def extractRevsLabel(filetype, pattern):
review = open('../data/chres_review.json')
revsKW = open('../data/revsKW'+filetype+'.json','w+')
revs = []
lab = []
i = 1
if pattern=='AllWords':
for index,line in enumerate(review):
jre = json.loads(line)
jstar = jre['stars']
text = jre['text']
ws = tp.removeStopPunc(text)
lab.append(jstar)
revs.append(ws)
print i
# if i==50: break
i += 1
else:
for index,line in enumerate(review):
jre = json.loads(line)
jstar = jre['stars']
text = jre['text']
tagText = tp.posTag(text)
adj = tp.posExtract(tagText,pattern)
adjs = ' '.join(adj)
ws = tp.removeStopPunc(adjs)
lab.append(jstar)
revs.append(ws)
print i
# if i==50: break
i += 1
np.save('../data/label'+filetype+'.npy', np.array(lab))
json.dump(revs, revsKW)
review.close()
revsKW.close()
return 1
def extractRevsLabel(filetype, pattern):
review = open('../data/chres_review.json')
revsKW = open('../data/revsKW' + filetype + '.json', 'w+')
revs = []
lab = []
i = 1
if pattern == 'AllWords':
for index, line in enumerate(review):
jre = json.loads(line)
jstar = jre['stars']
text = jre['text']
ws = tp.removeStopPunc(text)
lab.append(jstar)
revs.append(ws)
print i
# if i==50: break
i += 1
else:
for index, line in enumerate(review):
jre = json.loads(line)
jstar = jre['stars']
text = jre['text']
tagText = tp.posTag(text)
adj = tp.posExtract(tagText, pattern)
adjs = ' '.join(adj)
ws = tp.removeStopPunc(adjs)
lab.append(jstar)
revs.append(ws)
print i
# if i==50: break
i += 1
np.save('../data/label' + filetype + '.npy', np.array(lab))
json.dump(revs, revsKW)
review.close()
revsKW.close()
return 1
vocabulary = open('../data/vocabularyNN.txt','w+')
label = open('../data/labelNN.txt','w+')
data = open('../data/dataNN.txt','w+')
voca = []
revs = []
lab = []
dat = []
i=1
for line in review:
jre = json.loads(line)
jstar = jre['stars']
text = jre['text']
tagText = tp.posTag(text)
adj = tp.posExtract(tagText,'NN.*')
adjs = ' '.join(adj)
ws = tp.removeStopPunc(adjs)
if ws!=[]:
lab.append(jstar)
revs.append(ws)
voca.extend(ws)
print i
# if i==5: break
i += 1
print lab
for i in lab:
