Ch 04. 카운트 기반의 문서표현.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Chapter 4. 카운트 기반의 문서표현\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2 BOW 기반의 카운트 벡터 생성"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package movie_reviews to\n",
      "[nltk_data]     C:\\Users\\sk8er\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package movie_reviews is already up-to-date!\n",
      "[nltk_data] Downloading package punkt to\n",
      "[nltk_data]     C:\\Users\\sk8er\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package punkt is already up-to-date!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import nltk\n",
    "nltk.download('movie_reviews')\n",
    "nltk.download('punkt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#review count: 2000\n",
      "#samples of file ids: ['neg/cv000_29416.txt', 'neg/cv001_19502.txt', 'neg/cv002_17424.txt', 'neg/cv003_12683.txt', 'neg/cv004_12641.txt', 'neg/cv005_29357.txt', 'neg/cv006_17022.txt', 'neg/cv007_4992.txt', 'neg/cv008_29326.txt', 'neg/cv009_29417.txt']\n",
      "#categories of reviews: ['neg', 'pos']\n",
      "#num of \"neg\" reviews: 1000\n",
      "#num of \"pos\" reviews: 1000\n",
      "#id of the first review: neg/cv000_29416.txt\n",
      "#first review content:\n",
      " plot : two teen couples go to a church party , drink and then drive . \n",
      "they get into an accident . \n",
      "one of the guys dies , but his girlfriend continues to see him in her life , and has nightmares . \n",
      "w\n",
      "\n",
      "#sentence tokenization result: [['plot', ':', 'two', 'teen', 'couples', 'go', 'to', 'a', 'church', 'party', ',', 'drink', 'and', 'then', 'drive', '.'], ['they', 'get', 'into', 'an', 'accident', '.']]\n",
      "#word tokenization result: ['plot', ':', 'two', 'teen', 'couples', 'go', 'to', 'a', 'church', 'party', ',', 'drink', 'and', 'then', 'drive', '.', 'they', 'get', 'into', 'an']\n"
     ]
    }
   ],
   "source": [
    "from nltk.corpus import movie_reviews\n",
    "\n",
    "print('#review count:', len(movie_reviews.fileids())) #영화 리뷰 문서의 id를 반환\n",
    "print('#samples of file ids:', movie_reviews.fileids()[:10]) #id를 10개까지만 출력\n",
    "print('#categories of reviews:', movie_reviews.categories()) # label, 즉 긍정인지 부정인지에 대한 분류\n",
    "print('#num of \"neg\" reviews:', len(movie_reviews.fileids(categories='neg'))) #label이 부정인 문서들의 id를 반환\n",
    "print('#num of \"pos\" reviews:', len(movie_reviews.fileids(categories='pos'))) #label이 긍정인 문서들의 id를 반환\n",
    "fileid = movie_reviews.fileids()[0] #첫번째 문서의 id를 반환\n",
    "print('#id of the first review:', fileid)\n",
    "print('#first review content:\\n', movie_reviews.raw(fileid)[:200]) #첫번째 문서의 내용을 200자까지만 출력\n",
    "print()\n",
    "print('#sentence tokenization result:', movie_reviews.sents(fileid)[:2]) #첫번째 문서를 sentence tokenize한 결과 중 앞 두 문장\n",
    "print('#word tokenization result:', movie_reviews.words(fileid)[:20]) #첫번째 문서를 word tokenize한 결과 중 앞 스무 단어"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['plot', ':', 'two', 'teen', 'couples', 'go', 'to', 'a', 'church', 'party', ',', 'drink', 'and', 'then', 'drive', '.', 'they', 'get', 'into', 'an', 'accident', '.', 'one', 'of', 'the', 'guys', 'dies', ',', 'but', 'his', 'girlfriend', 'continues', 'to', 'see', 'him', 'in', 'her', 'life', ',', 'and', 'has', 'nightmares', '.', 'what', \"'\", 's', 'the', 'deal', '?', 'watch']\n"
     ]
    }
   ],
   "source": [
    "documents = [list(movie_reviews.words(fileid)) for fileid in movie_reviews.fileids()]\n",
    "print(documents[0][:50]) #첫째 문서의 앞 50개 단어를 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count of ',': 77717, count of 'the': 76529, count of '.': 65876, count of 'a': 38106, count of 'and': 35576, count of 'of': 34123, count of 'to': 31937, count of ''': 30585, count of 'is': 25195, count of 'in': 21822, "
     ]
    }
   ],
   "source": [
    "word_count = {}\n",
    "for text in documents:\n",
    "    for word in text:\n",
    "        word_count[word] = word_count.get(word, 0) + 1\n",
    "\n",
    "sorted_features = sorted(word_count, key=word_count.get, reverse=True)\n",
    "for word in sorted_features[:10]:\n",
    "    print(f\"count of '{word}': {word_count[word]}\", end=', ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "num of features: 43030\n",
      "count of 'film': 8935, count of 'one': 5791, count of 'movie': 5538, count of 'like': 3690, count of 'even': 2564, count of 'time': 2409, count of 'good': 2407, count of 'story': 2136, count of 'would': 2084, count of 'much': 2049, "
     ]
    }
   ],
   "source": [
    "from nltk.tokenize import RegexpTokenizer\n",
    "from nltk.corpus import stopwords #일반적으로 분석대상이 아닌 단어들\n",
    "\n",
    "tokenizer = RegexpTokenizer(\"[\\w']{3,}\") # 정규포현식으로 토크나이저를 정의\n",
    "english_stops = set(stopwords.words('english')) #영어 불용어를 가져옴\n",
    "\n",
    "#words() 대신 raw()를 이용해 원문을 가져옴\n",
    "documents = [movie_reviews.raw(fileid) for fileid in movie_reviews.fileids()] \n",
    "\n",
    "# stopwords의 적용과 토큰화를 동시에 수행.\n",
    "tokens = [[token for token in tokenizer.tokenize(doc) if token not in english_stops] for doc in documents]\n",
    "\n",
    "word_count = {}\n",
    "for text in tokens:\n",
    "    for word in text:\n",
    "        word_count[word] = word_count.get(word, 0) + 1\n",
    "\n",
    "sorted_features = sorted(word_count, key=word_count.get, reverse=True)\n",
    "\n",
    "print('num of features:', len(sorted_features))\n",
    "for word in sorted_features[:10]:\n",
    "    print(f\"count of '{word}': {word_count[word]}\", end=', ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_features = sorted_features[:1000] #빈도가 높은 상위 1000개의 단어만 추출하여 features를 구성"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 2, 0, 1, 0]\n"
     ]
    }
   ],
   "source": [
    "def document_features(document, word_features):\n",
    "    word_count = {}\n",
    "    for word in document: #document에 있는 단어들에 대해 빈도수를 먼저 계산\n",
    "        word_count[word] = word_count.get(word, 0) + 1\n",
    "        \n",
    "    features = []\n",
    "    for word in word_features: #word_features의 단어에 대해 계산된 빈도수를 feature에 추가\n",
    "        features.append(word_count.get(word, 0)) #빈도가 없는 단어는 0을 입력\n",
