brunch_dataset_processing.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "3695e29a",
   "metadata": {},
   "source": [
    "# 브런치 데이터셋 프로젝트\n",
    "\n",
    "수집한 데이터셋을 배포할 수 있도록 처리"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b6abd3e",
   "metadata": {},
   "source": [
    "## raw_dataset에 있는 파일 불러오기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "080a145b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "data_path =\"./raw_dataset\"\n",
    "file_list = os.listdir(data_path)\n",
    "\n",
    "file_path_list = []\n",
    "for file in file_list:\n",
    "    file_path_list.append(data_path + '/' + file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f1f86718",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import re\n",
    "\n",
    "df_dict = {}\n",
    "for file in file_path_list:\n",
    "    df = pd.read_csv(file)\n",
    "    keyword = re.search('./raw_dataset/(.+?)_dataset', file)\n",
    "    df_dict[keyword.group(1)] = df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d0cda5d",
   "metadata": {},
   "source": [
    "## 데이터 확인"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68f90927",
   "metadata": {},
   "source": [
    "### 수집에 사용한 키워드 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "71db18b7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "keywords = df_dict.keys()\n",
    "keywords = list(keywords)\n",
    "#keywords"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "d82ba52c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "118"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(keywords)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "95812853",
   "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>title</th>\n",
       "      <th>sub_title</th>\n",
       "      <th>body_text</th>\n",
       "      <th>keyword</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>내 몸이 나에게 말을 걸다.</td>\n",
       "      <td>&lt;중년의 진로수업&gt;</td>\n",
       "      <td>\\n요즘은 모든 것이 스마트. 아침에 눈을 떠서 핸드폰을 열면 어플이 정확한 수면시...</td>\n",
       "      <td>['중년', '건강', '일상']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>하체에서 전달되는 원인모를 통증 '좌골신경통'에 대해서</td>\n",
       "      <td>소리없이 찾아오는 불편한 소식 좌골신경통</td>\n",
       "      <td>\\n 어느 날 갑자기 관절주변에서 통증이 발생된다면 운동부족과 더불어 오랜시간 잘못...</td>\n",
       "      <td>['건강', '노인', '운동']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>아찔한 순간</td>\n",
       "      <td>구급차를 또 타게 되다니</td>\n",
       "      <td>\\n명절 연휴가 끝나고 일상으로 돌아가는 첫날인 오늘\\n나는 하루의 휴가가 더 있어...</td>\n",
       "      <td>['구급차', '건강', '반성']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>하루 한 권 독서</td>\n",
       "      <td>[당신은 뇌를 고칠 수 있다]-톰 브라이언</td>\n",
       "      <td>\\n지금 건강한 내가 5년 뒤에도 건강할 거라는 보장을 할 수 있을까? 건강에 대한...</td>\n",
       "      <td>['저자', '건강']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>정신병동에서 일하는 심리학자</td>\n",
       "      <td>심리선생님 만나는 날!</td>\n",
       "      <td>\\n글 내용에 포함되는 모든 사례는 새롭게 구성하고 재가공된 사례입니다. 상담자는 ...</td>\n",
       "      <td>['심리학', '건강', '감성에세이']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>폭식증! 이렇게 극복하세요</td>\n",
       "      <td>식이장애, 폭식습관 극복 할 수 있는 완벽한 방법</td>\n",
       "      <td>\\n폭식을 하면서 확실하게 느낀 게 있다. 허함을 채우려는 욕구, 먹으면 안 된다는...</td>\n",
       "      <td>['건강', '심리', '멘탈']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>도수치료 10회권을 끊었다</td>\n",
       "      <td>바른 자세의 중요성</td>\n",
       "      <td>\\n\\n처음에는 '잠을 이상하게 자서 그런가'라고 가벼이 생각했다. 그런데 시간이 ...</td>\n",
       "      <td>['건강', '건강관리', '생활건강']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>덤으로 주어진 하루</td>\n",
       "      <td>매너리즘의 극복</td>\n",
       "      <td>\\n“여전히 다른 사람들과 교류하는 건 힘드신가요?”\\n\\n“남들과 별로 이야기하고...</td>\n",
       "      <td>['에세이', '건강', '가족']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>지금 떠나도 나쁘지 않을 것 같아</td>\n",
       "      <td>방향 전환</td>\n",
       "      <td>\\n“나 지금 정말 맨 정신이야. 정말 진지하게 생각해 봤는데 지금까지 내가 살아온...</td>\n",
       "      <td>['에세이', '건강', '가족']</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>담당선생님을 바꾸겠습니다</td>\n",
       "      <td>귀 기울여야 할 대상</td>\n",
       "      <td>\\n다시 만난 담당의는 남편에게 다시 걷지 못할 것이라 말했던 사실조차 기억하지 못...</td>\n",
       "      <td>['에세이', '건강', '가족']</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                            title                    sub_title  \\\n",
       "0                내 몸이 나에게 말을 걸다.                    <중년의 진로수업>   \n",
       "1  하체에서 전달되는 원인모를 통증 '좌골신경통'에 대해서       소리없이 찾아오는 불편한 소식 좌골신경통   \n",
       "2                          아찔한 순간                구급차를 또 타게 되다니   \n",
       "3                       하루 한 권 독서      [당신은 뇌를 고칠 수 있다]-톰 브라이언   \n",
       "4                 정신병동에서 일하는 심리학자                 심리선생님 만나는 날!   \n",
       "5                  폭식증! 이렇게 극복하세요  식이장애, 폭식습관 극복 할 수 있는 완벽한 방법   \n",
       "6                  도수치료 10회권을 끊었다                   바른 자세의 중요성   \n",
       "7                      덤으로 주어진 하루                     매너리즘의 극복   \n",
       "8              지금 떠나도 나쁘지 않을 것 같아                        방향 전환   \n",
       "9                   담당선생님을 바꾸겠습니다                  귀 기울여야 할 대상   \n",
       "\n",
       "                                           body_text                 keyword  \n",
       "0  \\n요즘은 모든 것이 스마트. 아침에 눈을 떠서 핸드폰을 열면 어플이 정확한 수면시...      ['중년', '건강', '일상']  \n",
       "1  \\n 어느 날 갑자기 관절주변에서 통증이 발생된다면 운동부족과 더불어 오랜시간 잘못...      ['건강', '노인', '운동']  \n",
       "2  \\n명절 연휴가 끝나고 일상으로 돌아가는 첫날인 오늘\\n나는 하루의 휴가가 더 있어...     ['구급차', '건강', '반성']  \n",
       "3  \\n지금 건강한 내가 5년 뒤에도 건강할 거라는 보장을 할 수 있을까? 건강에 대한...            ['저자', '건강']  \n",
       "4  \\n글 내용에 포함되는 모든 사례는 새롭게 구성하고 재가공된 사례입니다. 상담자는 ...  ['심리학', '건강', '감성에세이']  \n",
       "5  \\n폭식을 하면서 확실하게 느낀 게 있다. 허함을 채우려는 욕구, 먹으면 안 된다는...      ['건강', '심리', '멘탈']  \n",
       "6  \\n\\n처음에는 '잠을 이상하게 자서 그런가'라고 가벼이 생각했다. 그런데 시간이 ...  ['건강', '건강관리', '생활건강']  \n",
       "7  \\n“여전히 다른 사람들과 교류하는 건 힘드신가요?”\\n\\n“남들과 별로 이야기하고...     ['에세이', '건강', '가족']  \n",
       "8  \\n“나 지금 정말 맨 정신이야. 정말 진지하게 생각해 봤는데 지금까지 내가 살아온...     ['에세이', '건강', '가족']  \n",
       "9  \\n다시 만난 담당의는 남편에게 다시 걷지 못할 것이라 말했던 사실조차 기억하지 못...     ['에세이', '건강', '가족']  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_dict['건강'].loc[range(10), ['title','sub_title', 'body_text', 'keyword']]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "648f4c31",
   "metadata": {},
   "source": [
    "### 전체 에세이 수 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a926262c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "전체 에세이 수: 1108528\n"
     ]
    }
   ],
   "source": [
    "num_essay = 0\n",
    "num_essay_tup = []\n",
    "for keyword in keywords:\n",
