形態素解析、単語分析、ワードクラウド

!pip install janome

!pip install wikipedia

!pip install japanize-matplotlib

!apt-get -y install fonts-ipafont-gothic

#文章の分かち書き

from janome.tokenizer import Tokenizer

t = Tokenizer()

bunsyou=""#@param{type:"string"}

for token in t.tokenize(bunsyou):

  print(token)

#Wikipediaからのテキスト読み込み

import wikipedia

keyword = ""#@param{type:"string"}

wikipedia.set_lang("ja")

search_response = wikipedia.search(keyword)

for sr in search_response:

  print(search_response.index(sr),sr)

keyword_index=#@param{type:"number"}

print('読み込みページ：',search_response[keyword_index])

page_data = wikipedia.page(search_response[keyword_index])

docs=page_data.content

docs

#単語の出現頻度の計算

from collections import Counter

import collections

import re

import matplotlib.pyplot as plt

import japanize_matplotlib 

word=[]

for token in t.tokenize(docs):

  hinshi='\u540D\u8A5E'#@param{type:"string"}

  if re.match(hinshi, token.part_of_speech):

    word.append(token.surface)

c_kosuu=30#@param{type:"integer"}

c=collections.Counter(word).most_common(c_kosuu)

values, counts = zip(*c)

print(values)

print(counts)

plt.bar(values,counts)

plt.show()

#単語の出現頻度の可視化（ワードクラウド）

from wordcloud import WordCloud

text=" ".join(word)

fpath = '/usr/share/fonts/truetype/fonts-japanese-gothic.ttf'

wordcloud = WordCloud(font_path=fpath,width=800, height=400).generate(text)

plt.figure(figsize=(40,30))

plt.imshow(wordcloud)

plt.axis("off")

plt.show()