    "    return features\n",
    "\n",
    "word_features_ex = ['one', 'two', 'teen', 'couples', 'solo']\n",
    "doc_ex = ['two', 'two', 'couples']\n",
    "print(document_features(doc_ex, word_features_ex))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(film, 5), (one, 3), (movie, 6), (like, 3), (even, 3), (time, 0), (good, 2), (story, 0), (would, 1), (much, 0), (also, 1), (get, 3), (character, 1), (two, 2), (well, 1), (first, 0), (characters, 1), (see, 2), (way, 3), (make, 5), "
     ]
    }
   ],
   "source": [
    "feature_sets = [document_features(d, word_features) for d in tokens]\n",
    "\n",
    "# 첫째 feature set의 내용을 앞 20개만 word_features의 단어와 함께 출력\n",
    "for i in range(20):\n",
    "    print(f'({word_features[i]}, {feature_sets[0][i]})', end=', ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]\n"
     ]
    }
   ],
   "source": [
    "print(feature_sets[0][-20:]) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.3 사이킷런을 이용한 카운트 벡터 생성"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### CountVectorizer\n",
    "\n",
    "http://scikit-learn.org/stable/modules/feature_extraction.html#text-feature-extraction  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# data 준비, movie_reviews.raw()를 사용하여 raw text를 추출\n",
    "reviews = [movie_reviews.raw(fileid) for fileid in movie_reviews.fileids()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CountVectorizer(vocabulary=['film', 'one', 'movie', 'like', 'even', 'time',\n",
      "                            'good', 'story', 'would', 'much', 'also', 'get',\n",
      "                            'character', 'two', 'well', 'first', 'characters',\n",
      "                            'see', 'way', 'make', 'life', 'really', 'films',\n",
      "                            'plot', 'little', 'people', 'could', 'bad', 'scene',\n",
      "                            'never', ...])\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "\n",
    "#cv = CountVectorizer() #모든 매개변수에 디폴트 값을 사용하는 경우\n",
    "\n",
    "#앞에서 생성한 word_features를 이용하여 특성 집합을 지정하는 경우\n",
    "cv = CountVectorizer(vocabulary=word_features) \n",
    "\n",
    "#cv = CountVectorizer(max_features=1000) #특성 집합을 지정하지 않고 최대 특성의 수를 지정하는 경우\n",
    "print(cv) #객체에 사용된 인수들을 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['film' 'one' 'movie' 'like' 'even' 'time' 'good' 'story' 'would' 'much'\n",
      " 'also' 'get' 'character' 'two' 'well' 'first' 'characters' 'see' 'way'\n",
      " 'make']\n",
      "['film', 'one', 'movie', 'like', 'even', 'time', 'good', 'story', 'would', 'much', 'also', 'get', 'character', 'two', 'well', 'first', 'characters', 'see', 'way', 'make']\n"
     ]
    }
   ],
   "source": [
    "reviews_cv = cv.fit_transform(reviews) #reviews를 이용하여 count vector를 학습하고, 변환\n",
    "print(cv.get_feature_names_out()[:20]) # count vector에 사용된 feature 이름을 반환\n",
    "print(word_features[:20]) # 비교를 위해 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#type of count vectors: <class 'scipy.sparse.csr.csr_matrix'>\n",
      "#shape of count vectors: (2000, 1000)\n",
      "#sample of count vector:\n",
      "  (0, 0)\t6\n",
