def __init__(self, datasets_path, corpus_name, parse_type, lang='english'):
self.datasets_path = datasets_path
self.corpus_name = corpus_name
self.corpus_path = path.join(datasets_path, corpus_name)
self.docs_path = path.join(self.corpus_path, "docs")
self.topics_file = path.join(self.corpus_path, "topics.xml")
self.models_path = path.join(self.corpus_path, "models")
self.smodels_path = path.join(self.corpus_path, "smodels")
self.jar_path = path.join(PROJECT_PATH, "summarizer", "jars")
os.environ['CLASSPATH'] = self.jar_path
self.cleaned_path = path.join(datasets_path, "processed")
if parse_type == 'parse':
if lang == 'english':
self.parser = stanford.StanfordParser(
model_path="%s/englishPCFG.ser.gz" % (self.jar_path))
if lang == 'german':
self.parser = stanford.StanfordParser(
model_path="%s/germanPCFG.ser.gz" % (self.jar_path))
# self.cleaned_path = path.join(datasets_path, "processed.parse")
if parse_type == 'props': # TODO
if lang == 'english':
self.props_parser = ClausIE.get_instance()
if lang == 'german':
self.parser = stanford.StanfordParser(
model_path="%s/germanPCFG.ser.gz" % (self.jar_path))
def __init__(self):
try:
self.cl = ClausIE.get_instance(
jar_filename=
'/data/defacto/github/fever/clausie/clausie/clausie.jar')
except Exception as error:
raise error
def main():
sents = process_data_from_input_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
# print(r)
# bob = add_person('Bob')
# print(bob.likes,bob.has)
# joe = add_person('Joe')
# print(joe.likes,joe.has)
# mary = add_person('Mary')
# print(mary.likes,mary.has)
# print(persons)
# print(pets)
# print(trips)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
answer_question(question)
def enumerateSPtuples_Translate(self,
keyword=None,
maxTweet=99999,
reader=dataReader()):
reader = reader
c = reader.cursor_tweet(keyword)
tweetContent = {}
SPtuples = []
translator = googleTranslatror()
cl = ClausIE.get_instance(jar_filename='./clausie/clausie.jar')
for tweetIndex, item in enumerate(c):
print tweetIndex + 1, ':'
content = item['Content'].replace(u'\u200b', '')
content = content.replace(u'\xa0', '')
content = content.replace(u'\u5168\u6587', '')
content = content.encode('utf-8')
content = re.sub(r'#.*?#|@.*?\s|\[.*?\]|\s|【|】|全文|的秒拍视频|(|)|“|”',
'', content)
print content
content = translator.translate_zhCN2en(content)
print content
if tweetIndex + 1 == maxTweet:
break
return tweetContent, SPtuples
def process_data_from_input(file_path):
sents = get_data_from_file(file_path)
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
print('triples processed')
def process_data_from_input_file():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
# print(t)
process_relation_triplet(t)
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
print(r)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
q_trip = cl.extract_triples([preprocess_question(question)])[0]
# (WHO, has, PET)
# here's one just for dogs
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a {} named {}.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
print(answer.format(person.name, 'dog', pet.name))
def process_data_from_input_file(path='assignment_01.data'):
sents = get_data_from_file(path)
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
def process_data_from_input_file(path='assignment_01.data'):
sents = get_data_from_file(path)
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
try:
process_relation_triplet(triples)
except Exception as e:
print(
"""There was an error when processing following triplet in the data file: {}\nSENT: {}\nERROR: {}\n\n"""
.format(t, sents[int(t.index)], e))
def load_data():
input_data = process_data_from_input_file() # get input data
cl = ClausIE.get_instance() # setup the NLP thing
triples = cl.extract_triples(
input_data) # turn the input data into some triples
for t in triples:
try:
r = process_relation_triplet(t)
except:
pass
return cl, triples
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
# print(r)
print(trips)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
answer_question(question)
def answer_questions(string):
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
q_trip = cl.extract_triples([preprocess_question(string)])[0]
answers = []
print(q_trip)
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a dog.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
answer = (answer.format(person.name))
answers.append(answer)
if q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a cat.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
answer = (answer.format(person.name))
answers.append(answer)
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
print(r)
#print(persons)
#print(pets)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
question = question.lower()
q_trip = cl.extract_triples([preprocess_question(question)])[0]
# person_name = q_trip.subject.name
# print(person_name)
# print(q_trip.subject.label=='PROPN')
# (WHO, has, PET)
# here's one just for dogs
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a {} named {}.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
print(answer.format(person.name, 'dog', pet.name))
# here's one for cats
elif q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a {} named {}.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
print (answer.format(person.name, 'cat', pet.name))
# for this Does John have a cat? question
# we need to get the subject in the quesion and than, get_person_pet(person.name)==subject's name
elif q_trip.subject.startswith('does') and q_trip.object == ('cat' or 'dog'):
answer = '{}, {} does not have a {}.'
new_subj_doc = q_trip.subject
people = new_subj_doc[5:].capitalize()
#new_subj_doc = nlp(unicode(q_trip.subject))
#name_doc = new_subj_doc.char_span(0, len(q_trip.subject))
# people = [token.text for token in name_doc if token.ent_type_ == 'PERSON']
# print (people)
# s_people = [token.text for token in subj_doc if token.ent_type_ == 'PERSON']
# for person in persons:
if people in persons:
pet = get_persons_pet(people)
if q_trip.object == pet:
print(answer.format('Yes', people, q_trip.object))
else:
print(answer.format('No', people, q_trip.object))
else:
print (answer.format('No', people, q_trip.object))
def generate_triplet(file_path='./assignment_01.data'):
sents = get_data_from_file(file_path)
cl = ClausIE.get_instance()
triples = [(sent, cl.extract_triples([sent])) for sent in sents]
return triples
def answer_question(question_string):
answers = []
cl = ClausIE.get_instance()
try:
q_trip = cl.extract_triples([preprocess_question(question_string)])[0]
sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
doc = nlp(unicode(sentence))
for t in doc:
if t.pos_ == 'VERB' and t.head == t:
root = t
#print(q_trip)
#print(root.lemma_)
# (WHO, has, PET)
if q_trip.subject.lower(
) == 'who' and q_trip.object == 'dog' or q_trip.object == 'cat':
answer = '{} has a {} named {}.'
for person in persons:
pet = get_persons_pet(person.name)
if q_trip.object == 'dog' and pet and pet.type == 'dog':
answers.append(answer.format(person.name, 'dog', pet.name))
elif q_trip.object == 'cat' and pet and pet.type == 'cat':
answers.append(answer.format(person.name, 'cat', pet.name))
for answer in answers:
print(answer)
# (Who, go, SOMEWHERE)
if q_trip.subject.lower() == 'who' and (root.lemma_ == 'take'
or root.lemma_ == 'fly'
or root.lemma_ == 'go'
or root.lemma_ == 'leave'):
answer = '{} goes to {}.'
for a in [e.text for e in doc.ents if e.label_ == 'GPE']:
for person in persons:
for b in person.travels:
if b != None:
if b.departs_to == a:
answers.append(answer.format(person.name, a))
if answers == []:
print("No one is going to", a)
else:
for answer in answers:
print(answer)
# (Does, PERSON, like, PERSON)
if not has_question_word(question_string):
temp = q_trip.subject.split(' ')
flag = False
for a in temp:
if a in [e.text for e in doc.ents if e.label_ == 'PERSON']:
try:
select_person(a).likes
except AttributeError:
print(q_trip.object,
"is not a person in the database.")
break
if select_person(q_trip.object) in select_person(a).likes:
print("Yes.")
flag = True
if flag == False:
print("No.")
# (When is <person> [going to|flying to|traveling to] <place>)
if q_trip.object.lower().endswith('when'):
flag = False
a = select_person(q_trip.subject)
if a != None:
for b in a.travels:
for c in [e.text for e in doc.ents if e.label_ == 'GPE']:
if b != None:
if b.departs_to == c:
print(b.departs_on)
flag = True
else:
print(q_trip.subject, "is not going to anywhere.")
else:
print(q_trip.subject,
"is not in the database. Try another person name.")
flag = True
try:
c
except NameError:
print("It is not even a place!!")
flag = True
if flag == False:
print(q_trip.subject, "is not going to", c)
# (Who likes <person>?)
if q_trip.subject.lower() == 'who' and root.lemma_ == 'like':
answer = '{} like(s) {}.'
people = []
temp = ''
for person in persons:
for like in person.likes:
if like.name == q_trip.object:
people.append(person)
for a in people:
temp += a.name
temp += ', '
temp = temp[:-2]
if temp == '':
print(answer.format('No one', q_trip.object))
else:
print(answer.format(temp, q_trip.object))
# (Who does <person> like?)
if q_trip.object.lower() == 'who':
answer = '{} likes {}.'
temp = ''
try:
for a in select_person(q_trip.subject).likes:
temp += a.name
temp += ', '
temp = temp[:-2]
print(answer.format(q_trip.subject, temp))
except AttributeError:
print(q_trip.subject, "is not a person in the database.")
except ValueError:
answer = 'The name of {}\'s {} is {}.'
ques = question_string.split(' ')
petType = ques[-1][:-1]
personName = ques[-2][:-2]
try:
for a in select_person(personName).has:
if a.type == petType:
print(answer.format(personName, petType, a.name))
except AttributeError:
print(personName, "is not a person in the database.")
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
# print(r)
#print(persons)
#print(pets)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
question = question.lower()
q_trip = cl.extract_triples([preprocess_question(question)])[0]
q_doc = nlp(unicode(preprocess_question(question)))
# (WHO, has, PET)
# here's one just for dogs
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a {}.'
for person in persons:
pet = person.has
if 'dog' in pet:
print(answer.format(person.name, 'dog'))
elif q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a {}.'
for person in persons:
pet = person.has
if 'cat' in pet:
print(answer.format(person.name, 'cat'))
# here's one for cats
elif q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a {}.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
print(answer.format(person.name, 'cat'))
# for this Does John have a cat? question
# we need to get the subject in the quesion and than, get_person_pet(person.name)==subject's name
elif (q_trip.object.startswith('like')) and (q_trip.subject == 'does'):
answer = '{}, {} {} {}.'
# new_subj_doc = nlp(unicode(q_trip.subject))
# # new_subj_doc = nlp(unicode(q_doc.subject))
p1 = q_trip.predicate.capitalize()
p2 = q_trip.object[5:].capitalize()
host_person = [
str(person.likes) for person in persons if person.name == p1
][0]
if p2 in host_person:
print(answer.format('Yes', p1, 'likes', p2))
else:
print(answer.format('No', p1, 'does not like', p2))
# if p2 in select_person(p1).likes:
# print('yes')
# else:
# print ('no')
# people = [str(e.text) for e in q_doc.ents]
# new_subj_doc_childern = new_subj_doc[0].children
#people = str([t.text.capitalize() for t in new_subj_doc_childern])
# print (people)
elif q_trip.subject.startswith('does') and (q_trip.object == ('cat')
or q_trip.object == ('dog')):
answer = '{}, {} does not have a {}.'
new_subj_doc = nlp(unicode(q_trip.subject))
#people = new_subj_doc[5:].capitalize()
# ents = list(new_subj_doc.ents)
# people = [ents[i].text for i in range(len(ents)) if ents[i].label_=='PERSON']
new_subj_doc_childern = new_subj_doc[0].children
people = str([t.text.capitalize() for t in new_subj_doc_childern])
if people in persons:
pet = get_persons_pet(people)
if q_trip.object == pet:
print(answer.format('Yes', people, q_trip.object))
else:
print(answer.format('No', people, q_trip.object))
else:
print(answer.format('No', people, q_trip.object))
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
#print(r)
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
q_trip = cl.extract_triples([preprocess_question(question)])[0]
q_sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
q_doc = nlp(unicode(q_sentence))
# (WHO, has, PET)
# here's one just for dogs
# Q.1> Who has a <pet type>?
if q_trip.subject.lower() == 'who' and (q_trip.object == 'dog' or q_trip.object == 'cat'):
answer = '{} has a {} named {}.'
print()
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog' and q_trip.object == 'dog':
print(answer.format(person.name, 'dog', pet.name))
elif pet and pet.type == 'cat' and q_trip.object == 'cat':
print(answer.format(person.name,'cat',pet.name))
# Q.2> Does <person> like <person>?
elif 'does' in q_trip.subject.lower() and 'like' in q_sentence:
answer = '{} likes {}'
print()
person_a = [token.text for token in q_doc.ents if token.label_== 'PERSON'][0]
select_a = select_person(person_a)
person_b = q_trip.object
select_b = select_person(person_b)
if select_b in select_a.likes:
print(answer.format(select_a,select_b))
else:
print('No relation given')
# Q.3> Who is [going to|flying to|traveling to] <place>?
elif q_trip.subject.lower() == 'who' and 'GPE' in [e.label_ for e in q_doc.ents]:
answer = '{} is traveling to {}'
print()
check = 0
travel_d = str([t.text for t in q_doc.ents if t.label_=="GPE"][0])
for person in persons:
if travel_d in [trip.departs_to for trip in person.travels]:
print(answer.format(person.name,travel_d))
check = 1
if check==0:
print(answer.format('Nobody',travel_d))
# Q.4> When is <person> [going to|flying to|traveling to] <place>?
elif 'when' in q_sentence.lower() and 'GPE' in [e.label_ for e in q_doc.ents]:
answer = '{} is traveling to {} on {}'
print()
person_t = str([t.text for t in q_doc.ents if (t.label_=="PERSON" or t.label_=="ORG")][0])
dest = str([t.text for t in q_doc.ents if t.label_=="GPE"][0])
check=0
for person in persons:
if person_t == person.name:
if dest in [trip.departs_to for trip in person.travels]:
print(answer.format(person_t, dest, trip.departs_on))
check =1
if check==0:
print(person_t+' is NOT traveling to '+dest+' anytime soon')
# Q.5> Who likes <person>?
elif 'who' in q_sentence.lower() and ('ORG' in [e.label_ for e in q_doc.ents] or 'PERSON' in [e.label_ for e in q_doc.ents]):
person_l = str([t.text for t in q_doc.ents if (t.label_=='PERSON' or t.label_=='ORG')][0])
like_l = []
for person in persons:
if person_l==person.name:
for p in person.likes:
like_l.append(p)
if len(like_l)>0:
print(person_l+' is liked by the following people:')
for p in like_l:
print(p)
elif len(like_l)==1:
print(person_l+' is liked by '+like_l[0])
else:
print('Nobody likes '+person_l)
# Q.6> Who does <person> like?
elif q_sentence.subject.lower() == 'who' and 'does' in q_sentence.lower():
person_lb = str()
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
# print(r)
#print(persons)
#print(pets)
traveldic = ['travel', 'fly', 'drive', 'go', 'visit']
#traveldic = {'travel to': 'take a trip', # verbs here are lemmatized
#'fly to': 'take a trip',
#'drive to': 'take a trip',
#'go to': 'take a trip',
#}
question = ' '
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
question2 = question
question = question.lower()
q_trip = cl.extract_triples([preprocess_question(question)])[0]
# q_doc = nlp(unicode((question)))
doc = nlp(unicode(question2))
# (WHO, has, PET)
# here's one just for dogs
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a {}.'
for person in persons:
pet = person.has
if 'dog' in pet:
print(answer.format(person.name, 'dog'))
elif q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a {}.'
for person in persons:
pet = person.has
if 'cat' in pet:
print(answer.format(person.name, 'cat'))
# here's one for cats
elif q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a {}.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
print (answer.format(person.name, 'cat'))
# Does someone like someone else?
elif (q_trip.object.startswith('like')) and (q_trip.subject == 'does'):
answer = '{}, {} {} {}.'
doc = nlp(unicode(question2))
personnames = [entity.text for entity in doc.ents if entity.label_=='PERSON']
p1 = str(personnames[0])
p2 = str(personnames[1])
host_person = [str(person.likes) for person in persons if person.name == p1]
if p2 in host_person:
print (answer.format('Yes',p1,'likes', p2))
else:
print (answer.format('No', p1,'does not like', p2))
# Who likes <person>?
elif (q_trip.subject == 'who') and (q_trip.predicate == 'likes'):
answer = '{} {} {}'
doc = nlp(unicode(question2))
personnames = [entity.text for entity in doc.ents if entity.label_ == 'PERSON']
p1 = str(personnames[0])
personlikep1 = []
for person in persons:
personlike = person.likes
for p in set(personlike):
if p1 == p.name:
personlikep1.append(person.name)
ref = []
if personlikep1 != ref:
print (answer.format(' and '.join(personlikep1), 'likes',p1))
else:
print ('Sorry, we don\'t know.')
# Who does <person> like?
elif (q_trip.object == 'who') and (q_trip.predicate.endswith('like')):
answer = '{} {} {}'
# doc = nlp(unicode(question2))
personnames = [entity.text for entity in doc.ents if entity.label_ == 'PERSON']
p1 = str(personnames[0])
likes = [person.likes for person in persons if person.name == p1][0]
p1likes = [p.name for p in set(likes) ]
ref = []
if p1likes != ref:
print(answer.format(p1, 'likes', ' and '.join(p1likes)))
else:
print ('Sorry, we don\'t know.')
# Who is [going to | flying to | traveling to | visiting] < place >?
elif (str(t.lemma_) for t in doc if str(t.lemma) in traveldic ) and (q_trip.subject == 'who'):
answer = '{} {} {}'
persontrip = []
ref = []
qplace = [str(entity.text) for entity in doc.ents if entity.label_ == 'GPE']
for person in persons:
qtrip = get_persons_trip(person.name)
#print (person.travels.date)
if qtrip and ( qplace[0] in qtrip.place ):
persontrip.append(person.name)
# print (person.name)
if persontrip != ref:
print (answer.format(' and '.join(persontrip), 'is traveling to', qplace[0]))
else:
print ('Sorry, we don\'t know.')
# Does <person> likes <Pet>?
elif q_trip.subject.startswith('does') and ( q_trip.object == ('cat') or q_trip.object == ('dog')):
answer = '{}, {} does not have a {}.'
new_subj_doc = nlp(unicode(q_trip.subject))
#people = new_subj_doc[5:].capitalize()
# ents = list(new_subj_doc.ents)
# people = [ents[i].text for i in range(len(ents)) if ents[i].label_=='PERSON']
new_subj_doc_childern = new_subj_doc[0].children
people = str([t.text.capitalize() for t in new_subj_doc_childern])
if people in persons:
pet = get_persons_pet(people)
if q_trip.object == pet:
print(answer.format('Yes', people, q_trip.object))
else:
print(answer.format('No', people, q_trip.object))
else:
print (answer.format('No', people, q_trip.object))
else:
print('We don\'t have the material to answer the question.')
def __init__(self):
try:
self.cl = ClausIE.get_instance(
jar_filename='/home/guest/git/DeFactoNLP/clausie/clausie.jar')
except Exception as error:
raise error
from __future__ import print_function
import re
import en_core_web_sm
from pyclausie import ClausIE
nlp = en_core_web_sm.load()
CL_OBJ = ClausIE.get_instance()
re_spaces = re.compile(r'\s+')
class Person(object):
def __init__(self, name, likes=None, has=None, travels=None):
"""
:param name: the person's name
:type name: basestring
:param likes: (Optional) an initial list of likes
:type likes: list
:param dislikes: (Optional) an initial list of likes
:type dislikes: list
:param has: (Optional) an initial list of things the person has
:type has: list
:param travels: (Optional) an initial list of the person's travels
:type travels: list
"""
self.name = name
self.likes = [] if likes is None else likes
self.has = [] if has is None else has
self.travels = [] if travels is None else travels
def main():
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
def main(question):
sents = process_data_from_input_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
for t in triples:
r = process_relation_triplet(t)
answers = []
#question = '?'
#question = raw_input("Please enter your question: [type exit to finish the query] \n")
if question[-1] != '?':
answers.append('This is not a question... please try again')
else:
ques = nlp(unicode(question))
if "What" not in [str(word) for word in ques]:
q_trip = cl.extract_triples([preprocess_question(question)])[0]
else:
q_trip = cl.extract_triples(["Hamid likes Ali"])[0]
"""
1. Who has a <pet_type>?
Checking the wh-word
Checking the pet type mentioned in the question
"""
if q_trip.subject.lower() == 'who' and q_trip.object.lower() == 'dog':
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
answers.append('{} has a {} named {}.'.format(
person.name, 'dog', pet.name))
# here's one just for cats
if q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
answers.append('{} has a {} named {}.'.format(
person.name, 'cat', pet.name))
"""
2. Who is [going to/flying to/traveling to/visiting] <place>?
Checking the wh-word
Checking the verb which is used in the question
"""
if q_trip.subject.lower() == 'who':
check_list = (('going to' in question), ('flying to' in question),
('traveling to' in question), ('visiting'
in question))
if any(check_list):
accepted_trips = [
trip for trip in trips if trip.location in q_trip.object
]
if len(accepted_trips) > 0:
for asked_trip in accepted_trips:
passenger = asked_trip.person
location = asked_trip.location
answers.append('{} is going to {}.'.format(
passenger, asked_trip.location))
else:
answers.append("I don't know")
"""
3. (does <person> like <person>?)
Check if the quesion yes/no or not
Check if the second person in the quesion is in the "like-list" of the first person
"""
if str(ques[0]).lower() == 'does':
accepted_subject = [
person for person in persons if person.name in q_trip.subject
]
if len(accepted_subject) > 0:
first = accepted_subject[0]
accepted_object = [
person for person in persons
if person.name in q_trip.object
]
if len(accepted_object) > 0:
second = accepted_object[0]
if second in first.likes:
answers.append('Yes! {} likes {}.'.format(
first, second))
else:
answers.append('No! {} does not like {}.'.format(
first, second))
else:
answers.append("I don't know")
else:
answers.append("I don't know".format(ques.ents[0]))
"""
4. (What's the name of <person>'s <pet_type>?)
