中文词频统计

主要代码如下：
排序出高频词
# -*- coding: utf-8 -*-"""
from wordcloud 
import WordCloudimport matplotlib.pyplot as plt
import jieba
article = open('hlm.txt',encoding='UTF-8').read()
dele = {'。','！','？','】','“','”','（','）',' ','》','《','，'}
jieba.add_word('贾宝玉')
words = list(jieba.cut(article))
articleDict = {}
articleSet = set(words)-dele
for w in articleSet:    
if len(w)>1:        
articleDict[w] = words.count(w)
articlelist = sorted(articleDict.items(),key = lambda x:x[1], reverse = True)
cut_text = " ".join(words)
'print(cut_hlm)'
mywc = WordCloud().generate(cut_hlm)
plt.imshow(mywc)plt.axis("off")
plt.show()
'''
for i in range(20):    
print(articlelist[i])
import pandas as pd
pd.DataFrame(data=articlelist).to_csv('test.csv',encoding='UTF-8')
'''

红楼梦高频词：

红楼梦整篇小说：

成功排序出高频词：

生成词云如下：