    "    #print(f\"{keyword}:{len(df_dict[keyword])}\")\n",
    "    num_essay_tup.append((keyword,len(df_dict[keyword])))\n",
    "    num_essay += len(df_dict[keyword])\n",
    "print(f\"전체 에세이 수: {num_essay}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5affad8c",
   "metadata": {},
   "source": [
    "### 키워드별 에세이 수 시각화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c7f87417",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x4000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import font_manager, rc\n",
    "\n",
    "font_path = \"C:\\Windows\\Fonts\\malgun.ttf\"\n",
    "font = font_manager.FontProperties(fname=font_path).get_name()\n",
    "rc('font', family=font)\n",
    "\n",
    "x = [tup[0] for tup in num_essay_tup]\n",
    "y = [tup[1] for tup in num_essay_tup]\n",
    "\n",
    "plt.figure(figsize=(10, 40))\n",
    "plt.barh(x[::-1],y[::-1])\n",
    "plt.savefig('keyword_num_essay.png', dpi=300)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a96c9a1",
   "metadata": {},
   "source": [
    "### 하나의 데이터프레임으로 합치기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "4d838c5f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_list = df_dict.values()\n",
    "total_df = pd.concat(df_list, ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "54741a80",
   "metadata": {
    "scrolled": true
   },
   "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>title</th>\n",
       "      <th>sub_title</th>\n",
       "      <th>body_text</th>\n",
       "      <th>keyword</th>\n",
       "      <th>likes</th>\n",
       "      <th>num_comments</th>\n",
       "      <th>post_date</th>\n",
       "      <th>post_url</th>\n",
       "      <th>author</th>\n",
       "      <th>author_id</th>\n",
       "      <th>author_belong</th>\n",
       "      <th>author_desc</th>\n",
       "      <th>num_subscription</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>컴퓨터 비전 기술 활용 AI개발 사례</td>\n",
       "      <td>스노우같은 셀카앱을 만들고 싶다면</td>\n",
       "      <td>\\n\\n\\n오늘 다뤄볼 주제는 컴퓨터 비전이라는 기술입니다.\\n\\n우리가 흔히 알고...</td>\n",
       "      <td>['컴퓨터비전', '비전', 'AI']</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@brunchk1wj/57</td>\n",
       "      <td>김성장</td>\n",
       "      <td>@brunchk1wj</td>\n",
       "      <td>CEO</td>\n",
       "      <td>AI 솔루션 스타트업, 디피니트 대표 김도환입니다.</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AI 어시스턴트의 보이스를 통한 이모티콘 활용 방법</td>\n",
       "      <td>Part 2. 탐구생활 #2</td>\n",
       "      <td>\\n현대 사회의 커뮤니케이션\\n\\n현대 사회는 스마트폰을 통해서 시공간적 제약의 한...</td>\n",
       "      <td>['이모티콘', 'AI', 'UX디자인']</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@soonkyujang/119</td>\n",
       "      <td>Qsigner</td>\n",
       "      <td>@soonkyujang</td>\n",
       "      <td>디자이너</td>\n",
       "      <td>일상의 소소한 소재를 탐구하고, 호기심을 상식으로 써내려가는 잡지식의 UX 디자이너.</td>\n",
       "      <td>2329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>AI기반 프레젠테이션 자동생성 솔루션 tome 소개</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n아래 글은 Generative Tech 스터디 1기에서 작성한 글입니다. Gen...</td>\n",
       "      <td>['프레젠테이션', '생산성', 'AI']</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@taewookim/75</td>\n",
       "      <td>TaeWoo Kim</td>\n",
       "      <td>@taewookim</td>\n",
       "      <td>CEO</td>\n",
       "      <td>내 지식과 경험을 수익으로 만드는 가장 쉬운 방법 에브리크리 (링크트리+ 내상품판매...</td>\n",
       "      <td>1384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>계산기같은 ChatGPT</td>\n",
       "      <td>검색을 넘어서는 원샷원킬일까?</td>\n",
       "      <td>\\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화...</td>\n",
       "      <td>['검색', 'AI', '계산기']</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@ioojoo/236</td>\n",
       "      <td>OOJOO</td>\n",
       "      <td>@ioojoo</td>\n",
       "      <td>출간작가</td>\n",
       "      <td>디지털 기술이 사회와 산업 그리고 기업과 개인에 가져다 주는 변화와 이에 대한 대응...</td>\n",
       "      <td>2248</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>AI로 이미지 만들기 (매우 쉬움)</td>\n",
       "      <td>Midjourney 사용 후기</td>\n",
       "      <td>\\n요즘 AI로 인터넷이 난리도 아니다. 특히 챗GPT. OpenAI사에서 신규로 ...</td>\n",
       "      <td>['AI', '인공지능', '챗봇']</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@khakiparka/32</td>\n",
       "      <td>카키파카</td>\n",
       "      <td>@khakiparka</td>\n",
       "      <td>에세이스트</td>\n",
       "      <td>Put on a parka and walk.</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108523</th>\n",
       "      <td>박지연</td>\n",
       "      <td>마녀를 만난 사회초년생</td>\n",
       "      <td>\\n지연은 운이 좋게도 졸업 직전에 취업에 성공했다. 대기업 계열사인 모 광고회사의...</td>\n",
       "      <td>['마녀', '사회생활', '회사']</td>\n",
       "      <td>1</td>\n",
       "      <td>19</td>\n",
       "      <td>2015-07-21</td>\n",
       "      <td>https://brunch.co.kr/@daljasee/20</td>\n",
       "      <td>고수리</td>\n",
       "      <td>@daljasee</td>\n",
       "      <td>에세이스트</td>\n",
       "      <td>책과 펜과 밤과 마음을 쓰는 에세이스트</td>\n",
       "      <td>19000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108524</th>\n",
       "      <td>나의 이직 경험기.</td>\n",
       "      <td>긴 글을 쓰는 건 어려운 일이네요.</td>\n",
       "      <td>\\n나는 세 곳의 직장에서 일을 했고, 네 번째 직장에서 일하고 있어요.\\n\\n나는...</td>\n",
       "      <td>['이직', '직장', '회사']</td>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>2015-07-04</td>\n",
       "      <td>https://brunch.co.kr/@daeheejang/6</td>\n",
       "      <td>장대희</td>\n",
       "      <td>@daeheejang</td>\n",
       "      <td>개발자</td>\n",
       "      <td>개발자</td>\n",
       "      <td>196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108525</th>\n",
       "      <td>아빠, 아버지.</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n내가 어릴적엔 아버지는 \\n힘세고, 무섭고, 과묵하고... \\n매일같이 일만하는...</td>\n",
       "      <td>['주말', '여행', '회사']</td>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>2015-07-01</td>\n",
       "      <td>https://brunch.co.kr/@legoilgi/4</td>\n",
       "      <td>레고일기 Legoilgi</td>\n",
       "      <td>@legoilgi</td>\n",
       "      <td>디자이너</td>\n",
       "      <td>레고로 일상을 기록하다.</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108526</th>\n",
       "      <td>창작자는 꼭 혼자여야 할까?</td>\n",
       "      <td>We-freekey와 Freekey.co 에 대한 블라블라</td>\n",
       "      <td>\\n스토리 제목은 \"팝아티스트의 회사 만들기(이전 제목)\"인데 실제 회사 만들기는 ...</td>\n",
       "      <td>['혼자여야', '회사', '창작자']</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2015-07-01</td>\n",
       "      <td>https://brunch.co.kr/@reeccu/3</td>\n",
       "      <td>릭킴 Rickkim</td>\n",
       "      <td>@reeccu</td>\n",
       "      <td>아티스트</td>\n",
       "      <td>내 안에 있는 무언가를 현실로 가져오고 싶어 발버둥치는 사람입니다. 이야기를 하고 ...</td>\n",
       "      <td>1101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108527</th>\n",
       "      <td>2015년 5월 3일의 나.</td>\n",
       "      <td>누군가의 시점에서 바라본 나. 릭의 첫 번째 인터뷰.</td>\n",
       "      <td>\\n지난번 인터뷰이 모집 시에 미리 신청을 받아 둔 팝아티스트 Rick 씨, 인터뷰...</td>\n",
       "      <td>['인터뷰', '회사', '직장']</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2015-06-25</td>\n",
       "      <td>https://brunch.co.kr/@reeccu/1</td>\n",
       "      <td>릭킴 Rickkim</td>\n",
       "      <td>@reeccu</td>\n",
       "      <td>아티스트</td>\n",
       "      <td>내 안에 있는 무언가를 현실로 가져오고 싶어 발버둥치는 사람입니다. 이야기를 하고 ...</td>\n",
       "      <td>1101</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1108528 rows × 13 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                title                         sub_title  \\\n",
       "0               컴퓨터 비전 기술 활용 AI개발 사례                 스노우같은 셀카앱을 만들고 싶다면   \n",
       "1        AI 어시스턴트의 보이스를 통한 이모티콘 활용 방법                   Part 2. 탐구생활 #2   \n",
       "2        AI기반 프레젠테이션 자동생성 솔루션 tome 소개                               NaN   \n",
       "3                       계산기같은 ChatGPT                  검색을 넘어서는 원샷원킬일까?   \n",
       "4                 AI로 이미지 만들기 (매우 쉬움)                  Midjourney 사용 후기   \n",
       "...                               ...                               ...   \n",
       "1108523                           박지연                      마녀를 만난 사회초년생   \n",
       "1108524                    나의 이직 경험기.               긴 글을 쓰는 건 어려운 일이네요.   \n",
       "1108525                      아빠, 아버지.                               NaN   \n",
       "1108526               창작자는 꼭 혼자여야 할까?  We-freekey와 Freekey.co 에 대한 블라블라   \n",
       "1108527               2015년 5월 3일의 나.     누군가의 시점에서 바라본 나. 릭의 첫 번째 인터뷰.   \n",
       "\n",
       "                                                 body_text  \\\n",
       "0        \\n\\n\\n오늘 다뤄볼 주제는 컴퓨터 비전이라는 기술입니다.\\n\\n우리가 흔히 알고...   \n",
       "1        \\n현대 사회의 커뮤니케이션\\n\\n현대 사회는 스마트폰을 통해서 시공간적 제약의 한...   \n",
       "2        \\n아래 글은 Generative Tech 스터디 1기에서 작성한 글입니다. Gen...   \n",
       "3        \\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화...   \n",
       "4        \\n요즘 AI로 인터넷이 난리도 아니다. 특히 챗GPT. OpenAI사에서 신규로 ...   \n",
       "...                                                    ...   \n",
       "1108523  \\n지연은 운이 좋게도 졸업 직전에 취업에 성공했다. 대기업 계열사인 모 광고회사의...   \n",
       "1108524  \\n나는 세 곳의 직장에서 일을 했고, 네 번째 직장에서 일하고 있어요.\\n\\n나는...   \n",
       "1108525  \\n내가 어릴적엔 아버지는 \\n힘세고, 무섭고, 과묵하고... \\n매일같이 일만하는...   \n",
       "1108526  \\n스토리 제목은 \"팝아티스트의 회사 만들기(이전 제목)\"인데 실제 회사 만들기는 ...   \n",
       "1108527  \\n지난번 인터뷰이 모집 시에 미리 신청을 받아 둔 팝아티스트 Rick 씨, 인터뷰...   \n",
       "\n",
       "                         keyword  likes  num_comments    post_date  \\\n",
       "0          ['컴퓨터비전', '비전', 'AI']      4             0   2023-01-25   \n",
       "1        ['이모티콘', 'AI', 'UX디자인']      8             0   2023-01-25   \n",
       "2        ['프레젠테이션', '생산성', 'AI']      7             0   2023-01-25   \n",
       "3            ['검색', 'AI', '계산기']     10             0   2023-01-25   \n",
       "4           ['AI', '인공지능', '챗봇']     15             1   2023-01-25   \n",
       "...                          ...    ...           ...          ...   \n",
       "1108523     ['마녀', '사회생활', '회사']      1            19   2015-07-21   \n",
       "1108524       ['이직', '직장', '회사']      1            12   2015-07-04   \n",
       "1108525       ['주말', '여행', '회사']      0            11   2015-07-01   \n",
       "1108526    ['혼자여야', '회사', '창작자']      1             2   2015-07-01   \n",
       "1108527      ['인터뷰', '회사', '직장']      2             3   2015-06-25   \n",
       "\n",
       "                                      post_url         author     author_id  \\\n",
       "0          https://brunch.co.kr/@brunchk1wj/57            김성장   @brunchk1wj   \n",
       "1        https://brunch.co.kr/@soonkyujang/119        Qsigner  @soonkyujang   \n",
       "2           https://brunch.co.kr/@taewookim/75     TaeWoo Kim    @taewookim   \n",
       "3             https://brunch.co.kr/@ioojoo/236          OOJOO       @ioojoo   \n",
       "4          https://brunch.co.kr/@khakiparka/32           카키파카   @khakiparka   \n",
       "...                                        ...            ...           ...   \n",
       "1108523      https://brunch.co.kr/@daljasee/20            고수리     @daljasee   \n",
       "1108524     https://brunch.co.kr/@daeheejang/6            장대희   @daeheejang   \n",
       "1108525       https://brunch.co.kr/@legoilgi/4  레고일기 Legoilgi     @legoilgi   \n",
       "1108526         https://brunch.co.kr/@reeccu/3     릭킴 Rickkim       @reeccu   \n",
       "1108527         https://brunch.co.kr/@reeccu/1     릭킴 Rickkim       @reeccu   \n",
       "\n",
       "        author_belong                                        author_desc  \\\n",
       "0                 CEO                       AI 솔루션 스타트업, 디피니트 대표 김도환입니다.   \n",
       "1                디자이너    일상의 소소한 소재를 탐구하고, 호기심을 상식으로 써내려가는 잡지식의 UX 디자이너.   \n",
       "2                 CEO  내 지식과 경험을 수익으로 만드는 가장 쉬운 방법 에브리크리 (링크트리+ 내상품판매...   \n",
       "3                출간작가  디지털 기술이 사회와 산업 그리고 기업과 개인에 가져다 주는 변화와 이에 대한 대응...   \n",
       "4               에세이스트                           Put on a parka and walk.   \n",
       "...               ...                                                ...   \n",
       "1108523         에세이스트                              책과 펜과 밤과 마음을 쓰는 에세이스트   \n",
       "1108524           개발자                                                개발자   \n",
       "1108525          디자이너                                     레고로 일상을 기록하다.    \n",
       "1108526          아티스트  내 안에 있는 무언가를 현실로 가져오고 싶어 발버둥치는 사람입니다. 이야기를 하고 ...   \n",
       "1108527          아티스트  내 안에 있는 무언가를 현실로 가져오고 싶어 발버둥치는 사람입니다. 이야기를 하고 ...   \n",
       "\n",
       "         num_subscription  \n",
       "0                      38  \n",
       "1                    2329  \n",
       "2                    1384  \n",
       "3                    2248  \n",
       "4                      10  \n",
       "...                   ...  \n",
       "1108523             19000  \n",
       "1108524               196  \n",
       "1108525                45  \n",
       "1108526              1101  \n",
       "1108527              1101  \n",
       "\n",
       "[1108528 rows x 13 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f01e657",
   "metadata": {},
   "source": [
