def comment_Analysis(self, commmentlist): #对每条评价进行打分，最高1.0，分越低说明评价越消极
xm.set_stopword('手机关键词停词表.txt')
totalscore = 0
for commtent in commmentlist:
totalscore = totalscore + xm.sentiment(commtent)
avescore = totalscore / len(commmentlist)
return avescore
def get_data():
html = login()
if (html == False):
get_cookie()
html = login()
else:
soup = BeautifulSoup(html.content, 'lxml')
friends = soup.findAll(attrs={'class': 'f-name q_namecard'})
msgs = soup.findAll(attrs={'class': 'f-info'})
with open('data.csv', 'a+', encoding='utf-8') as fo:
for i in range(len(friends) - 1):
num = str(friends[i].attrs['href'])
doc = msgs[i].text
score = xmnlp.sentiment(doc)
print(doc)
print('Score: ', score)
res = ""
if score > 0.49:
res = '积极'
else:
res = '消极'
print("%s,%s,%s,%d,%s" %
(friends[i].text, num.split('/')[-1], doc, score, res))
fo.writelines("%s,%s,%s,%s,%s\n" %
(friends[i].text, num.split('/')[-1], doc,
str(score), res))
fo.close()
def create(cls, cfg):
# type: (RasaNLUModelConfig) -> JiebaTokenizer
#模块的算法加载。
import xmnlp
#xmnlp.set_stopword('/path/to/stopword.txt') # 用户自定义停用词
#下次补上具体creat的代码
doc = """这件衣服的质量也太差了吧！"""
print('Load sentiment analyzer.Text: ', doc, 'score',
xmnlp.sentiment(doc))
component_conf = cfg.for_component(cls.name, cls.defaults)
analyzer = xmnlp.sentiment
return cls(component_conf, analyzer)
def addscore(chart):
# add a column containing the sentiment score on the original 4-column chart
chart1 = copy.deepcopy(chart)
for vvv in range(len(chart1)):
if len((chart1[vvv])[3]) == 0:
chart1[vvv].append("blank")
chart1[vvv].append("blank")
chart1[vvv].append("blank")
else:
x = (chart1[vvv])[3]
s = SnowNLP(chart1[vvv][3])
t1 = xmnlp.sentiment(x)
t2 = s.sentiments
t3 = (t1 + t2) / 2
chart1[vvv].append(t1)
chart1[vvv].append(t2)
chart1[vvv].append(t3)
return chart1
def analysis(self, filein_path, fileout_path):
isFileInExists = os.path.exists(filein_path)
isFileOutExists = os.path.exists(fileout_path)
if isFileInExists is False:
print 'in file: {0} not exits.'.format(filein_path)
return
if isFileOutExists is False:
print 'out file: {0} not exits.'.format(fileout_path)
self.writeToCSVWithoutHeader(fileout_path, [
'share_number', 'comment_number', 'url', 'title', 'sentiment'
])
print 'create an new out file: {0}.'.format(fileout_path)
in_content = self.readFromCSV(filein_path)
in_content.pop(0)
for item in in_content:
s = xmnlp.sentiment(item[3])
self.writeToCSVWithoutHeader(
fileout_path, [item[0], item[1], item[2], item[3], s])
print "{0}--{1}".format(item[3], s)
def get_pos_neg(text_list):
positive_texts = []
negative_texts = []
for text in text_list:
text = text.replace('<','')
text = text.replace(' ','')
if len(text)>0:
score1 = SnowNLP(text).sentiments
score2 = xmnlp.sentiment(text)
tmp = "".join(jieba.analyse.textrank(text))
if len(tmp)>0:
score3 = SnowNLP(tmp).sentiments
if score1<0.45 and score2<0.45 and score3<0.45:
negative_texts.append(text)
if score1>0.55 and score2>0.55 and score3>0.55:
positive_texts.append(text)
if len(tmp)==0:
if score1<0.45 and score2<0.45:
negative_texts.append(text)
if score1>0.55 and score2>0.55:
positive_texts.append(text)
return positive_texts, negative_texts
def comment_extract(self, commentlist):
rank = []
BEST = []
WORST = []
for comment in commentlist: #删除长度小于30的评论
if len(comment) < 12:
commentlist.remove(comment)
for comment in commentlist:
group = (xm.sentiment(comment), comment)
rank.append(group)
ranklist = sorted(rank)
num = int(len(ranklist) * 0.05)
bestComment = ranklist[num - 1:num + 4]
for comment in bestComment:
BEST.append(comment[1])
worstComment = ranklist[-num - 6:-num - 1]
for comment in worstComment:
WORST.append(comment[1])
return WORST, BEST #返回五条评分高的评价和五条比较低分的评价
def test_sentiment():
score = xmnlp.sentiment('这酒店真心不错')
assert score > 0.5
情感计算
/ naive bayes /
"""
print(descr)
doc = """真伤心"""
doc2 = """天气太好了，我们去钓鱼吧"""
print('\n++++++++++++++++++++++++ usage 1 ++++++++++++++++++++++++\n')
"""
1. 使用类来进行操作
"""
from xmnlp import XmNLP
xm = XmNLP(doc, stopword=True)
print('Text: ', doc)
print('Score: ', xm.sentiment())
print('Text: ', doc2)
print('Score: ', xm.sentiment(doc2))
print('\n++++++++++++++++++++++++ usage 2 ++++++++++++++++++++++++\n')
import xmnlp
print('Text: ', doc)
print('Score: ', xmnlp.sentiment(doc))
print('Text: ', doc2)
print('Score: ', xmnlp.sentiment(doc2))
def test_sentiment():
score = xmnlp.sentiment('这酒店真心不错哦')
assert score[1] > 0.5
score = xmnlp.sentiment('这酒店真心太差了')
assert score[1] < 0.5