Check if the wh-quesion is what
Check if the person has pet or not
Report the name of the pet
"""
#This part is for dogs
if 'What' in [str(word) for word in ques
] and 'dog' in [str(word) for word in ques]:
name = str(ques.ents[0])
accepted_names = [
person for person in persons if person.name == name
]
if len(accepted_names) > 0:
owner = accepted_names[0]
if len(owner.has) > 0 and owner.has[0].type == "dog":
answers.append("{}'s {}'s name is {}.".format(
owner.name, "dog", owner.has[0].name))
else:
answers.append("{} does not have a pet".format(owner))
else:
answers.append("We have no information about this person")
#This part is for cats
if 'What' in [str(word) for word in ques
] and 'cat' in [str(word) for word in ques]:
name = str(ques.ents[0])
accepted_names = [
person for person in persons if person.name == name
]
if len(accepted_names) > 0:
owner = accepted_names[0]
#print(owner.name)
if len(owner.has) > 0 and owner.has[0].type == "cat":
answers.append("{}'s {}'s name is {}.".format(
owner.name, "cat", owner.has[0].name))
else:
#answers.append("{} does not have a pet".format(owner))
answers.append("I don't know")
else:
answers.append("I don't know")
"""
5. When is <person> [going to/flying to/traveling to/visiting] <place>?
Check if the wh quesion is when
Checking the lemma_ of the root
Inquiring the trip list and "location"; I have defined a class for any trip
"""
if 'When' in list_maker(ques):
check_list = (('going to' in question), ('flying to' in question),
('traveling to' in question), ('visiting'
in question))
if any(check_list):
accepted_trips = [
trip for trip in trips if trip.location in q_trip.object
]
if len(accepted_trips) > 0:
for asked_trip in accepted_trips:
if asked_trip.person in question.split(' '):
time = asked_trip.time
answers.append('In {}.'.format(time))
else:
answers.append("I don't know")
"""
6. (Who likes <person>?)
Check the wh word; also check that the quesion contain like
Check the "like list" of the other player
"""
if q_trip.subject.lower() == 'who' and str(
q_trip.predicate) == 'likes':
accepted_names = [
person for person in persons if person.name == q_trip.object
]
if len(accepted_names) > 0:
loved = accepted_names[0]
lovers = []
for person in persons:
if loved in person.likes:
lovers.append(person)
if len(lovers) > 1:
answers.append('{},and {} like {}.'.format(
', '.join([str(name) for name in lovers[0:-1]]),
lovers[-1], loved))
elif len(lovers) == 1:
answers.append('{} likes {}.'.format(lovers[0], loved))
else:
answers.append('{} likes {}.'.format("no one", loved))
else:
answers.append("I don't know")
"""
7. (Who does <person> like?)
Check the wh word; also check that the quesion contain like
Check the "like list" of the mentioned person
"""
if q_trip.object.lower() == 'who' and q_trip.predicate == 'does like':
accepted_names = [
person for person in persons if person.name == q_trip.subject
]
if len(accepted_names) > 0:
lover = accepted_names[0]
if len(lover.likes) > 1:
answers.append('{} likes {},and {}.'.format(
lover,
', '.join([str(name) for name in lover.likes[0:-1]]),
lover.likes[-1]))
elif len(lover.likes) == 1:
answers.append('{} likes {}.'.format(
lover, lover.likes[0]))
else:
answers.append('{} likes {}.'.format("no one", loved))
else:
lover = None
answers.append("I don't know")
return answers
def answer_question(question_string):
cl = ClausIE.get_instance()
# Q1: Who has a <pet_type>? (e.g. Who has a dog?)
q_trip = cl.extract_triples([preprocess_question(question_string)])[0]
sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
doc = nlp(unicode(sentence))
for t in doc:
if t.pos_ == 'VERB' and t.head == t:
root = t
if q_trip.subject.lower() == 'who' and ('dog' in q_trip.object
or 'cat' in q_trip.object):
answer = '{} has a {}.'
pet_type = 'dog' if 'dog' in q_trip.object else 'cat'
for person in persons:
pet = get_persons_pet(person.name)
if pet:
if pet.type == pet_type:
print(answer.format(person.name, pet_type, pet.name))
# Q2: Who is [going to|flying to|traveling to] <place>? (e.g. Who is flying to Japan?)
if q_trip.subject.lower() == 'who' and (root.lemma_ == 'go'
or root.lemma_ == 'fly'
or root.lemma_ == 'travel'):
answer = '{} is going to {}'
place_doc = nlp(unicode(q_trip.object))
for e in place_doc.ents:
if e.label_ == 'GPE':
place = e.text
# print(place)
for where in trips:
# print(where.place)
if where.place[0] == place:
# a = get_place(where)
print(answer.format(where.traveller, where.place[0]))
# Q3: Does <person> like <person>? (e.g. Does Bob like Sally?)
if 'does' in q_trip.subject.lower() and 'like' in sentence:
list = [e.text for e in doc.ents if e.label_ == 'PERSON']
person_sub = list[0]
person_obj = list[1]
x = 0
for person in persons:
if person.name == person_sub:
for person1 in person.likes:
if person1.name == person_obj:
x = 1
if x:
print("Yes.")