    "### 데이터 프레임 속성 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "e9c5d5ce",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1108528 entries, 0 to 1108527\n",
      "Data columns (total 13 columns):\n",
      " #   Column            Non-Null Count    Dtype \n",
      "---  ------            --------------    ----- \n",
      " 0   title             1108488 non-null  object\n",
      " 1   sub_title         815452 non-null   object\n",
      " 2   body_text         1102783 non-null  object\n",
      " 3   keyword           1108528 non-null  object\n",
      " 4   likes             1108528 non-null  int64 \n",
      " 5   num_comments      1108528 non-null  int64 \n",
      " 6   post_date         1108528 non-null  object\n",
      " 7   post_url          1108528 non-null  object\n",
      " 8   author            1108408 non-null  object\n",
      " 9   author_id         1108528 non-null  object\n",
      " 10  author_belong     845850 non-null   object\n",
      " 11  author_desc       1107648 non-null  object\n",
      " 12  num_subscription  1108528 non-null  int64 \n",
      "dtypes: int64(3), object(10)\n",
      "memory usage: 109.9+ MB\n"
     ]
    }
   ],
   "source": [
    "total_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8af4c201",
   "metadata": {},
   "source": [
    "### 결측치 확인\n",
    "\n",
    "각 column의 결측치 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "0de13354",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "title                   40\n",
       "sub_title           293076\n",
       "body_text             5745\n",
       "keyword                  0\n",
       "likes                    0\n",
       "num_comments             0\n",
       "post_date                0\n",
       "post_url                 0\n",
       "author                 120\n",
       "author_id                0\n",
       "author_belong       262678\n",
       "author_desc            880\n",
       "num_subscription         0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a472ea1b",
   "metadata": {},
   "source": [
    "#### 결측치 제거하기\n",
    "* body_text가 null인 행\n",
    "* title이 null인 행"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "83e1fd69",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "title                    0\n",
       "sub_title           290979\n",
       "body_text                0\n",
       "keyword                  0\n",
       "likes                    0\n",
       "num_comments             0\n",
       "post_date                0\n",
       "post_url                 0\n",
       "author                 120\n",
       "author_id                0\n",
       "author_belong       261613\n",
       "author_desc            879\n",
       "num_subscription         0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_df = total_df.dropna(subset=['body_text'])\n",
    "total_df = total_df.dropna(subset=['title'])\n",
    "total_df.reset_index(inplace=True, drop=True) # 인덱스 초기화\n",
    "total_df.isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b497737a",
   "metadata": {},
   "source": [
    "### url 기준 중복 제거 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a5276f77",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1102743 entries, 0 to 1102742\n",
      "Data columns (total 13 columns):\n",
      " #   Column            Non-Null Count    Dtype \n",
      "---  ------            --------------    ----- \n",
      " 0   title             1102743 non-null  object\n",
      " 1   sub_title         811764 non-null   object\n",
      " 2   body_text         1102743 non-null  object\n",
      " 3   keyword           1102743 non-null  object\n",
      " 4   likes             1102743 non-null  int64 \n",
      " 5   num_comments      1102743 non-null  int64 \n",
      " 6   post_date         1102743 non-null  object\n",
      " 7   post_url          1102743 non-null  object\n",
      " 8   author            1102623 non-null  object\n",
      " 9   author_id         1102743 non-null  object\n",
      " 10  author_belong     841130 non-null   object\n",
      " 11  author_desc       1101864 non-null  object\n",
      " 12  num_subscription  1102743 non-null  int64 \n",
      "dtypes: int64(3), object(10)\n",
      "memory usage: 109.4+ MB\n"
     ]
    }
   ],
   "source": [
    "total_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "2b293b85",
   "metadata": {},
   "outputs": [],
   "source": [
    "# url 기준 중복 제거\n",
    "total_df = total_df.drop_duplicates(['post_url'])\n",
    "total_df.reset_index(inplace=True, drop=True) # 인덱스 초기화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "72e654fa",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 857807 entries, 0 to 857806\n",
      "Data columns (total 13 columns):\n",
      " #   Column            Non-Null Count   Dtype \n",
      "---  ------            --------------   ----- \n",
      " 0   title             857807 non-null  object\n",
      " 1   sub_title         629829 non-null  object\n",
      " 2   body_text         857807 non-null  object\n",
      " 3   keyword           857807 non-null  object\n",
      " 4   likes             857807 non-null  int64 \n",
      " 5   num_comments      857807 non-null  int64 \n",
      " 6   post_date         857807 non-null  object\n",
      " 7   post_url          857807 non-null  object\n",
      " 8   author            857717 non-null  object\n",
      " 9   author_id         857807 non-null  object\n",
      " 10  author_belong     652065 non-null  object\n",
      " 11  author_desc       857098 non-null  object\n",
      " 12  num_subscription  857807 non-null  int64 \n",
      "dtypes: int64(3), object(10)\n",
      "memory usage: 85.1+ MB\n"
     ]
    }
   ],
   "source": [
    "total_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2cd50b3",
   "metadata": {},
   "source": [
    "총 **857,807**개의 에세이 수집됨"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "36dcf583",
   "metadata": {},
   "source": [
    "### 좋아요(likes)가 10개 이상인 행만 추출\n",
    "\n",
    "사람들이 누른 좋아요(하트)는 글의 퀄리티를 나타내는 지표가 될 수 있음. \n",
    "\n",
    "고퀄리티의 에세이를 추출하기 위해 좋아요가 10개 이상인 글만 추출"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "id": "a0450e6e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "423967"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "high_quality_essay = total_df['likes'] >= 10\n",
    "high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "85c18c6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "hqe_df = total_df[high_quality_essay]\n",
    "hqe_df = hqe_df.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "c4bd879e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 423967 entries, 0 to 423966\n",
