Python批量处理中文文本并解析关键词

本文的使用环境在Ubuntu17.10、Pycharm 2017.3编辑器，以及Python3.6.3。
首先介绍几个处理文本的模块。

• Wordcloud, 可以将文本中的关键词总结并输出词云
  https://github.com/amueller/word_cloud
• jieba，像英语文章的单词那样将中文文本分成若干个词组
• docx，python处理doc、docx文本的模块
  http://python-docx.readthedocs.io/en/latest/index.html
• snownlp，可以对文本进行正负面情感分析取值
  https://github.com/isnowfy/snownlp
• pyLDAvis，直观地在默认网页显示关键词的可交互动态图
• textrank4zh，可以总结关键词出现的频率、关键词、关键句
  https://github.com/letiantian/TextRank4ZH

---

模块安装和导入

pip3 install wordcloud
pip3 install jieba
pip3 install python-docx
pip3 install snownlp
pip3 install pyLDAvis
pip3 install textRank4zh

读入文本

若是docx文本，以下为处理一个文本的代码，批量处理可用数组储存文件名再遍历

import docxfile=docx.Document("example.docx")#可修改文件名text=" "for para in file.paragraphs:    text+=para.text

若是单个txt文本，

text = "example.txt" #可以修改文件名with open(text,"r",encoding='gbk',errors='ignore') as f:    mytext=f.read()

生成词云

from os import pathfrom wordcloud import WordCloudimport osd = path.dirname(__file__)#该文件所在的文件夹路径font=os.path.join(os.path.dirname(__file__), "arialuni.ttf")#arialuni.ttf为字体的包# 读入文本mytext = "santi.txt"   #可更换文件名with open(mytext,"r",encoding='gbk',errors='ignore') as f:    text=f.read()# Generate a word cloud image加载词云wordcloud = WordCloud().generate(text)# Display the generated image:# the matplotlib way:import matplotlib.pyplot as pltplt.imshow(wordcloud)plt.axis("off")# lower max_font_sizewordcloud = WordCloud(font_path=font,max_font_size=40).generate(text)plt.figure()plt.imshow(wordcloud)plt.axis("off")plt.show()# The pil way (if you don't have matplotlib)#image = wordcloud.to_image()#image.show()

提取关键词和分类

from sklearn.feature_extraction.text import TfidfVectorizer,CountVectorizerfrom sklearn.decomposition import LatentDirichletAllocation#从文本中提取1000个最重要的特征关键词n_features = 1000tf_vectorizer = CountVectorizer(strip_accents = 'unicode',                                max_features=n_features,                                stop_words='english',                                max_df = 0.5,                                min_df = 10)tf = tf_vectorizer.fit_transform(df.content_cutted)#用LDA算法将主题分为5类n_topics=5lda = LatentDirichletAllocation(n_topics=n_topics,max_iter=50,                               learning_method='online',                               learning_offset=50,                               random_state=0)lda.fit(tf)#输出五个主题的前若干个关键词的函数定义def print_top_words(model,feature_names,n_top_words):    for topic_idx,topic in enumerate(model.components_):        print("Topic #%d:" % topic_idx)        print(" ".join([feature_names[i]                        for i in topic.argsort()[:-n_top_words -1 :-1]]))        print()#输出每个主题的前20个关键词n_top_words=20tf_feature_names=tf_vectorizer.get_feature_names()print_top_words(lda,tf_feature_names,n_top_words)#关键词和主题生成可交互的动态图import pyLDAvisimport pyLDAvis.sklearnpyLDAvis.enable_notebook()pyLDAvis.sklearn.prepare(lda, tf, tf_vectorizer)data = pyLDAvis.sklearn.prepare(lda, tf, tf_vectorizer)pyLDAvis.show(data)

统计关键词出现的频率、关键词组和摘要

可直接参考github https://github.com/letiantian/TextRank4ZH。

情感分析

可直接参考github https://github.com/isnowfy/snownlp。