def buildTestData():
xtest = []
sp = SenticPhrase('')
for i, data in enumerate(testDS):
data1 = []
lexScore = 0
window = 5
atPos = int(testSents[i][1].rsplit('-', maxsplit=1)[1])
for lind in range(atPos - window, atPos):
if lind >= 0:
if data[1][lind] in posLex:
lexScore += 1
elif data[1][lind] in negLex or data[1][lind] in negWords:
lexScore -= 1
for lind in range(atPos + 1, atPos + window + 1):
if lind >= len(data[1]):
break
else:
if data[1][lind] in posLex:
lexScore += 1
elif data[1][lind] in negLex or data[1][lind] in negWords:
lexScore -= 1
senScore = sp.get_polarity(testSents[i][0])
protoVec = np.append(testXtidf[i].A[0], [lexScore, senScore])
xtest.append(protoVec)
xtest = np.array(xtest)
return xtest
def buildTrainingData():
X1 = []
yForSk = []
sp = SenticPhrase('')
for i, data in enumerate(dataset):
data1 = []
# creating lexicon score with window of 5 to the left and right of aspect
lexScore = 0
window = 2
atPos = int(sentences[i][1].rsplit('-', maxsplit=1)[1])
for lind in range(atPos - window, atPos):
if lind >= 0:
if data[1][lind] in posLex:
lexScore += 1
elif data[1][lind] in negLex or data[1][lind] in negWords:
lexScore -= 1
for lind in range(atPos + 1, atPos + window + 1):
if lind >= len(data[1]):
break
else:
if data[1][lind] in posLex:
lexScore += 1
elif data[1][lind] in negLex or data[1][lind] in negWords:
lexScore -= 1
senScore = sp.get_polarity(sentences[i][0])
protoVec = np.append(Xtidf[i].A[0], [lexScore, senScore])
X1.append(protoVec)
yForSk.append(data[-1])
X1 = np.array(X1)
yForSk = np.array(yForSk)
return X1, yForSk
def sentyment(wordlist):
sp = SenticPhrase(wordlist)
if sp.info(wordlist)['sentiment'] == 'strong negative':
text_sentiment = -2
elif sp.info(wordlist)['sentiment'] == 'weak negative':
text_sentiment = -1
elif sp.info(wordlist)['sentiment'] == 'neutral':
text_sentiment = 0
elif sp.info(wordlist)['sentiment'] == 'weak positive':
text_sentiment = 1
elif sp.info(wordlist)['sentiment'] == 'strong positive':
text_sentiment = 2
return text_sentiment
def sentyki(wordlist):
sp = SenticPhrase(wordlist)
try:
text_aptitude = sp.info(wordlist)['sentics']['aptitude']
except KeyError:
text_aptitude = 0
try:
text_pleasantness = sp.info(wordlist)['sentics']['pleasantness']
except KeyError:
text_pleasantness = 0
try:
text_attention = sp.info(wordlist)['sentics']['attention']
except KeyError:
text_attention = 0
try:
text_sensitivity = sp.info(wordlist)['sentics']['sensitivity']
except KeyError:
text_sensitivity = 0
return text_aptitude, text_attention, text_pleasantness, text_sensitivity
results[1] = results[1].union(
set([ancestor.lemma_.lower() for ancestor in token.ancestors if len(ancestor) > 2]).union(
set([child.lemma_.lower() for child in token.head.children if child.text != token.text and len(child) > 2])
)
)
results[4] += 1
elif token.lemma_.lower() in hedging_words:
results[2] = results[1].union(
set([ancestor.lemma_.lower() for ancestor in token.ancestors if len(ancestor) > 2]).union(
set([child.lemma_.lower() for child in token.head.children if child.text != token.text and len(child) > 2])
)
)
results[5] += 1
return tuple(results)
sp = SenticPhrase("Hello, World!")
def get_sentics(sent):
"""
input: Spacy processed sentence
output: a tuple containing the polarity score and a list of sentic values
(pleasantness, attention, sensitiviy, aptitude )
"""
info = sp.info(sent)
# Sometimes sentic doesn't returns any sentics values, seems to be only when purely neutral.
# Some sort of tag to make sure this is true could help with classiciation! (if all 0's not enough)
sentics = {"pleasantness":0, "attention":0, "sensitivity":0, "aptitude":0}
sentics.update(info["sentics"])
return [info['polarity'], sentics['aptitude'], sentics['attention'], sentics['sensitivity'], sentics['pleasantness']]
padding=pad,
truncating=pad)
x_test_pad = pad_sequences(x_test_tokens,
maxlen=max_tokens,
padding=pad,
truncating=pad)
x_train_pad[0]
# In[11]:
#Create a polarity Array
x_train_polarityScore = []
for comment in x_train:
comment = SenticPhrase(comment)
x_train_polarityScore.append(comment.get_polarity())
x_train_polarityScore = np.array(x_train_polarityScore)
# In[12]:
x_test_polarityScore = []
for comment_test in x_test:
comment_test = SenticPhrase(comment_test)
try:
x_test_polarityScore.append(comment_test.get_polarity())
except:
x_test_polarityScore.append(0)
x_test_polarityScore = np.array(x_test_polarityScore)
else:
dict[name] += 1
for i in range(1,len(crew1)):
for name in crew1[i]:
if name not in dict:
dict[name] = 1
else:
dict[name] += 1
crew1 = []
for key in dict:
if dict[key]>=1:
crew1.append(key)
crew.append(crew1)
for i in range(len(movie_content)):
sp = SenticPhrase(movie_content[i])
pol += sp.get_polarity()
n += 1
polarity.append(pol/n)
else:
polarity.append(0.00)
crew.append([])
print(crew)
print("\n\n",len(crew),len(title))
print('\n\n\n\n\n\t\t\t\t\t\t\t\tMOVIE REVIEW')
for i in range(len(title)):
print('Title: ',title[i])
print('Stars: ',stars[i])
print('Polarity: ',polarity[i])
print('Crew:\n')
for name in crew[i]: