def test_neutral(to_stdout=True):
"""
neutral表示作者的话
:param to_stdout:
:return:
"""
print 'test neutral'
sentiment.load('../data/impurity_classifier')
result_file = None
if not to_stdout:
result_file = codecs.open('../data/result.csv', 'w', encoding='gbk', errors='ignore')
with codecs.open('../data/clean_neutral.txt', encoding='utf-8') as neutral_file:
for line in neutral_file:
line = line.strip()
prob = sentiment.classify(line)
if 0.8 > prob > 0.2:
if to_stdout:
print (line + ',' + str(prob > 0.5 and 1 or 0) + ',' + str(prob) + cur_linesep).encode('gbk')
raw_input('press enter to continue')
else:
result_file.write(line + ',' + str(prob > 0.5 and 1 or 0) + ',' + str(prob) + cur_linesep)
if not to_stdout:
result_file.close()
def get_type_good_courses(self):
#print("in")
course_type = self.get_name("sst")
sql = f"match(c:Course) return c.title"
ans = self.graph.run(sql)
results = []
for course in ans :
sql = f"match(c:Course)-[] ->(ce:Course_Type) where c.title = '{course}' return ce.name"
type_ = self.graph.run(sql)[0]
if type_ == course_type :
sql = f"match(c:Course)-[]->() where c.title = '{course}' return c.rating"
rating = self.graph.run(sql)[0]
score = sentiment.classify(rating)
if score >= 0.4:
print(course,score)
results.append(course)
if len(results)>0:
res_ = "、".join(results)
else :
return "该类型下没有对应的课程"
res = course_type + "评价不错的课程有: " + res_ + "等～"
return res
def get_school_good_courses(self):
school = self.get_name("ut")
course_type = self.get_name("sst")
sql = f"match(s:School)-[] -> (c:Course) where s.name = '{school}' return c.title"
courses = self.graph.run(sql)
results = []
for course in courses:
sql = f"match(c:Course)-[] ->(ce:Course_Type) where c.title = '{course}' return ce.name"
type_ = self.graph.run(sql)[0]
if type_ == course_type:
sql = f"match (c:Course)-[]->() where c.title='{course}' return c.rating"
rating = self.graph.run(sql)[0]
score = sentiment.classify(rating)
if score >= 0.4:
print(course,score)
results.append(course)
if len(results) == 0:
return "该院系没有对应的课程"
else :
res = "、".join(results)
ret = school + "开设的" + "评价不错的" + course_type + "有: " + res + "等～"
return ret
def classify(cases):
"""
测试列表中每个句子的
:param cases:
:return:
"""
for case in cases:
case = clean_impurity(case)
sentiment.load('../data/impurity_classifier')
prob = sentiment.classify(case)
print (case + ',' + str(prob > 0.5 and 1 or 0) + ',' + str(prob) + cur_linesep).encode('gbk')
def emotionCalculate(self, name, sent, comment_score):
# 情感分析，（已经确定name在句子str里面了）累加该评论中含有该名字的短句感情值
emotion = 0.0
num = 0
s_em = 0
for s in re.split(',|\.|;|，|。|；|！', sent):
if s.find(name) != -1:
em = sentiment.classify(sent.decode('utf8')) - 0.5
s_em += em
num += 1
print s
print "em: " + str(em)
if num >= 1:
emotion += s_em * 1.0 / num
result = 0.5 * int(comment_score) + 5 * emotion
print "0.5 * " + comment_score + ", 5 *" + str(emotion) + " => " + str(result)
return result
def test_sentiment():
print 'test model'
sentiment.load('../data/train_impurity_classifier')
print 'test_negative'
# with codecs.open('../data/test_negative.txt', encoding='utf-8') as negative_file:
# for line in negative_file:
# if sentiment.classify(line) > 0.1:
# print line,
raw_input('press enter to continue')
print 'test_positive'
with codecs.open('../data/test_positive.txt', encoding='utf-8') as positive_file:
for line in positive_file:
if sentiment.classify(line) < 0.5:
print line,
def testEmotionModel(self, path):
# 测试训练出来的情感模型，如果模型打分0.6一下，并且该评论分低于3分，认为正确
num = 0.0
with open(path) as file:
lines = file.readlines()
for line in lines:
arr = line.split("\001")
com = arr[5].replace("\"", "").replace("\n", "")
score = int(arr[4]) / 10
em = sentiment.classify(com)
print com
print "score is " + str(score) + " emotion is " + str(em)
if em < 0.6 and score < 3:
num += 1.0
else:
if em >= 0.6 and score >= 3:
num += 1.0
return num
from snownlp import SnowNLP
from snownlp import sentiment
import jieba
jieba.load_userdict('words.txt')
# string='原标题：市场信心不足，大盘反弹受阻'
string = '信心'
sent = sentiment.Sentiment()
words_list = sentiment.Sentiment.handle(sent, string)
# score=SnowNLP('原标题：市场信心不足，大盘反弹受阻')
score = SnowNLP('信心')
print(words_list)
s = score.sentiments
# s=sentiment.Sentiment.classify(sent,'原标题：市场信心不足，大盘反弹受阻')
print(s)
text = '原标题：市场信心不足，大盘反弹受阻'
print(sentiment.classify(text))
# #设置jieba自定义词库
# import jieba
# jieba.load_userdict('words.txt') #自己准备的常用词词典
# #优化后的lcut
# a=jieba.lcut('需要分词的文本,市场信心不足')
# a=SnowNLP(a)
# print(a.sentiments)
# # print(list(a))
目前健保已通過治療藥物，可減緩肺功能下降近五成，急性惡化機率減少六成八，如持續治療，能降低死亡風險四成三；已有患者用藥後惡化情形趨緩，從無法活動到每日步行一點五公里，提高生活品質，引起全台胸腔內科醫師大規模搜查疑似病例，欲揪出潛在患者。
張時杰表示，初步篩檢可透過公司企業或自費進行胸部Ｘ光合併吹氣肺功能檢查，後續再以臨床診斷、高解析度電腦斷層確診。
菜瓜布肺的危險因子，包含逾五十歲、直系家人或近親曾有病史、吸菸、暴露於化學工廠環境作業員等風險較高。
'''
from snownlp import normal
from snownlp import seg
from snownlp import sentiment
from snownlp.summary import textrank
if __name__ == '__main__':
sents = normal.get_sentences(text)
doc = []
#summary
for sent in sents:
words = seg.seg(sent)
words = normal.filter_stop(words)
doc.append(words)
rank = textrank.TextRank(doc)
rank.solve()
for index in rank.top_index(5):
print(sents[index])
#probability of the sentiment
pro = sentiment.classify(text)
print(pro)