else:
print("No")
# Q4: When is <person> [going to|flying to|traveling to] <place>?
if 'when' in sentence.lower() and (root.lemma_ == 'go' or root.lemma_
== 'fly' or root.lemma_ == 'travel'):
answer = "{} is going to {} in {}."
personD = [
e.text for e in doc.ents
if e.label_ == 'PERSON' or (e.label_ == 'ORG')
][0]
for e in doc.ents:
if e.label_ == 'GPE':
place = e.text
# print(place)
for where in trips:
# print(where.place[0])
# print(where.traveller)
if where.place[0] == place and where.traveller == personD:
print(
answer.format(where.traveller, where.place[0],
where.time[0]))
# 5) Who likes <person>?
if 'who' in q_trip.subject.lower() and root.lemma_ == 'like':
qdoc = nlp(unicode(sentence))
personB = [e.text for e in qdoc.ents if e.label_ == 'PERSON'][0]
for person in persons:
for personC in person.likes:
if personC.name == personB:
print(person)
# 6) Who does <person> like?
if 'does' in q_trip.predicate.lower() and root.lemma_ == 'like':
qdoc = nlp(unicode(sentence))
personD = [
e.text for e in qdoc.ents
if e.label_ == 'PERSON' or (e.label_ == 'ORG')
][0]
m = add_person(personD)
for person in m.likes:
print(person)
def answer_question(question_string):
answered = 1
if question_string.startswith('what') or question_string.startswith('What'):
# For Bonus (5 pts): What's the name of <person>'s <pet_type>? (e.g. What's the name of Mike's dog?)
doc=nlp(unicode(question_string))
answer="{}'s {}'s name is {}"
a=1
for e in doc.ents:
if e.label_=='PERSON':
person=add_person(e.text)
pet_type='dog' if 'dog' in question_string else 'cat'
for x in person.has:
if x.type==pet_type:
# if x.name:
print(answer.format(person.name,pet_type,x.name))
a=0
answered=0
# else:
# print()
if a:
print(""+person.name+" has not a "+pet_type)
answered = 0
else:
IE = ClausIE.get_instance()
q_trip = IE.extract_triples([preprocess_question(question_string)])[0]
# (WHO, has, PET)
# here's one just for dogs
sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
# print("triplet.subject: " + q_trip.subject)
# print("triplet.predicate: " + q_trip.predicate)
# print("triplet.object: " + q_trip.object)
doc = nlp(unicode(sentence))
for t in doc:
if t.pos_ == 'VERB' and t.head == t:
root = t
# print(root)
# elif t.pos_ == 'NOUN'
# also, if only one sentence
# root = doc[:].root
# For q1:Who has a <pet_type>? (e.g. Who has a dog?)
if q_trip.subject.lower() == 'who' and ('dog' in q_trip.object or 'cat' in q_trip.object):
answer = '{} has a {} named {}.'
pet_type = 'dog' if 'dog' in q_trip.object else 'cat'
x = 1
for person in persons:
pet = get_persons_pet(person.name)
if pet:
if pet.type == pet_type:
print(answer.format(person.name, pet_type, pet.name))
x = 0
answered = 0
if x:
print('Nobody has a' + pet_type)
answered = 0
# for answer in answers:
# print answer
# For q2: Who is [going to|flying to|traveling to] <place>? (e.g. Who is flying to Japan?)
if q_trip.subject.lower() == 'who' and (
root.lemma_ == 'go' or root.lemma_ == 'fly' or root.lemma_ == 'travel'):
answer = '{} is going to {}, time:{}.'
# place=t.text for t in doc.ents if t.label_=='GPE'
place_doc = nlp(unicode(q_trip.object))
a = 1
for x in place_doc.ents:
if x.label_ == 'GPE':
place = x.text
for trip in trips:
if trip.to == place:
print(answer.format(trip.name, trip.to, trip.time))
a = 0
answered = 0
if a:
print('Nobody is going to ' + x.text)
answered = 0
# For q3: Does <person> like <person>? (e.g. Does Bob like Sally?)
if 'does' in sentence.lower() and 'like' in sentence and 'who' not in sentence.lower():
list = [e.text for e in doc.ents if e.label_ == 'PERSON' or e.label_ == 'ORG']
person_sub = list[0]
person_obj = list[1]
x = 0
for person in persons:
if person.name == person_sub:
for person1 in person.likes:
if person1.name == person_obj:
# print(x)
x = 1
# print(x)
# print(x)
if x:
print("Yes.")
answered = 0
else:
print("No.")
answered = 0
# For q4:When is <person> [going to|flying to|traveling to] <place>?
if 'when' in q_trip.object.lower() and (root.lemma_ == 'go' or root.lemma_ == 'fly' or root.lemma_ == 'travel'):
answer = '{} is going to {} in {}.'