      "Data columns (total 14 columns):\n",
      " #   Column            Non-Null Count   Dtype \n",
      "---  ------            --------------   ----- \n",
      " 0   index             423967 non-null  int64 \n",
      " 1   title             423967 non-null  object\n",
      " 2   sub_title         311981 non-null  object\n",
      " 3   body_text         423967 non-null  object\n",
      " 4   keyword           423967 non-null  object\n",
      " 5   likes             423967 non-null  int64 \n",
      " 6   num_comments      423967 non-null  int64 \n",
      " 7   post_date         423967 non-null  object\n",
      " 8   post_url          423967 non-null  object\n",
      " 9   author            423894 non-null  object\n",
      " 10  author_id         423967 non-null  object\n",
      " 11  author_belong     342634 non-null  object\n",
      " 12  author_desc       423879 non-null  object\n",
      " 13  num_subscription  423967 non-null  int64 \n",
      "dtypes: int64(4), object(10)\n",
      "memory usage: 45.3+ MB\n"
     ]
    }
   ],
   "source": [
    "hqe_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a2f51cf2",
   "metadata": {},
   "source": [
    "### 필요한 컬럼만 선택하기\n",
    "\n",
    "에세이 생성을 위한 데이터셋에는\n",
    "\n",
    "* title\n",
    "* sub_title\n",
    "* body_text\n",
    "* keyword\n",
    "* likes\n",
    "* post_date\n",
    "* post_url\n",
    "* author_id\n",
    "\n",
    "컬럼만 포함시킬 것이다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "e1012ea8",
   "metadata": {
    "scrolled": true
   },
   "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>title</th>\n",
       "      <th>sub_title</th>\n",
       "      <th>body_text</th>\n",
       "      <th>keyword</th>\n",
       "      <th>likes</th>\n",
       "      <th>post_date</th>\n",
       "      <th>post_url</th>\n",
       "      <th>author_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>계산기같은 ChatGPT</td>\n",
       "      <td>검색을 넘어서는 원샷원킬일까?</td>\n",
       "      <td>\\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화...</td>\n",
       "      <td>['검색', 'AI', '계산기']</td>\n",
       "      <td>10</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@ioojoo/236</td>\n",
       "      <td>@ioojoo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>AI로 이미지 만들기 (매우 쉬움)</td>\n",
       "      <td>Midjourney 사용 후기</td>\n",
       "      <td>\\n요즘 AI로 인터넷이 난리도 아니다. 특히 챗GPT. OpenAI사에서 신규로 ...</td>\n",
       "      <td>['AI', '인공지능', '챗봇']</td>\n",
       "      <td>15</td>\n",
       "      <td>2023-01-25</td>\n",
       "      <td>https://brunch.co.kr/@khakiparka/32</td>\n",
       "      <td>@khakiparka</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>AI 시스템을 강화하는 휴먼인더루프(HITL)</td>\n",
       "      <td>'Human-In-The-Loop'의 정의와 빅테크기업의 솔루션을 설명</td>\n",
       "      <td>\\n이 글은 제가 삼성 SDS &lt; 인사이트 리포트 &gt; 2022년 11월 30일에 기...</td>\n",
       "      <td>['AI', '루프', '휴먼']</td>\n",
       "      <td>10</td>\n",
       "      <td>2023-01-23</td>\n",
       "      <td>https://brunch.co.kr/@ywkim36/102</td>\n",
       "      <td>@ywkim36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>AI로 할 수 있는 걸 평가할 때 던져 볼 만한 질문들</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\nAI는 잘할 수 있는 것도 많지만 여전히 한계도 있다. AI로 이런 것도 할 수...</td>\n",
       "      <td>['AI', '책이야기', 'IT']</td>\n",
       "      <td>11</td>\n",
       "      <td>2023-01-23</td>\n",
       "      <td>https://brunch.co.kr/@delight412/532</td>\n",
       "      <td>@delight412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>나는 생각한다. 고로 존재한다.</td>\n",
       "      <td>[10년 후 더 빛나는 책] 커넥톰 (승현준 지음)</td>\n",
       "      <td>\\n인류의 난제 중 하나는, 인간의 뇌가 동작하는 원리를 밝혀내는 것이다. 인간의 ...</td>\n",
       "      <td>['AI', '뇌과학', '과학기술']</td>\n",
       "      <td>12</td>\n",
       "      <td>2023-01-22</td>\n",
       "      <td>https://brunch.co.kr/@waybridge/88</td>\n",
       "      <td>@waybridge</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423962</th>\n",
       "      <td>14. 미지근한 피로</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n\\n\\n1. \\n\\n내가 태어났을 때, 엄마 뱃속을 헤집고 나오는 건 정말 피곤...</td>\n",
       "      <td>['피로', '자기계발', '회사']</td>\n",
       "      <td>29</td>\n",
       "      <td>2015-10-16</td>\n",
       "      <td>https://brunch.co.kr/@suhanjang/23</td>\n",
       "      <td>@suhanjang</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423963</th>\n",
       "      <td>11. '칼퇴' 없는 세상</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n\\n\\n1. \\n복도 쪽이 술렁였다. 사람들의 시선이 일제히 한 곳으로 쏠렸다....</td>\n",
       "      <td>['칼퇴', '관리', '회사']</td>\n",
       "      <td>29</td>\n",
       "      <td>2015-09-23</td>\n",
       "      <td>https://brunch.co.kr/@suhanjang/10</td>\n",
       "      <td>@suhanjang</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423964</th>\n",
       "      <td>10. 눈치보니즘</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n\\n\\n1.\\n입사한지 한 달 후 쯤의 일이다. \\n어쩌다 본사 신입사원 간담회...</td>\n",
       "      <td>['눈치', '신입', '회사']</td>\n",
       "      <td>36</td>\n",
       "      <td>2015-09-21</td>\n",
       "      <td>https://brunch.co.kr/@suhanjang/16</td>\n",
       "      <td>@suhanjang</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423965</th>\n",
       "      <td>07. 피고용자 사회</td>\n",
       "      <td>NaN</td>\n",
       "      <td>\\n\\n\\n1. 피고용자\\n\\n'피고용자(employee)'라는 말은 1940년대 ...</td>\n",
       "      <td>['회사', '고용', '자기계발']</td>\n",
       "      <td>35</td>\n",
       "      <td>2015-09-10</td>\n",
       "      <td>https://brunch.co.kr/@suhanjang/9</td>\n",
       "      <td>@suhanjang</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423966</th>\n",
       "      <td>프리랜서가 멋지다고? 웃기시네!</td>\n",
       "      <td>지극히 개인적인 프리랜서 생존기 (1) 나는 어떻게 세상 밖으로 나왔나</td>\n",
       "      <td>\\n2013년 11월 30일,\\n두번째로 적을 두었던 회사에서 나왔다.\\n\\n그리고...</td>\n",
       "      <td>['회사', '프리랜서']</td>\n",
       "      <td>10</td>\n",
       "      <td>2015-08-10</td>\n",
       "      <td>https://brunch.co.kr/@nergul01/2</td>\n",
       "      <td>@nergul01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>423967 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 title  \\\n",
       "0                        계산기같은 ChatGPT   \n",
       "1                  AI로 이미지 만들기 (매우 쉬움)   \n",
       "2            AI 시스템을 강화하는 휴먼인더루프(HITL)   \n",
       "3       AI로 할 수 있는 걸 평가할 때 던져 볼 만한 질문들   \n",
       "4                    나는 생각한다. 고로 존재한다.   \n",
       "...                                ...   \n",
       "423962                     14. 미지근한 피로   \n",
       "423963                  11. '칼퇴' 없는 세상   \n",
       "423964                       10. 눈치보니즘   \n",
       "423965                     07. 피고용자 사회   \n",
       "423966               프리랜서가 멋지다고? 웃기시네!   \n",
       "\n",
       "                                      sub_title  \\\n",
       "0                              검색을 넘어서는 원샷원킬일까?   \n",
       "1                              Midjourney 사용 후기   \n",
       "2       'Human-In-The-Loop'의 정의와 빅테크기업의 솔루션을 설명   \n",
       "3                                           NaN   \n",
       "4                  [10년 후 더 빛나는 책] 커넥톰 (승현준 지음)   \n",
       "...                                         ...   \n",
       "423962                                      NaN   \n",
       "423963                                      NaN   \n",
       "423964                                      NaN   \n",
       "423965                                      NaN   \n",
       "423966  지극히 개인적인 프리랜서 생존기 (1) 나는 어떻게 세상 밖으로 나왔나   \n",
       "\n",
       "                                                body_text  \\\n",
       "0       \\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화...   \n",
       "1       \\n요즘 AI로 인터넷이 난리도 아니다. 특히 챗GPT. OpenAI사에서 신규로 ...   \n",
       "2       \\n이 글은 제가 삼성 SDS < 인사이트 리포트 > 2022년 11월 30일에 기...   \n",
       "3       \\nAI는 잘할 수 있는 것도 많지만 여전히 한계도 있다. AI로 이런 것도 할 수...   \n",
       "4       \\n인류의 난제 중 하나는, 인간의 뇌가 동작하는 원리를 밝혀내는 것이다. 인간의 ...   \n",
       "...                                                   ...   \n",
       "423962  \\n\\n\\n1. \\n\\n내가 태어났을 때, 엄마 뱃속을 헤집고 나오는 건 정말 피곤...   \n",
       "423963  \\n\\n\\n1. \\n복도 쪽이 술렁였다. 사람들의 시선이 일제히 한 곳으로 쏠렸다....   \n",
       "423964  \\n\\n\\n1.\\n입사한지 한 달 후 쯤의 일이다. \\n어쩌다 본사 신입사원 간담회...   \n",
       "423965  \\n\\n\\n1. 피고용자\\n\\n'피고용자(employee)'라는 말은 1940년대 ...   \n",
       "423966  \\n2013년 11월 30일,\\n두번째로 적을 두었던 회사에서 나왔다.\\n\\n그리고...   \n",
       "\n",
       "                      keyword  likes    post_date  \\\n",
       "0         ['검색', 'AI', '계산기']     10   2023-01-25   \n",
       "1        ['AI', '인공지능', '챗봇']     15   2023-01-25   \n",
       "2          ['AI', '루프', '휴먼']     10   2023-01-23   \n",
       "3        ['AI', '책이야기', 'IT']     11   2023-01-23   \n",
       "4       ['AI', '뇌과학', '과학기술']     12   2023-01-22   \n",
       "...                       ...    ...          ...   \n",
       "423962   ['피로', '자기계발', '회사']     29   2015-10-16   \n",
       "423963     ['칼퇴', '관리', '회사']     29   2015-09-23   \n",
       "423964     ['눈치', '신입', '회사']     36   2015-09-21   \n",
       "423965   ['회사', '고용', '자기계발']     35   2015-09-10   \n",
       "423966         ['회사', '프리랜서']     10   2015-08-10   \n",
       "\n",
       "                                    post_url    author_id  \n",
       "0           https://brunch.co.kr/@ioojoo/236      @ioojoo  \n",
       "1        https://brunch.co.kr/@khakiparka/32  @khakiparka  \n",
       "2          https://brunch.co.kr/@ywkim36/102     @ywkim36  \n",
       "3       https://brunch.co.kr/@delight412/532  @delight412  \n",
       "4         https://brunch.co.kr/@waybridge/88   @waybridge  \n",
       "...                                      ...          ...  \n",
       "423962    https://brunch.co.kr/@suhanjang/23   @suhanjang  \n",
       "423963    https://brunch.co.kr/@suhanjang/10   @suhanjang  \n",
       "423964    https://brunch.co.kr/@suhanjang/16   @suhanjang  \n",
       "423965     https://brunch.co.kr/@suhanjang/9   @suhanjang  \n",
       "423966      https://brunch.co.kr/@nergul01/2    @nergul01  \n",
       "\n",
       "[423967 rows x 8 columns]"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hqe_df = hqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "hqe_df = hqe_df.reset_index()\n",
    "hqe_df = hqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "hqe_df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2bab473e",
   "metadata": {},
   "source": [
    "### 좋아요가 10 이상인 글 JSONL 형식으로 데이터 저장하기\n",
    "\n",
    "데이터 프레임의 각 row를 딕셔너리로 변환해 딕셔너리 리스트를 만든 후, jsonl로 변환"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "b88d9993",
   "metadata": {},
   "outputs": [],
   "source": [
    "hqe10_json_data = hqe_df.to_dict('records')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "f7df003c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'title': '계산기같은 ChatGPT',\n",
       " 'sub_title': '검색을 넘어서는 원샷원킬일까?',\n",
       " 'body_text': '\\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화형 지식AI인 ChatGPT.\\nhttps://v.daum.net/v/20230123093058969\\n과연 20년째 최고의 인터넷 경험인 검색을 대체까지 할 수 있을까?\\n난 단연코 아니라고 본다. 보완재라며 모를까.\\n\\nChatGPT는 20~30여분 검색어 입력 후 결과를 찾아 나서며 머리를 굴려 정리를 해서 답을 찾아가는 과정을 1분으로 줄여줄만큼 탐색→정리→요약을 기가 막히게 해주는 것은 사실이다. 게다가 같은 질문에 매번 조금씩 다른 답을 주기 때문에 좀 더 다양한 관점의 정보를 얻고자 하면 같은 질문을 다시 하면 된다. 게다가, 앞 질문에 이어서 추가 질문을 하면 그 내역을 기반으로 답을 주기 때문에 지속적으로 꼬리에 꼬리를 물며 질문을 던지면서 심층 답변을 찾아갈 수 있다.\\n그런 ChatGPT 사용도 한계가 있다.\\n1. 좋은 답변을 위해서는 제대로 된 질문을 해야 한다. 검색어 입력하는 것보다 더 자세하게 궁금한 것을 자세하게 문장으로 질문을 구성해야 한다. 한마디로 keyword가 아닌 prompt를 신경써서 쳐 넣어야 한다. 찾고 싶은 단어 하나를 입력하기는 쉽지만, 필요한 답을 찾기 위한 좋은 질문을 문장으로 구성하는 것은 쉽지 않다.\\n\\n2. 좋은 답변인지 알쏭달쏭하다는 것이다. 검색은 결과물들을 찾아 다니며 다양한 기사, 의견 그리고 댓글들을 보면서 여러 관점을 들여다볼 수 있다. 그 과정에서 옳고 그름 그리고 다양한 시각들이 있음을 깨닫고 내 생각 정리를 할 수 있다. 하지만, ChatGPT는 모범 답안처럼 보여지는 해설을 제시하므로 이걸 곧이 곧대로 받아들이면 오답을 진리로 받아들이는 착오에 빠질 수 있다.\\n\\n그것에 익숙해지면 우리도 앞으로 한계에 직면할 것이다.\\n계산기를 두드리면 원하는 답을 즉시 찾을 수 있지만 늘 계산기를 사용할 수 있는 것은 아니다. 대부분 우리 일상 속에서의 계산은 수 많은 상황에서 암산으로 즉시 계산해야 하는 경우가 대부분이다. 계산기에 의존하면 간단한 계산을 스스로 해낼 수 없다. 산수를 배우는 이유가 있다. 검색은 산수와 같고, 암산과 같다. ChatGPT에 의존하면 나만의 공식을 가질 수 없다. 기계가 토해내는 답을 진리로 알고 다른 생각을 펼칠 수 없고, 여러 의견이 있을 수 있음을 망각할 수 있다. 그 과정에서 내 생각도 자리 잡을 수 있는 기회를 잃게 된다.\\n\\n그래서, ChatGPT는 현재의 대화 방식의 인터페이스로는 온전한 서비스가 될 수 없다.\\n기존의 검색에 보완제로서 연계되거나 특정 도메인에서만 제한적으로 사용될 것이다. 이미 https://you.com과 같은 검색과의 통합 UI로 선보이기도 하고, MS의 bing에 ChatGPT가 연계하는 것을 고려하고 있기도 하다. 또, 아마도 구글 역시LAMDA2를 검색엔진에 활용하거나 별도의 서비스 영역에 도입하는 것을 고려할 것이다.\\n☞ 구글의 Sparrow\\nhttps://www.aitimes.com/news/articleView.html?idxno=149050\\n\\n쇼핑할 때, 판매원이 옆에 따라 다니면서 필요한 것을 즉시 제시하는 것도 때론 유용하지만.. 대개는 그냥 이것저것 만져보고 보면서 혼자 살펴보다가 판매원의 도움을 받는 것이 더 나은 것처럼, 검색하다가 필요로 할 때 AI가 자연스럽게 도와주는 것이 훨씬 더 낫지 않을까.',\n",
       " 'keyword': \"['검색', 'AI', '계산기']\",\n",
       " 'likes': 10,\n",
       " 'post_date': ' 2023-01-25',\n",
       " 'post_url': 'https://brunch.co.kr/@ioojoo/236',\n",
       " 'author_id': '@ioojoo'}"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hqe10_json_data[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e9498dc9",
   "metadata": {},
   "source": [