      "  (0, 1)\t3\n",
      "  (0, 2)\t6\n",
      "  (0, 3)\t3\n",
      "  (0, 4)\t3\n",
      "  (0, 6)\t2\n",
      "  (0, 8)\t1\n"
     ]
    }
   ],
   "source": [
    "print('#type of count vectors:', type(reviews_cv))\n",
    "print('#shape of count vectors:', reviews_cv.shape)\n",
    "print('#sample of count vector:')\n",
    "print(reviews_cv[0, :10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<2000x1000 sparse matrix of type '<class 'numpy.int64'>'\n",
       "\twith 252984 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews_cv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[5, 3, 6, 3, 3, 0, 2, 0, 1, 0, 1, 3, 1, 2, 1, 0, 1, 2, 3, 5]\n",
      "[6 3 6 3 3 0 2 0 1 0 1 3 2 2 1 0 1 2 3 5]\n"
     ]
    }
   ],
   "source": [
    "print(feature_sets[0][:20]) #절 앞에서 직접 계산한 카운트 벡터\n",
    "print(reviews_cv.toarray()[0, :20]) #변환된 결과의 첫째 feature set 중에서 앞 20개를 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "film:6, one:3, movie:6, like:3, even:3, time:0, good:2, story:0, would:1, much:0, also:1, get:3, character:2, two:2, well:1, first:0, characters:1, see:2, way:3, make:5, "
     ]
    }
   ],
   "source": [
    "for word, count in zip(cv.get_feature_names_out()[:20], reviews_cv[0].toarray()[0, :20]):\n",
    "    print(f'{word}:{count}', end=', ')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.4 한국어 텍스트의 카운트 벡터 변환\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>review</th>\n",
       "      <th>rating</th>\n",
       "      <th>date</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>돈 들인건 티가 나지만 보는 내내 하품만</td>\n",
       "      <td>1</td>\n",
       "      <td>2018.10.29</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>몰입할수밖에 없다. 어렵게 생각할 필요없다. 내가 전투에 참여한듯 손에 땀이남.</td>\n",
       "      <td>10</td>\n",
       "      <td>2018.10.26</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>이전 작품에 비해 더 화려하고 스케일도 커졌지만.... 전국 맛집의 음식들을 한데 ...</td>\n",
       "      <td>8</td>\n",
       "      <td>2018.10.24</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>이 정도면 볼만하다고 할 수 있음!</td>\n",
       "      <td>8</td>\n",
       "      <td>2018.10.22</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>재미있다</td>\n",
       "      <td>10</td>\n",
       "      <td>2018.10.20</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>나는 재밌게 봄</td>\n",
       "      <td>10</td>\n",
       "      <td>2018.10.14</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.5점은 줄 수 없냐?</td>\n",
       "      <td>0</td>\n",
       "      <td>2018.10.10</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>헐..다 죽었어....나중에 앤트맨 보다가도 깜놀...</td>\n",
       "      <td>10</td>\n",
       "      <td>2018.10.08</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>충격 결말</td>\n",
       "      <td>9</td>\n",
       "      <td>2018.10.06</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>응집력</td>\n",
       "      <td>8</td>\n",
       "      <td>2018.10.05</td>\n",
       "      <td>인피니티 워</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                              review  rating        date  \\\n",
       "0                             돈 들인건 티가 나지만 보는 내내 하품만       1  2018.10.29   \n",
       "1       몰입할수밖에 없다. 어렵게 생각할 필요없다. 내가 전투에 참여한듯 손에 땀이남.      10  2018.10.26   \n",
       "2  이전 작품에 비해 더 화려하고 스케일도 커졌지만.... 전국 맛집의 음식들을 한데 ...       8  2018.10.24   \n",
       "3                                이 정도면 볼만하다고 할 수 있음!       8  2018.10.22   \n",
       "4                                               재미있다      10  2018.10.20   \n",
       "5                                           나는 재밌게 봄      10  2018.10.14   \n",
       "6                                      0.5점은 줄 수 없냐?       0  2018.10.10   \n",
       "7                     헐..다 죽었어....나중에 앤트맨 보다가도 깜놀...      10  2018.10.08   \n",
       "8                                              충격 결말       9  2018.10.06   \n",
       "9                                                응집력       8  2018.10.05   \n",
       "\n",
       "    title  \n",
       "0  인피니티 워  \n",
       "1  인피니티 워  \n",
       "2  인피니티 워  \n",
       "3  인피니티 워  \n",
       "4  인피니티 워  \n",
       "5  인피니티 워  \n",
       "6  인피니티 워  \n",
       "7  인피니티 워  \n",
       "8  인피니티 워  \n",
       "9  인피니티 워  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "df = pd.read_csv('./data/daum_movie_review.csv')\n",
    "df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['10점' '18' '1987' '1도' '1점' '1점도' '2시간' '2시간이' '2편' '5점' '6점' '7점' '8점'\n",
      " 'cg' 'cg가' 'cg는' 'cg도' 'cg만' 'good' 'of' 'ㅋㅋ' 'ㅋㅋㅋ' 'ㅋㅋㅋㅋ' 'ㅎㅎ' 'ㅎㅎㅎ'\n",
      " 'ㅜㅜ' 'ㅠㅠ' 'ㅠㅠㅠ' 'ㅡㅡ' '가는' '가는줄' '가면' '가서' '가슴' '가슴아픈' '가슴이' '가장' '가족'\n",
      " '가족과' '가족들과' '가족의' '가족이' '가지고' '간만에' '갈수록' '감독' '감독님' '감독은' '감독의' '감독이'\n",
      " '감동' '감동과' '감동도' '감동은' '감동을' '감동이' '감동입니다' '감동적' '감동적이고' '감동적인' '감사드립니다'\n",
      " '감사합니다' '감정이' '갑자기' '갔는데' '갔다가' '강철비' '강추' '강추합니다' '같고' '같네요' '같다' '같습니다'\n",
      " '같아' '같아요' '같은' '같은데' '같음' '같이' '개연성' '개연성이' '개인적으로' '거의' '겁나' '것도' '것은'\n",
      " '것을' '것이' '것이다' '겨울왕국' '결국' '결말' '결말이' '계속' '고맙습니다' '곤지암' '공포' '공포를'\n",
      " '공포영화' '관객']\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "daum_cv = CountVectorizer(max_features=1000)\n",
    "\n",
    "daum_DTM = daum_cv.fit_transform(df.review) #review를 이용하여 count vector를 학습하고, 변환\n",
    "print(daum_cv.get_feature_names_out()[:100]) # count vector에 사용된 feature 이름을 반환"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#전체 형태소 결과: ['몰입', '할수밖에', '없다', '.', '어렵게', '생각', '할', '필요없다', '.', '내', '가', '전투', '에', '참여', '한', '듯', '손', '에', '땀', '이남', '.']\n",
      "#명사만 추출: ['몰입', '생각', '내', '전투', '참여', '듯', '손', '땀', '이남']\n",
      "#품사 태깅 결과 [('몰입', 'Noun'), ('할수밖에', 'Verb'), ('없다', 'Adjective'), ('.', 'Punctuation'), ('어렵게', 'Adjective'), ('생각', 'Noun'), ('할', 'Verb'), ('필요없다', 'Adjective'), ('.', 'Punctuation'), ('내', 'Noun'), ('가', 'Josa'), ('전투', 'Noun'), ('에', 'Josa'), ('참여', 'Noun'), ('한', 'Determiner'), ('듯', 'Noun'), ('손', 'Noun'), ('에', 'Josa'), ('땀', 'Noun'), ('이남', 'Noun'), ('.', 'Punctuation')]\n"
     ]
    }
   ],
   "source": [
    "from konlpy.tag import Okt #konlpy에서 Twitter 형태소 분석기를 import\n",
    "twitter_tag = Okt()\n",
    "\n",
    "print('#전체 형태소 결과:', twitter_tag.morphs(df.review[1]))\n",
    "print('#명사만 추출:', twitter_tag.nouns(df.review[1]))\n",
