I want to lead an easy life.

!apt install aptitude swig
!aptitude install mecab libmecab-dev mecab-ipadic-utf8 git make curl xz-utils file -y
!pip install mecab-python3 unidic-lite
!git clone --depth 1 https://github.com/neologd/mecab-ipadic-neologd.git
!echo yes | mecab-ipadic-neologd/bin/install-mecab-ipadic-neologd -n -a

import requests
import subprocess
import MeCab
from bs4 import BeautifulSoup
from time import sleep

# 本文をダウンロード
def novel_text_dler(url):
  headers = {
    'User-Agent':
    'Mozilla/5.0 (Windows NT 10.0; WOW64; Trident/7.0; rv:11.0) like Gecko'
  }
  r = requests.get(url, headers=headers)
  r.encoding = r.apparent_encoding
  soup =  BeautifulSoup(r.text)
  honbun = soup.find_all('div', class_='novel_view')
  novel = ''
  for text in honbun:
    novel += text.text
  sleep(1)
  return novel

# 形態素解析
def keitaiso(text):
  cmd = 'echo `mecab-config --dicdir`"/mecab-ipadic-neologd"'
  path = (subprocess.Popen(cmd, stdout=subprocess.PIPE, shell=True).communicate()[0]).decode('utf-8')
  tagger = MeCab.Tagger('-d {0}'.format(path))
  tagger.parse('')
  node = tagger.parseToNode(text)
  word = ''
  pre_feature = ''
  while node:
    # 名詞、形容詞、動詞、形容動詞であるか判定
    HANTEI = "名詞" in node.feature
    HANTEI = "形容詞" in node.feature or HANTEI
    HANTEI = "動詞" in node.feature or HANTEI
    HANTEI = "形容動詞" in node.feature or HANTEI
    # 以下に該当する場合は除外（ストップワード）
    HANTEI = (not "代名詞" in node.feature) and HANTEI
    HANTEI = (not "助動詞" in node.feature) and HANTEI
    HANTEI = (not "非自立" in node.feature) and HANTEI
    HANTEI = (not "数" in node.feature) and HANTEI
    HANTEI = (not "人名" in node.feature) and HANTEI
    if HANTEI:
      if ("名詞接続" in pre_feature and "名詞" in node.feature) or ("接尾" in node.feature):
        word += '{0}'.format(node.surface)
      else:
        word += ' {0}'.format(node.surface)
      #print('{0} {1}'.format(node.surface, node.feature))
    pre_feature = node.feature
    node = node.next
  return word[1:]

# 累計ランキングTop50の1から5話を取得
def novel_total_50():
  headers = {
    'User-Agent':
    'Mozilla/5.0 (Windows NT 10.0; WOW64; Trident/7.0; rv:11.0) like Gecko'
  }
  r = requests.get('http://yomou.syosetu.com/rank/list/type/total_total/', headers=headers)
  r.encoding = r.apparent_encoding
  soup =  BeautifulSoup(r.text)
  rank_index = soup.find_all('div', class_='rank_h')
  sleep(1)
  with open('drive/My Drive/Colab Notebooks/syosetu/novel_datas.txt', 'w') as f:
    for rank in range(50):
      link = rank_index[rank].find('a')
      url = link.get('href')
      # 後でわかりやすいようにURLと小説のタイトルを設定
      f.write(url + '\t' + link.get_text() + '\t')
      for chapter in range(5):
        f.write(keitaiso(novel_text_dler(url + str(chapter + 1) + '/')))
      f.write('\n')

novel_total_50()

from gensim.models.doc2vec import Doc2Vec, TaggedDocument

# 空白で単語を区切り、改行で文書を区切っているテキストデータ
with open('drive/My Drive/Colab Notebooks/syosetu/novel_datas.txt', 'r') as f:
  # 文書ごとに単語を分割してリストにする
  trainings = [TaggedDocument(words = data.split('\t')[2].split(), tags=[i]) for i, data in enumerate(f)]

# 学習の実行
m = Doc2Vec(documents=trainings, dm=1, vector_size=200, window=8, min_alpha=1e-4, min_count=5, sample=1e-3, workers=4, epochs=40)
# モデルを保存
m.save('drive/My Drive/Colab Notebooks/syosetu/doc2vec.model')

m = Doc2Vec.load('drive/My Drive/Colab Notebooks/syosetu/doc2vec.model')
# 0番目の小説に似ている小説は？（0番目の小説のタイトルは、「転生したらスライムだった件」）
print(m.docvecs.most_similar(0))

# 28番目の小説に似ている小説は？（28番目の小説のタイトルは、「公爵令嬢の嗜み」）
print(m.docvecs.most_similar(28))

# 魔法に似た単語は？
print(m.wv.most_similar('魔法'))

# スライムに似た単語は？
print(m.wv.most_similar('スライム'))

# 魔法に水を足す
print(m.wv.most_similar(positive=['魔法', '水']))
# 魔法から最強を引く
print(m.wv.most_similar(positive=['魔法'], negative=['最強']))

# 本来は小説の本文を形態素解析するべきだが、あえて少ない文字で検索してみる
x = m.infer_vector(keitaiso('主人公最弱 ヒロイン最強'))
print(m.docvecs.most_similar([x]))