    "keyword가 리스트가 아닌 문자열로 저장되었다. 이를 리스트로 변환하여 저장할 필요가 있다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "9eea3bf9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 문자열로 저장된 리스트를 파이썬 리스트 타입으로 변환해주는 함수\n",
    "def str_to_list(cell):\n",
    "    cell = ''.join(c for c in cell if c not in \"'[]\")\n",
    "    cell = cell.split(', ')\n",
    "    return cell"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "bad6d13d",
   "metadata": {},
   "outputs": [],
   "source": [
    "for json in hqe10_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "bcaa92d8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'title': '계산기같은 ChatGPT',\n",
       " 'sub_title': '검색을 넘어서는 원샷원킬일까?',\n",
       " 'body_text': '\\n구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화형 지식AI인 ChatGPT.\\nhttps://v.daum.net/v/20230123093058969\\n과연 20년째 최고의 인터넷 경험인 검색을 대체까지 할 수 있을까?\\n난 단연코 아니라고 본다. 보완재라며 모를까.\\n\\nChatGPT는 20~30여분 검색어 입력 후 결과를 찾아 나서며 머리를 굴려 정리를 해서 답을 찾아가는 과정을 1분으로 줄여줄만큼 탐색→정리→요약을 기가 막히게 해주는 것은 사실이다. 게다가 같은 질문에 매번 조금씩 다른 답을 주기 때문에 좀 더 다양한 관점의 정보를 얻고자 하면 같은 질문을 다시 하면 된다. 게다가, 앞 질문에 이어서 추가 질문을 하면 그 내역을 기반으로 답을 주기 때문에 지속적으로 꼬리에 꼬리를 물며 질문을 던지면서 심층 답변을 찾아갈 수 있다.\\n그런 ChatGPT 사용도 한계가 있다.\\n1. 좋은 답변을 위해서는 제대로 된 질문을 해야 한다. 검색어 입력하는 것보다 더 자세하게 궁금한 것을 자세하게 문장으로 질문을 구성해야 한다. 한마디로 keyword가 아닌 prompt를 신경써서 쳐 넣어야 한다. 찾고 싶은 단어 하나를 입력하기는 쉽지만, 필요한 답을 찾기 위한 좋은 질문을 문장으로 구성하는 것은 쉽지 않다.\\n\\n2. 좋은 답변인지 알쏭달쏭하다는 것이다. 검색은 결과물들을 찾아 다니며 다양한 기사, 의견 그리고 댓글들을 보면서 여러 관점을 들여다볼 수 있다. 그 과정에서 옳고 그름 그리고 다양한 시각들이 있음을 깨닫고 내 생각 정리를 할 수 있다. 하지만, ChatGPT는 모범 답안처럼 보여지는 해설을 제시하므로 이걸 곧이 곧대로 받아들이면 오답을 진리로 받아들이는 착오에 빠질 수 있다.\\n\\n그것에 익숙해지면 우리도 앞으로 한계에 직면할 것이다.\\n계산기를 두드리면 원하는 답을 즉시 찾을 수 있지만 늘 계산기를 사용할 수 있는 것은 아니다. 대부분 우리 일상 속에서의 계산은 수 많은 상황에서 암산으로 즉시 계산해야 하는 경우가 대부분이다. 계산기에 의존하면 간단한 계산을 스스로 해낼 수 없다. 산수를 배우는 이유가 있다. 검색은 산수와 같고, 암산과 같다. ChatGPT에 의존하면 나만의 공식을 가질 수 없다. 기계가 토해내는 답을 진리로 알고 다른 생각을 펼칠 수 없고, 여러 의견이 있을 수 있음을 망각할 수 있다. 그 과정에서 내 생각도 자리 잡을 수 있는 기회를 잃게 된다.\\n\\n그래서, ChatGPT는 현재의 대화 방식의 인터페이스로는 온전한 서비스가 될 수 없다.\\n기존의 검색에 보완제로서 연계되거나 특정 도메인에서만 제한적으로 사용될 것이다. 이미 https://you.com과 같은 검색과의 통합 UI로 선보이기도 하고, MS의 bing에 ChatGPT가 연계하는 것을 고려하고 있기도 하다. 또, 아마도 구글 역시LAMDA2를 검색엔진에 활용하거나 별도의 서비스 영역에 도입하는 것을 고려할 것이다.\\n☞ 구글의 Sparrow\\nhttps://www.aitimes.com/news/articleView.html?idxno=149050\\n\\n쇼핑할 때, 판매원이 옆에 따라 다니면서 필요한 것을 즉시 제시하는 것도 때론 유용하지만.. 대개는 그냥 이것저것 만져보고 보면서 혼자 살펴보다가 판매원의 도움을 받는 것이 더 나은 것처럼, 검색하다가 필요로 할 때 AI가 자연스럽게 도와주는 것이 훨씬 더 낫지 않을까.',\n",
       " 'keyword': ['검색', 'AI', '계산기'],\n",
       " 'likes': 10,\n",
       " 'post_date': ' 2023-01-25',\n",
       " 'post_url': 'https://brunch.co.kr/@ioojoo/236',\n",
       " 'author_id': '@ioojoo'}"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hqe10_json_data[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a67443b4",
   "metadata": {},
   "source": [
    "리스트로 저장되었다.\n",
    "\n",
    "현재 hqe10_json_data는 딕셔너리의 리스트이다. \n",
    "\n",
    "하나의 엘리먼트를 한 줄에 저장하는 **jsonl** 형식으로 최종데이터셋을 저장하자.\n",
    "\n",
    "jsonl로 저장하기 위해선 먼저 dict 객체를 dumps를 통해 json객체로 변환한 후, 이를 writelines()를 통해 써주면 된다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "ca84067b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "423967"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(hqe10_json_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "id": "135024b9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict"
      ]
     },
     "execution_count": 133,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(hqe10_json_data[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0809ddf0",
   "metadata": {},
   "source": [
    "형식에 맞게 잘 저장되고 잘 불러올 수 있는지 확인하기 위해 5개로 테스트.\n",
    "\n",
    "참고로 jsonl을 읽어오기 위해서는 jsonlines 라이브러리를 설치해야 한다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "50025b8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "with open(\"brunch_dataset_10_sample.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe10_json_data[:5]:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "69d2d381",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting jsonlines\n",
      "  Downloading jsonlines-3.1.0-py3-none-any.whl (8.6 kB)\n",
      "Requirement already satisfied: attrs>=19.2.0 in c:\\users\\docdoca\\appdata\\local\\programs\\python\\python311\\lib\\site-packages (from jsonlines) (22.2.0)\n",
      "Installing collected packages: jsonlines\n",
      "Successfully installed jsonlines-3.1.0\n"
     ]
    }
   ],
   "source": [
    "!pip install jsonlines"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "904734fd",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "계산기같은 ChatGPT\n",
      "\n",
      "구글마저 긴장케할 정도로 파죽지세로 뜨거운 감자가 되고 있는 OpenAI의 대화형 지식AI인 ChatGPT.\n",
      "https://v.daum.net/v/20230123093058969\n",
      "과연 20년째 최고의 인터넷 경험인 검색을 대체까지 할 수 있을까?\n",
      "난 단연코 아니라고 본다. 보완재라며 모를까.\n",
      "\n",
      "ChatGPT는 20~30여분 검색어 입력 후 결과를 찾아 나서며 머리를 굴려 정리를 해서 답을 찾아가는 과정을 1분으로 줄여줄만큼 탐색→정리→요약을 기가 막히게 해주는 것은 사실이다. 게다가 같은 질문에 매번 조금씩 다른 답을 주기 때문에 좀 더 다양한 관점의 정보를 얻고자 하면 같은 질문을 다시 하면 된다. 게다가, 앞 질문에 이어서 추가 질문을 하면 그 내역을 기반으로 답을 주기 때문에 지속적으로 꼬리에 꼬리를 물며 질문을 던지면서 심층 답변을 찾아갈 수 있다.\n",
      "그런 ChatGPT 사용도 한계가 있다.\n",
      "1. 좋은 답변을 위해서는 제대로 된 질문을 해야 한다. 검색어 입력하는 것보다 더 자세하게 궁금한 것을 자세하게 문장으로 질문을 구성해야 한다. 한마디로 keyword가 아닌 prompt를 신경써서 쳐 넣어야 한다. 찾고 싶은 단어 하나를 입력하기는 쉽지만, 필요한 답을 찾기 위한 좋은 질문을 문장으로 구성하는 것은 쉽지 않다.\n",
      "\n",
      "2. 좋은 답변인지 알쏭달쏭하다는 것이다. 검색은 결과물들을 찾아 다니며 다양한 기사, 의견 그리고 댓글들을 보면서 여러 관점을 들여다볼 수 있다. 그 과정에서 옳고 그름 그리고 다양한 시각들이 있음을 깨닫고 내 생각 정리를 할 수 있다. 하지만, ChatGPT는 모범 답안처럼 보여지는 해설을 제시하므로 이걸 곧이 곧대로 받아들이면 오답을 진리로 받아들이는 착오에 빠질 수 있다.\n",
      "\n",
      "그것에 익숙해지면 우리도 앞으로 한계에 직면할 것이다.\n",
      "계산기를 두드리면 원하는 답을 즉시 찾을 수 있지만 늘 계산기를 사용할 수 있는 것은 아니다. 대부분 우리 일상 속에서의 계산은 수 많은 상황에서 암산으로 즉시 계산해야 하는 경우가 대부분이다. 계산기에 의존하면 간단한 계산을 스스로 해낼 수 없다. 산수를 배우는 이유가 있다. 검색은 산수와 같고, 암산과 같다. ChatGPT에 의존하면 나만의 공식을 가질 수 없다. 기계가 토해내는 답을 진리로 알고 다른 생각을 펼칠 수 없고, 여러 의견이 있을 수 있음을 망각할 수 있다. 그 과정에서 내 생각도 자리 잡을 수 있는 기회를 잃게 된다.\n",
      "\n",
      "그래서, ChatGPT는 현재의 대화 방식의 인터페이스로는 온전한 서비스가 될 수 없다.\n",
      "기존의 검색에 보완제로서 연계되거나 특정 도메인에서만 제한적으로 사용될 것이다. 이미 https://you.com과 같은 검색과의 통합 UI로 선보이기도 하고, MS의 bing에 ChatGPT가 연계하는 것을 고려하고 있기도 하다. 또, 아마도 구글 역시LAMDA2를 검색엔진에 활용하거나 별도의 서비스 영역에 도입하는 것을 고려할 것이다.\n",
      "☞ 구글의 Sparrow\n",
      "https://www.aitimes.com/news/articleView.html?idxno=149050\n",
      "\n",
      "쇼핑할 때, 판매원이 옆에 따라 다니면서 필요한 것을 즉시 제시하는 것도 때론 유용하지만.. 대개는 그냥 이것저것 만져보고 보면서 혼자 살펴보다가 판매원의 도움을 받는 것이 더 나은 것처럼, 검색하다가 필요로 할 때 AI가 자연스럽게 도와주는 것이 훨씬 더 낫지 않을까.\n",
      "['검색', 'AI', '계산기']\n",
      "\n",
      "\n",
      "--------------------------------------------------\n",
      "\n",
      "\n",
      "AI로 이미지 만들기 (매우 쉬움)\n",
      "\n",
