbs.append(token) return MinerUtility.get_span(story, verbs)
def remove_verbs(self, matrix, stories):
verbs = []
cases = matrix.index.values.tolist()
for case in cases:
pos = []
for story in stories:
for token in story.data:
if NLPUtility.case(token) == case:
pos.append(token)
if len(set(pos)) == 1 and NLPUtility.is_verb(pos[0]):
verbs.append(case)
for verb in verbs:
if matrix.loc[verb, 'sum'] > 0:
verbs.remove(verb)
return matrix[(-matrix.index.isin(verbs))]
def remove_verbs(self, matrix, stories): verbs = [] cases = matrix.index.values.tolist() for case in cases: pos = [] for story in stories: for token in story.data: if NLPUtility.case(token) == case: pos.append(token) if len(set(pos)) == 1 and NLPUtility.is_verb(pos[0]): verbs.append(case) for verb in verbs: if matrix.loc[verb, 'sum'] > 0: verbs.remove(verb) return matrix[(-matrix.index.isin(verbs))]
def get_mobj_and_mv(self, story, part='means'):
has_subj = False
simple = False
found_verb = False
found_obj = False
found_mv_phrase = False
subject = []
main_verb = []
main_object = []
mv_phrase = []
# Simple case if the subj and dobj are linked by a verb
for token in eval('story.' + str(part) + '.text'):
if NLPUtility.is_subject(token):
has_subj = True
subject = token
#BC if NLPUtility.is_verb(token.head):
if NLPUtility.is_verb(
token.head) and str.lower(token.head.text) != 'can':
found_verb = True
main_verb = token.head
break
if type(subject) is list:
subject = eval('story.' + str(part) + '.text')[0]
for token in eval('story.' + str(part) + '.text'):
if NLPUtility.is_dobj(token):
found_obj = True
if token.pos_ == "PRON": # If it is a pronoun, look for a preposition with a pobj
f = False
for child in token.head.children:
if child.dep_ == "prep" and child.right_edge.dep_ == "pobj" and not f:
token = child.right_edge
mv_phrase = [main_verb, child]
f = True
found_mv_phrase = True
elif token.pos_ == "ADJ" or token.pos_ == "ADV": # Set to right edge if there is an adj/adv as dobj, and possibly make a verb phrase
original_token = token
f = False
for child in token.children:
if child.dep_ == "prep" and not f:
for grandchild in child.children:
if grandchild.dep_ == "pobj":
mv_phrase = [main_verb, token, child]
token = grandchild
f = True
found_mv_phrase = True
if token.head == main_verb:
simple = True
main_object = token
break
# If the root of the sentence is a verb
if not simple:
for token in eval('story.' + str(part) + '.text'):
if token.dep_ == 'ROOT' and NLPUtility.is_verb(token):
found_verb = True
main_verb = token
break
# If no main verb could be found it is the second word (directly after 'I')
# Possibly a NLP error...
if not found_verb:
#BC main_verb = eval('story.' + str(part) + '.text')[1]
if str(part) == 'means' or str.lower(
eval('story.' + str(part) + '.text')[1].text) == 'can':
main_verb = eval('story.' + str(part) + '.text')[2]
else:
main_verb = eval('story.' + str(part) + '.text')[1]
# If the sentence contains no dobj it must be another obj
if not found_obj:
for token in eval('story.' + str(part) + '.text'):
if token.dep_[1:] == 'obj':
found_obj = True
main_object = token
break
# If none is found it points to the unknown 'system part'
# + get phrases for main_object and main_verb
if not found_obj and part == 'means':
main_object = story.system.main
if part == 'means':
story.means.main_verb.main = main_verb
story.means.main_object.main = main_object
if found_mv_phrase:
story.means.main_verb.phrase = MinerUtility.get_span(
story, mv_phrase, 'means.text')
story.means.main_verb.type = "II"
else:
story.ends.subject.main = subject
story.ends.main_verb.main = main_verb
story.ends.main_object.main = main_object
if found_mv_phrase:
story.ends.main_verb.phrase = MinerUtility.get_span(
story, mv_phrase, 'ends.text')
story.ends.main_verb.type = "II"
if type(main_object) is list or main_object == story.system.main:
story = eval('self.get_' + str(part) + '_phrases(story, ' +
str(found_mv_phrase) + ', False)')
else:
story = eval('self.get_' + str(part) + '_phrases(story, ' +
str(found_mv_phrase) + ')')
return story
def get_mobj_and_mv(self, story, part='means'):
has_subj = False
simple = False
found_verb = False
found_obj = False
found_mv_phrase = False
subject = []
main_verb = []
main_object = []
mv_phrase = []
# Simple case if the subj and dobj are linked by a verb
for token in eval('story.' + str(part) + '.text'):
if NLPUtility.is_subject(token):
has_subj = True
subject = token
if NLPUtility.is_verb(token.head):
found_verb = True
main_verb = token.head
break
if type(subject) is list:
subject = eval('story.' + str(part) + '.text')[0]
for token in eval('story.' + str(part) + '.text'):
if NLPUtility.is_dobj(token):
found_obj = True
if token.pos_ == "PRON": # If it is a pronoun, look for a preposition with a pobj
f = False
for child in token.head.children:
if child.dep_ == "prep" and child.right_edge.dep_ == "pobj" and not f:
token = child.right_edge
mv_phrase = [main_verb, child]
f = True
found_mv_phrase = True
elif token.pos_ == "ADJ" or token.pos_ == "ADV": # Set to right edge if there is an adj/adv as dobj, and possibly make a verb phrase
original_token = token
f = False
for child in token.children:
if child.dep_ == "prep" and not f:
for grandchild in child.children:
if grandchild.dep_ == "pobj":
mv_phrase = [main_verb, token, child]
token = grandchild
f = True
found_mv_phrase = True
if token.head == main_verb:
simple = True
main_object = token
break
# If the root of the sentence is a verb
if not simple:
for token in eval('story.' + str(part) + '.text'):
if token.dep_ == 'ROOT' and NLPUtility.is_verb(token):
found_verb = True
main_verb = token
break
# If no main verb could be found it is the second word (directly after 'I')
# Possibly a NLP error...
if not found_verb:
main_verb = eval('story.' + str(part) + '.text')[1]
# If the sentence contains no dobj it must be another obj
if not found_obj:
for token in eval('story.' + str(part) + '.text'):
if token.dep_[1:] == 'obj':
found_obj = True
main_object = token
break
# If none is found it points to the unknown 'system part'
# + get phrases for main_object and main_verb
if not found_obj and part == 'means':
main_object = story.system.main
if part == 'means':
story.means.main_verb.main = main_verb
story.means.main_object.main = main_object
if found_mv_phrase:
story.means.main_verb.phrase = MinerUtility.get_span(story, mv_phrase, 'means.text')
story.means.main_verb.type = "II"
else:
story.ends.subject.main = subject
story.ends.main_verb.main = main_verb
story.ends.main_object.main = main_object
if found_mv_phrase:
story.ends.main_verb.phrase = MinerUtility.get_span(story, mv_phrase, 'ends.text')
story.ends.main_verb.type = "II"
if main_object == story.system.main:
story = eval('self.get_' + str(part) + '_phrases(story, ' + str(found_mv_phrase) + ', False)')
else:
story = eval('self.get_' + str(part) + '_phrases(story, ' + str(found_mv_phrase) + ')')
return story