# answer1 = '{} is not going to {} in {}.'
# if q_trip.subject in [e.text for e in doc.ents if e.label_ == 'PERSON' or e.label_ == 'ORG']:
# print(0)
when_who = [str(e) for e in doc.ents if e.label_ == 'PERSON' or e.label_ == 'ORG']
when_all = ' '.join(when_who)
when_final = when_all.split(' ')
# if len(q4_persosn)<1:
# q4_persosn[0]='Mary'
# print(trips[0].name)
place1_doc = nlp(unicode(q_trip.object))
for e in place1_doc.ents:
if e.label_ == 'GPE':
q4_place = e.text
break
for x in when_final:
# print(x)
# print(q4_place)
if q_trip.subject=='Mary':
x='Mary'
# print(x)
for trip in trips:
if trip.to == q4_place:
if trip.name.name == x:
print(answer.format(trip.name, trip.to, trip.time))
answered = 0
# if x==trips[0].name.name:
# # print(1)
else:
for trip in trips:
if trip.to == q4_place:
if trip.name.name == x:
print(answer.format(trip.name, trip.to, trip.time))
answered = 0
# elif trip.name.name == x and trip.to != q4_place:
# print(answer1.format(trip.name, trip.to, trip.time))
# answered = 0
# For q5:Who likes <person>?
if 'like' in sentence and 'who' in q_trip.subject.lower():
if q_trip.object in [e.text for e in doc.ents if e.label_ == 'PERSON'or e.label_ == 'ORG']:
x = 1
person_like = q_trip.object
for person in persons:
for x in person.likes:
if x.name == person_like:
print(person.name)
x = 0
answered = 0
# if x:
# print('Nobody likes ' + person_like+'.')
# answered = 0
# For q6:Who does <person> like?
if 'like' in sentence and 'who' in q_trip.object.lower() and 'does' in sentence:
# print(1)
if 'Mary'in sentence:
# print(0)
person_like = add_person('Mary')
for a in person_like.likes:
print(a.name)
answered = 0
elif q_trip.subject in [e.text for e in doc.ents if e.label_ == 'PERSON'or e.label_ == 'ORG']:
person_like = add_person(q_trip.subject)
if len(person_like.likes) == 0:
print('' + person_like.name + ' likes nobody.')
answered = 0
else:
for a in person_like.likes:
print(a.name)
answered = 0
if answered:
print("Sorry, I don't know.")
def answer_question(question_string):
if 'What' in question_string:
e_doc = nlp(unicode(question_string))
for e in e_doc.ents:
#print(9)
if e.label_ == 'PERSON':
person = e.text
#print(person)
for x in persons:
if x.name == person:
for e in x.has:
#print(e.name)
if e.name != None:
print(e.name)
else:
cl = ClausIE.get_instance()
# q1
q_trip = cl.extract_triples([preprocess_question(question_string)])[0]
sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
doc = nlp(unicode(sentence))
# print(q_trip.object)
# print(q_trip.predicate)
# print(q_trip.subject)
if q_trip.subject.lower() == 'who' and ('dog' in q_trip.object
or 'cat' in q_trip.object):
answer = '{} has a {}.'
pet_type = 'dog' if 'dog' in q_trip.object else 'cat'
for person in persons:
pet = get_persons_pet(person.name)
if pet:
if pet.type == pet_type:
print(answer.format(person.name, pet_type, pet.name))
# q2
if q_trip.subject.lower() == 'who' and (root.lemma_ == 'go' or 'fly'
or 'travel'):
answer = '{} is going to {}, time:{}'
place_doc = nlp(unicode(q_trip.object))
for e in place_doc.ents:
if e.label_ == 'GPE':
place = e.text
# print(place)
for trip in trips:
if trip.to == place and len(trip.time) > 0:
print(answer.format(trip.name, trip.to, trip.time))
# q3
if 'does' in q_trip.subject.lower():
# answer = '{} likes {}'
sent = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
e_doc = nlp(unicode(sent))
list = [
str(e) for e in e_doc.ents
if e.label_ == 'PERSON' or e.label_ == 'ORG'
]
x = 0
# print(list)
subject = list[0]
object = q_trip.object
for person in persons:
# print(9)
if person.name == subject:
for personA in person.likes:
# print(personA.name)
if personA.name == object:
x = 1
# print(x)
if x:
print("y")
else:
print("N")
# q4
if 'when' in q_trip.object and (root.lemma_ == 'go' or root.lemma_
== 'fly' or root.lemma_ == 'travel'):
answer = '{} is going to {}, time:{}'
e_doc = nlp(unicode(q_trip.object))
for e in e_doc.ents:
if e.label_ == 'GPE':
place = e.text
# print(place)
for trip in trips:
if trip.to == place and len(trip.time) > 0:
print(answer.format(trip.name, trip.to, trip.time))
# q5
if 'who' in q_trip.subject.lower():
answer = '{} likes {}'
# e_doc = nlp(unicode(q_trip.object))
for e in doc.ents:
if e.label_ == 'PERSON':
person1 = e.text
#print(person1)
for person in persons:
for e in person.likes:
if e.name == q_trip.object:
print(person.name)
# q6
if root.lemma_ == 'like':
x = 1
for x in doc:
if x.text == "n't":
x = 0
if x:
fw_doc = nlp(unicode(q_trip.object))
if 'who' in q_trip.object.lower():
answer = '{} likes {}'
e_doc = nlp(unicode(question_string))
for e in e_doc.ents:
if e.label_ == 'PERSON':
person = e.text
for person in persons:
#print(person)
if person.name == q_trip.subject:
for e in person.likes:
print(e.name)
def process_what_petname_question(string):
cl = ClausIE.get_instance()
str = string.replace('What\'s', '').strip()
triples = cl.extract_triples([str])[0]
w_quetsion = triples.subject + ' ' + triples.predicate + ' ' + triples.object + ' ' + 'name'
return w_quetsion
def answer_question(question=' '):
while question[-1] != '?':
question = raw_input("Please enter your question: ")
if question[-1] != '?':
print('This is not a question... please try again')
cl = ClausIE.get_instance()
q_trip = cl.extract_triples([preprocess_question(question)])[0]
triplet_sentence = q_trip.subject + ' ' + q_trip.predicate + ' ' + q_trip.object
doc = nlp(unicode(triplet_sentence))
root = doc[:].root
# retrieve answers for questions like (WHO, has, PET)
# here's one just for dogs
# you should really check the verb... this is just an example not the best way to do things
if q_trip.subject.lower() == 'who' and root.lemma_ == 'have':
answer = '{} has a {} named {}.'