    "print('#품사 태깅 결과', twitter_tag.pos(df.review[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "나만의 토크나이저 결과: ['몰입', '할수밖에', '없다', '어렵게', '생각', '할', '필요없다', '내', '전투', '참여', '듯', '손', '땀', '이남']\n"
     ]
    }
   ],
   "source": [
    "def my_tokenizer(doc):\n",
    "    return [token for token, pos in twitter_tag.pos(doc) if pos in ['Noun', 'Verb', 'Adjective']]\n",
    "\n",
    "print(\"나만의 토크나이저 결과:\", my_tokenizer(df.review[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['가' '가는' '가는줄' '가면' '가서' '가슴' '가장' '가족' '가족영화' '가지' '가치' '각색' '간' '간다'\n",
      " '간만' '갈' '갈수록' '감' '감독' '감동' '감사' '감사합니다' '감상' '감성' '감정' '감탄' '갑자기' '갔는데'\n",
      " '갔다' '갔다가' '강' '강철' '강추' '같고' '같네요' '같다' '같습니다' '같아' '같아요' '같은' '같은데'\n",
      " '같음' '개' '개그' '개봉' '개연' '개인' '거' '거기' '거리' '거의' '걱정' '건' '건가' '건지' '걸'\n",
      " '겁니다' '것' '게' '겨울왕국' '결론' '결말' '경찰' '경험' '계속' '고' '고맙습니다' '고민' '고생' '곤지암'\n",
      " '곳' '공감' '공포' '공포영화' '과' '과거' '관' '관객' '관객수' '관람' '광주' '괜찮은' '교훈' '구성'\n",
      " '국내' '국민' '군인' '군함도' '굿' '권선' '귀신' '그' '그것' '그게' '그날' '그냥' '그닥' '그대로'\n",
      " '그때' '그래픽']\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "\n",
    "#토크나이저와 특성의 최대개수를 지정\n",
    "daum_cv = CountVectorizer(max_features=1000, tokenizer=my_tokenizer)\n",
    "#명사만 추출하고 싶은 경우에는 tokenizer에 'twitter_tag.nouns'를 바로 지정해도 됨\n",
    "\n",
    "daum_DTM = daum_cv.fit_transform(df.review) #review를 이용하여 count vector를 학습하고, 변환\n",
    "print(daum_cv.get_feature_names_out()[:100]) # count vector에 사용된 feature 이름을 반환"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<14725x1000 sparse matrix of type '<class 'numpy.int64'>'\n",
      "\twith 110800 stored elements in Compressed Sparse Row format>\n",
      "0.007524617996604414\n"
     ]
    }
   ],
   "source": [
    "print(repr(daum_DTM))\n",
    "print(110800/(14725*1000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "내 : 1, 듯 : 1, 몰입 : 1, 생각 : 1, 손 : 1, 없다 : 1, 할 : 1, "
     ]
    }
   ],
   "source": [
    "for word, count in zip(daum_cv.get_feature_names_out(), daum_DTM[1].toarray()[0]):\n",
    "    if count > 0:\n",
    "        print(word, ':', count, end=', ')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.5 카운트 벡터의 활용\n",
    "\n",
    "### 코사인 유사도(Cosine similarity)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl\n",
    "\n",
    "# 그래프에서 마이너스 폰트 깨지는 문제에 대한 대처\n",
    "mpl.rcParams['axes.unicode_minus'] = False\n",
    "\n",
    "import numpy as np\n",
    "x = np.arange(0,2*np.pi,0.1)   # start,stop,step\n",
    "y = np.cos(x)\n",
    "#print(x)\n",
    "plt.plot(x, y)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#대상 특성 행렬의 크기: (1, 1000)\n",
      "#유사도 계산 행렬의 크기: (1, 2000)\n",
      "#유사도 계산결과를 역순으로 정렬: [0.8367205630128807, 0.43817531290756406, 0.4080451370075411, 0.40727044884302327, 0.4060219836225451, 0.3999621981759778, 0.39965783997760135, 0.39566661804603703, 0.3945302295079114, 0.3911637170821695]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "\n",
    "start = len(reviews[0]) // 2 #첫째 리뷰의 문자수를 확인하고 뒤 절반을 가져오기 위해 중심점을 찾음\n",
    "source = reviews[0][-start:] #중심점으로부터 뒤 절반을 가져와서 비교할 문서를 생성\n",
    "\n",
    "source_cv = cv.transform([source]) #코사인 유사도는 카운트 벡터에 대해 계산하므로 벡터로 변환\n",
    "#transform은 반드시 리스트나 행렬 형태의 입력을 요구하므로 리스트로 만들어서 입력\n",
    "\n",
    "print(\"#대상 특성 행렬의 크기:\", source_cv.shape) #행렬의 크기를 확인, 문서가 하나이므로 (1, 1000)\n",
    "\n",
    "sim_result = cosine_similarity(source_cv, reviews_cv) #변환된 count vector와 기존 값들과의 similarity 계산\n",
    "\n",
    "print(\"#유사도 계산 행렬의 크기:\", sim_result.shape)\n",
    "print(\"#유사도 계산결과를 역순으로 정렬:\", sorted(sim_result[0], reverse=True)[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#가장 유사한 리뷰의 인덱스: 0\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "print('#가장 유사한 리뷰의 인덱스:', np.argmax(sim_result[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#가장 유사한 리뷰부터 정렬한 인덱스: [   0 1110 1570  687  628  112 1712 1393  524 1740]\n"
     ]
    }
   ],
   "source": [
    "print('#가장 유사한 리뷰부터 정렬한 인덱스:', (-sim_result[0]).argsort()[:10])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.6 TF-IDF로 성능을 높여보자"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#shape of tfidf matrix: (2000, 1000)\n",
      "#20 count score of the first review: [6 3 6 3 3 0 2 0 1 0 1 3 2 2 1 0 1 2 3 5]\n",
      "#20 tfidf score of the first review: [0.13556199 0.06700076 0.14998642 0.0772298  0.08608998 0.\n",
      " 0.0609124  0.         0.03126552 0.         0.03242315 0.09567082\n",
      " 0.06575035 0.06518293 0.03225625 0.         0.0345017  0.06863314\n",
      " 0.10042383 0.16727495]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "transformer = TfidfTransformer()\n",
    "transformer\n",
    "\n",
    "reviews_tfidf = transformer.fit_transform(reviews_cv)\n",
    "print('#shape of tfidf matrix:', reviews_tfidf.shape) #TF-IDF 행렬의 모양과 카운트 행렬의 모양이 일치하는 것을 확인\n",
    "\n",
    "#첫 리뷰의 카운트 벡터 중 앞 20개 값 출력\n",
    "print('#20 count score of the first review:', reviews_cv[0].toarray()[0][:20]) \n",
    "#첫 리뷰의 TF-IDF 벡터 중 앞 20개 값 출력\n",
    "print('#20 tfidf score of the first review:', reviews_tfidf[0].toarray()[0][:20])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#가장 유사한 리뷰의 인덱스: 0\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "tf = TfidfVectorizer(vocabulary=word_features)\n",
    "reviews_tf = tf.fit_transform(reviews)\n",
    "\n",
    "source_tf = tf.transform([source]) #코사인 유사도는 카운트 벡터에 대해 계산하므로 벡터로 변환\n",
    "#transform은 반드시 리스트나 행렬 형태의 입력을 요구하므로 리스트로 만들어서 입력\n",
    "\n",
    "sim_result_tf = cosine_similarity(source_tf, reviews_tf) #변환된 count vector와 기존 값들과의 similarity 계산\n",
    "\n",
    "print('#가장 유사한 리뷰의 인덱스:', np.argmax(sim_result_tf[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#카운트 벡터에 대해 가장 유사한 리뷰부터 정렬한 인덱스: [   0 1110 1570  687  628  112 1712 1393  524 1740]\n",
      "#TF-IDF 벡터에 대해 가장 유사한 리뷰부터 정렬한 인덱스: [   0 1110 1393 1570  645  323 1143  628 1676 1391]\n"
     ]
    }
   ],
   "source": [
    "print('#카운트 벡터에 대해 가장 유사한 리뷰부터 정렬한 인덱스:', (-sim_result[0]).argsort()[:10])\n",
    "print('#TF-IDF 벡터에 대해 가장 유사한 리뷰부터 정렬한 인덱스:', (-sim_result_tf[0]).argsort()[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": true,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}