文本情感分析
从上一篇完成了对新浪微博的爬取，以及模拟登录的问题，小编又开始研究对微博文
本的正面和反面分析，从网上搜索了好多方法，有机器学习和情感字典，可是机器学
习需要比较深的知识链，而小编还是小白，所以就选择了情感字典方法。好了，直接
上代码嘛，直接可以运行的。
前提需要安装相关的库jieba,
直接用pip install jieba
就能安装，
一、数据准备
先要准备情感字典，
- 情感字典，如：
最尼玛 -6.70400012637
扰民 -6.49756445867
fuck… -6.32963390433
RNM -6.21861284426
wcnmlgb -5.96710044003
2.5: -5.90459648251 - 停用字字典，如：
!，”，#，$，& - 副词，如：
百分之百 6
倍加 6
备至 6
不得了 6
不堪 6
不可开交 6
不亦乐乎 6 - 否定词，如：不，没，无，非，莫，弗，勿
如果需要可以去CSDN去下载，我已经上传了
二、情感分析
#!usr/bin/env python
#coding:utf-8
import jieba
class SentimentAnalysis:
#初始化
def __init__(self,sentiment,noword,adverb,stopword):
self.__readFile(sentiment,noword,adverb,stopword)
#读取相关文库
def __readFile(self,sentiment,noword,adverb,stopword):
self.__sentList = {}
self.__noword = []
self.__adverb = {}
self.__stopword = []
#情感词
sentList = open(sentiment,'rb')
for s in sentList.readlines():
try:
s = s.replace('\r\n','').replace('\n','')
self.__sentList[s.split(' ')[0]] = s.split(' ')[1]
except:
pass
sentList.close()
nowordList = open(noword,'rb')
for s in nowordList.readlines():
try:
s = s.replace('\r\n','').replace('\n','')
self.__noword.append(s)
except:
print "数据错误："+s
nowordList.close()
adverbList = open(adverb,'rb')
for s in adverbList.readlines():
try:
s = s.replace('\r\n','').replace('\n','')
self.__adverb[s.split(' ')[0]] = s.split(' ')[1]
except:
print "数据错误："+s
adverbList.close()
stopwordList = open(stopword,'rb')
for s in stopwordList.readlines():
try:
s = s.replace('\r\n','').replace('\n','')
self.__stopword.append(s)
except:
print "数据错误："+s
stopwordList.close()
def setSentence(self,sentence):
self.__sentence = sentence.lstrip()
#预处理
def preDetail(self):
wordsList = jieba.cut(self.__sentence, cut_all=False)
newWords = {}
i = 0
for w in wordsList:
if w not in self.__stopword:
newWords[str(i)] =w
i = i+1
senWord = {}
notWord = {}
degreeWord = {}
m = 0
for index in newWords.keys():
if newWords[index] in self.__sentList.keys() and newWords[index] not in self.__noword and newWords[index] not in self.__adverb.keys():
senWord[index] = self.__sentList[newWords[index].encode('utf-8')]
elif newWords[index] in self.__noword and newWords[index] not in self.__adverb.keys():
notWord[index] = -1
elif newWords[index] in self.__adverb.keys():
degreeWord[index] = self.__adverb[newWords[index].encode('utf-8')]
else:
senWord[index] = 0
return senWord,notWord,degreeWord,newWords
def getScore(self):
senWord,notWord,degreeWord,newWords = self.preDetail()
W = 1
score = 0
# 存所有情感词的位置的列表
senLoc = []
notLoc = []
degreeLoc = []
for i in senWord.keys():
senLoc.append(int(i))
for i in notWord.keys():
notLoc.append(int(i))
for i in degreeWord.keys():
degreeLoc.append(int(i))
senLoc.sort()
notLoc.sort()
degreeLoc.sort()
senloc = -1
for i in range(0, len(newWords)):
# 如果该词为情感词
if i in senLoc:
# loc为情感词位置列表的序号
senloc += 1
# 直接添加该情感词分数
score += W * float(senWord[str(i)])
# print "score = %f" % score
if senloc < len(senLoc) - 1:
# 判断该情感词与下一情感词之间是否有否定词或程度副词
# j为绝对位置
if senLoc[senloc] - senLoc[senloc + 1] > 1:
for j in range(senLoc[senloc]+1, senLoc[senloc + 1]):
# 如果有否定词
if j in notLoc:
W *= -1
# 如果有程度副词
elif j in degreeLoc:
W *= float(degreeWord[j])
else:
W = 1
# i定位至下一个情感词
if senloc < len(senLoc) - 1:
i = senLoc[senloc + 1]
return score
def getAnalysis():
return SentimentAnalysis('情感字典.txt', '否定词.txt', '副词.txt', '停用词.txt')
s = analysis.getAnalysis()
s.setSentence('句子')
#如果分数为正则为正面新闻
#如果位数为负则为负面新闻
print s.getScore()