if q_trip.object == 'dog':
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
print(answer.format(person.name, pet.type, pet.name))
elif q_trip.object == 'cat':
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
print(answer.format(person.name, pet.type, pet.name))
# retrieve answers for questions like (WHO, like, PERSON)
# again this is just an example, NOT the best way to do things. That's for you to figure out.
elif q_trip.subject.lower(
) == 'who' and root.lemma_ == 'like' and q_trip.object in [
e.text
for e in doc.ents if e.label_ == 'PERSON' or e.label_ == 'ORG'
]:
answer = '{} likes {}'
liked_person = select_person(q_trip.object)
for person in persons:
if person.name != q_trip.object and liked_person in person.likes:
print(answer.format(person.name, liked_person.name))
elif q_trip.subject.lower(
) == 'does' and q_trip.predicate == 'like' and q_trip.object in [
e.text for e in doc.ents if e.label_ == 'PERSON'
]:
person = select_trip(q_trip.subject)
liked_person = select_person(q_trip.object)
if liked_person in person.likes:
print(person.name, 'likes ', liked_person.name)
else:
print(person.name, 'does not like ', liked_person.name)
elif q_trip.subject.lower(
) == 'who' and q_trip.predicate == 'likes' and q_trip.object in [
e.text for e in doc.ents if e.label_ == 'PERSON'
]:
like_person = select_person(q_trip.object)
for person in persons:
if like_person in person.likes:
print(person.name, 'likes ', like_person.name)
else:
print("I don't know.")
def answer_questions(string):
sents = get_data_from_file()
cl = ClausIE.get_instance()
triples = cl.extract_triples(sents)
q_trip = cl.extract_triples([preprocess_question(string)])[0]
answers = []
# (WHO, has, PET)
# here's one just for dogs
print(q_trip)
if q_trip.subject.lower() == 'who' and q_trip.object == 'dog':
answer = '{} has a dog.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'dog':
answer = (answer.format(person.name))
answers.append(answer)
if q_trip.subject.lower() == 'who' and q_trip.object == 'cat':
answer = '{} has a cat.'
for person in persons:
pet = get_persons_pet(person.name)
if pet and pet.type == 'cat':
answer = (answer.format(person.name))
answers.append(answer)
if q_trip.subject.lower() == 'what' and 'name' in q_trip.object:
answer = '{} has a {} name {}'
for t in q_trip.object:
if t.pos_ == 'PERSON':
person = t.text
pet = get_persons_pet(person)
answer = answer.format(person, pet, pet.name)
if q_trip.subject.lower == 'who' and q_trip.predicate == 'likes':
answer = '{} likes {}'
for i in q_trip.object:
if i.pos == 'PERSON':
target = i
for people in persons:
if people.likes == target:
answer = answer.format(people.name, target)
answers.append(answer)
if q_trip.subject.lower() == 'does' and q_trip.predicate == 'like':
answer = '{} likes {}'
for t in q_trip.subect:
if t.pos_ == 'PERSON':
Sender = t
Receiver = q_trip.object
if Sender.likes == Receiver:
answer = answer.format(Sender, Receiver)
answers.append(answer)
if q_trip.subject.lower() == 'who' and (q_trip.predicate == 'going'
or q_trip.predicate == 'flying'
or q_trip.predicate == 'traveling'
or q_trip.predicate == 'visiting'):
destination = q_trip.object
answer = '{} is going on a trip to {}'
for person in persons:
if person.travels == destination:
answer = (answer.format(person.name), destination)
answers.append(answer)
if q_trip.subject.lower() == 'when' and (
q_trip.predicate == 'going' or q_trip.predicate == 'flying' or
q_trip.predicate == 'traveling' or q_trip.predicate == 'visiting'):
answer = '{} is going on a trip in {}'
for t in q_trip.object:
if t.pos_ == 'GPE':
destination = q_trip.object
for trips in Trip:
if Trip.departs_to == destination:
answer = answer.format(Trip.departs_on)
answers.append(answer)
for answer in answers:
if answers != []:
print(answer)
else:
print('I do not know')
from __future__ import print_function
import re
import spacy
from pyclausie import ClausIE
nlp = spacy.load('en_core_web_sm')
re_spaces = re.compile(r'\s+')
cl = ClausIE.get_instance()
class Person(object):
def __init__(self, name, likes=None, has=None, travels=None):
"""
:param name: the person's name
:type name: basestring
:param likes: (Optional) an initial list of likes
:type likes: list
:param dislikes: (Optional) an initial list of likes
:type dislikes: list
:param has: (Optional) an initial list of things the person has
:type has: list
:param travels: (Optional) an initial list of the person's travels
:type travels: list
"""
self.name = name
self.likes = [] if likes is None else likes
self.has = [] if has is None else has
self.travels = [] if travels is None else travels
def __repr__(self):
'''
Created on 2017.8.27
@author: ZhongqiLi
'''
from pyclausie import ClausIE
cl = ClausIE.get_instance(jar_filename='./clausie/clausie.jar')
for i in range(1):
S = ['I learned that the 2012 Sasquatch music festival is scheduled for May 25th until May 28.']
triples = cl.extract_triples(S)
for t in triples:
print t