      "요즘 AI로 인터넷이 난리도 아니다. 특히 챗GPT. OpenAI사에서 신규로 론칭한 AI 챗봇 프로그램인데 기세가 심상치 않다. 뼛속까지 문과인 나조차 AI분야에 관심을 갖게 만드는 공상과학과 유사한 엄청난 기술을 자랑한다. 프로토타입인 데다 트래픽이 많아 직접 사용해보진 못했으나 링크드인이나 유튜브를 통해 챗GPT의 능력을 쉽게 접할 수 있다. \"언어\"로 가능한 모든 영역을 수행할 수 있을뿐더러 코딩까지 대신 써준다. 예를 들어, 천만 관객용 영화의 플롯과 인물 구성을 써달라고 요청하면 몇 분 내로 완벽한 플롯과 인물 구성을 내놓는다. 뉴스 기사, 시나리오, 산문, 시, 소네트, 심지어는 마케팅 전략까지 써준다. 인터넷에 업로드되어 있는 무한에 가까운 정보 풀과 고도화된 머신러닝의 결합물이라고 볼 수 있다. \n",
      "\n",
      "어떤 질문에도 답을 내놓아 구글을 대체할 것이라는 목소리도 나온다. 구글은 사용자가 직접 정보를 모아 분석해서 답을 도출해야 하지만 챗GPT는 중간 과정을 생략하고 질문 하나로 바로 해답에 다다를 수 있게 해 준다. 잠재력을 인정받아 마이크로소프트로부터 막대한 투자를 여러 차례 받고 있기도 하다. (1월 24일 기준 1조 원 넘게 투자한 것으로 밝혀졌다.) 해당 프로그램에 대한 정보를 찾다 보면 놀라움의 연속이 아닐 수 없다. 하지만 걷잡을 수 없이 빠르게 진화하는 기술로 인한 불안과 공포도 분명 존재한다. 확실한 건 인류가 수십 년간 그려왔던 미래가 껑충 앞으로 다가와 더 선명해졌다는 것이다. AI의 세계에 몇 시간 정신을 팔다 보면 실제로 미래에 와 있다는 생각이 들 정도다. \n",
      "\n",
      "최근 ChatGPT의 인기로 인해 나의 감각도 AI를 향해 뻗어있다. 그래서 그런지, AI관련 기술에 대한 기사나 SNS 포스팅, 유튜브 영상을 많이 찾아보는 요즘이다. (알고리즘에 의해 의도적으로 노출되고 있는 거겠지만.) 그러던 중 며칠 전 한 링크드인 포스팅에서 또 다른 AI툴을 발견했다. 툴의 이름은 바로 Midjourney, 이미지를 만들어주는 프로그램이다. 구글 서칭 후 미드저니 웹사이트로 들어가 사용법을 익힌 후 몇 가지 명령을 시도해 보았다. \n",
      "\n",
      "먼저, 사용안내에 따라 미드저니의 디스코드 채널에 접속해 회원가입을 한 후 초청을 받는다. 그다음 접속자들의 놀이터 같은 형식의 세부 채널로 들어가 채팅창에 이미지 생성 명령을 내린다. 사용료는 없다. \n",
      "\n",
      "/imagine prompt [        ] \n",
      "\n",
      "[        ] 안에 생성하고 싶은 이미지의 \"요소\"를 영어로 입력하면 된다. 먼저 친구의 학교 과제를 도와줘 보기로 했다. \n",
      "\n",
      "조형물, 핑크색, 부드러운 질감, 나무를 입력해 보았다.\n",
      "단 몇 초 만에 위 네 개의 이미지를 생성해 냈다. 포함시킨 요소가 모두 적용된 것을 볼 수 있다. 하나의 이미지만 생성하는 것이 아니라 한 번에 네 개의 옵션 이미지를 내놓으며 마음에 들거나 변형시키고픈 이미지가 있으면 하나를 골라 추가 명령을 내릴 수 있다. 추가 명령은 두 가지로, \"업그레이드\"를 선택하면 해상도나 퀄리티를 높일 수 있고 \"변형\"을 선택하면 전체적인 느낌은 같으나 추가적인 요소를 덧붙일 수 있다. \n",
      "\n",
      "다음으로는, 이유는 모르겠으나 아래와 같은 특징을 입력해 보았다. \n",
      "\n",
      "젊은 남교수, 화창한 오후, 책상에 착석, 책장, 운동한 몸, 짧은 머리, 카메라 응시, 하얀 티셔츠, 태닝 된 피부.\n",
      ".\n",
      ".\n",
      ".\n",
      "이번에도 1분도 걸리지 않아 네 가지 옵션의 이미지를 생성해 주었다. 너무 현실감이 있어 이미지라기보다는 \"사진\"에 가까울 정도다. (특히, 오른쪽 상단 옵션의 경우)\n",
      "\n",
      "한번 이미지 생성 후 네 개중 하나를 선택해 업그레이드도 시켜보았다. \n",
      "입력 값: 젊은 나이의 형제, 오랜만의 만남, 동생보다 살짝 작은 형, 웃음 가득한 얼굴, 겨울, 캐나다 배경\n",
      "아주 훈훈한 형제를 만들어 줬다.\n",
      "\n",
      "시간 가는 줄 모르고 10번도 넘게 시도해 보니 약간의 노하우가 생겼다. \"리얼리스틱하게\", \"4K\", \"8K\" 등 퀄리티를 높이는 입력 값을 포함하면 더욱 사실적인 결과물을 얻을 수 있다. 다른 접속자들의 명령어도 실시간으로 모니터 할 수 있어 참고가 가능하다. 예를 들어, 어떤 사람은 카메라의 모델, 렌즈 두께, 조명 효과까지 세세히 입력하고 있었다. \n",
      "\n",
      "최대한 구체적으로 입력하면 할수록 더 좋은 결과물이 나온다. AI가 나의 명령어를 꼼꼼히 읽고 고민한다고 생각하고 사소한 부분이더라도 모두 입력하는 게 좋다. 요소마다 쉼표를 붙여가며 입력하지 않아도 되고 문장으로 묘사를 해도 된다. (마치 소설 속 인물 묘사처럼) 정해진 룰이 없어 굉장히 편했다. (처음에 /imagine prompt만 입력한 후에는 자유롭게 입력이 가능하다.) \n",
      "\n",
      "웃긴 건 나의 상상력과 창의력이 AI에 비해 미천하다는 느낌을 받은 것이다. 내가 창작하는 것도 아닌데 명령 단계에서부터 막힐 때가 종종 있었다. 어떤 이미지를 만들고 싶은지, 어떤 세부적인 요소가 들어가길 희망하는지, 어떤 느낌을 원하는지 나부터 확실하지 않은 경우가 많았다. 그리고 이미지 로딩이 완료되면 대부분의 경우는 내가 상상했던 결과물 보다 훨씬 더 기발한 결과물이 나왔다. \n",
      "\n",
      "챗GPT의 등장으로 인해 (실은 이것도 이전에 있었던 버전의 업그레이드 버전이다.) Midjourney와 같은 다른 AI 프로그램들도 덩달아 주목을 받고 있다. 근 몇 년, 늘 존재했던 기술이지만 이제야 비로소, 본격적으로 비전문가 일반인들도 AI를 인식하고 사용하기 시작하는 전환기가 도래한 것 같다. 격동적인 사회 변화가 예상되지만 일단은 이 신나는 현상을 흥미롭게 지켜봐야겠다.  \n",
      "['AI', '인공지능', '챗봇']\n",
      "\n",
      "\n",
      "--------------------------------------------------\n",
      "\n",
      "\n",
      "AI 시스템을 강화하는 휴먼인더루프(HITL)\n",
      "\n",
      "이 글은 제가 삼성 SDS < 인사이트 리포트 > 2022년 11월 30일에 기고/발행한 글입니다. 원본 글 'AI 시스템을 강화하는 휴먼인더루프(HUMAN IN THE LOOP)'의 서문만을 이곳 브런치에서 공유합니다.\n",
      "그리고 AI의 전체적인 강화프로세스를 알고 싶으신 분과, 특히 PM/PO, 기획자분들이 꼭 읽고 도움이 되길 바랍니다.\n",
      "다양한 비즈니스 영역에서 AI를 활용하여, 업무 프로세스를 혁신하여 초자동화를 이뤄내려는 시도가 늘고 있다. 초자동화는 그 기술 자체라기 보다는 ‘체계적인 문제 접근 방식’이다. 즉 AI와 머신러닝뿐만 아니라, 로봇 프로세스 자동화 (RPA: Robotic Process Automation), BPM / iBPMS (‘intelligent’ Business Process Management System), 통합 플랫폼 서비스 (iPaaS), low code / no code tools 및 다른 의사 결정도구나 조직의 베스트 프랙티스와 같은 기술, 도구, 플랫폼을 사용하여 최대한 많은 비즈니스 프로세스를 신속하게 찾아내고, 검증하여 자동화를 추진하는 체계적인 접근방법이라 할 수 있다.1 즉, 반복적이고 지루한 일련의 작업을 자동화된 시스템으로 대체하여, 인간이 더 가치 있는 업무에 집중할 수 있도록 하는 것이다.\n",
      "그러나 이 부분은 큰 필요조건이 만족되어야만 가능하다. AI의 작업 과정을 인간이 충분히 이해하고 신뢰하지 못한다면, 기업에서의 AI 도입과 효과는 매우 불투명해질 수밖에 없다. 기업의 입장에서는 AI를 사용하여 업무효율을 10% 올리는 것보다, 과정에서 예상하지 못하고 이해하기 어려운 상황을 어떻게 처리할는지가 더 중요하기 때문이다.\n",
      "예를 들어, AI에게 맡겨진 작업이 기업의 비용과 직간접적으로 연결이 되어 있어, 예상치 못한 금전 손실의 원인이 된다면, 아무리 많은 작업을 효율적으로 대신해주는 시스템이라 하더라도 신뢰하고 사용할 수는 없을 것이다. 또한 이것은 단순히 사용의 범위를 넘어, 전체 프로세스 중에 일어나는 중요한 결정에 대하여 AI의 예측 결과를 이해하고 수용할 수도 없게 된다. 즉 기업은 항상 혁신과 안정 중에 하나를 선택해야 한다면, 보수적이지만 안정적인 시스템을 선호할 수밖에 없는 조직이다.\n",
      "AI 서비스가 실제 업무 프로세스에 통합되기 위해서는 먼저 투명성, 공정성, 안정성이 확보되어야 작업 위임이 가능해진다. 현재 AI 모델의 예측 한계를 명확히 표시하고, 신뢰도가 낮거나 중요한 의사결정을 포함하는 데이터에 대해서 인간이 직접 리뷰하고 수정하여 품질을 높이는 과정은 AI 서비스에 대한 신뢰도를 높이는 전략이 된다. 이런 과정을 휴먼인더루프(Human-in-the-loop; 이하 HITL)라고 부른다.\n",
      "이처럼 HITL은 사용자가 AI 서비스를 수용할 수 있는 이해도를 높이고, 작업 과정의 효율화를 이끌어낼 수 있기에 기업의 초자동화를 이끄는 핵심 프로세스가 될 수 있다. HITL에 대해서 좀 더 자세히 정의하고, 활용 방안을 알아보자.\n",
      "\n",
      "신뢰할 수 있는 AI 업무 자동화\n",
      "휴먼인더루프 즉, HITL 접근을 좀 더 명확하게 정의하면 “AI를 활용해 추출한 데이터가 실제 중요한 비즈니스 프로세스에서 사용되기 전에, 사람이 데이터를 직접 검증 및 수정할 수 있도록 하는 시스템을 추가하여 데이터의 품질을 보장하기 위한 접근 방식”이다. HITL을 AI 도입 과정에서 시스템화하기 위해서는 작업 과정에서 발생할 수 있는 오류 케이스를 인간의 작업이 어떻게 더 쉽게 수행할 것인지에 대한 고민이 필요하다.아마존, 구글, 마이크로소프트와 같은 빅테크 플랫폼 기업의 경우엔, 자사의 AI 모델의 적용을 돕기 위한 관련 기능에 이미 HITL을 시스템화했다. 예를 들어, 구글은 Document AI에서 HITL과 관련된 다음의 4가지 기능을 제공하여, 검수 작업 관리에 대한 편의성을 높였다. • 신뢰도 기준값 필터 설정 • 검수를 손쉽게 도와주는 기능과 사용자 인터페이스 • 검수 인력 관리 기능 • 검수 작업 및 검수자에 관한 분석 기능먼저 사용자는 AI 작업 및 태스크 분류의 기준이 되는 신뢰도 점수를 직접 설정할 수 있다. 해당 신뢰도 점수를 기준으로 AI는 방대한 양의 문서에 대한 인식 작업을 진행한다. 이때 명확하게 인식하지 못한 문서에 대해서는 사용자가 주도적으로 수정할 수 있다.사용자의 인지 및 작업 흐름에 따라 맞춰 단순화한 사용자 인터페이스는 사람의 개입을 더욱 쉽게 만들어 데이터 품질 향상에 큰 도움이 된다. 적용 도메인 및 문서 특성에 따라 정확하게 인식하지 못한 예외 경우에는, 해당 문서를 분류한 뒤 선택적으로 확인하고 수정할 수 있게 시각적인 가이드를 제공한다. 예를 들어, 비용 지출 증빙 문서와 같은 정형화된 문서에서 날짜가 잘못 기입된 경우, 추출된 데이터에 빨간색으로 경고 표시를 한다. 또한 문서에서 인식된 영역을 표시하여 사용자가 맥락을 이해하고 수정할 수 있도록 지원한다.\n",
      "Document AI에서는 검수 작업을 사용자 스스로 진행할 수 있도록 하는 것뿐 아니라, 특정 검수 인력을 고용하거나, 내부 검수 인력을 관리한다는 점이 꽤나 인상적이다. AI 모델이 인식하지 못한 데이터에 대해서 인간이 검수 작업을 수행한다고 했을 때, 검수자의 변동, 태스크 할당을 동적으로 운영해야 하는 과정을 시스템적으로 해결하지 못한다면, 그 자체가 작업 오류 가능성을 높여 데이터의 품질 저하로 이어질 수 있다. 구글은 관리자의 대시보드에 검수 작업 할당과 인력, 비용 관리 기능을 제공하여, 관리자와 검수자의 커뮤니케이션 효율을 높였다. 또한 검수 작업과 검수 인력에 관한 분석 기능을 제공하여, AI와 인간 검수자의 작업 전환 과정과 효율을 동시에 높였다.\n",
      "AI 모델의 발전을 위한 데이터 완결성과 재학습\n",
      "\n",
      "확장적인 활용\n",
      "\n",
      "글 본문을 계속해서 읽으실 분들은 원문 인사이트 리포트 글로 이동하여 읽어주십시오.\n",
      "(글을 읽기 위해 추가적인 사용자 등록과 같은 과정은 필요없습니다.)\n",
      "\n",
      "*같은 글이 이곳에도 게재되었습니다.\n",
      "네이버 삼성SDS 포스트\n",
      "늘 여러분의 노력에 큰 응원을 보냅니다. 고맙습니다.\n",
      "['AI', '루프', '휴먼']\n",
      "\n",
      "\n",
      "--------------------------------------------------\n",
      "\n",
      "\n",
      "AI로 할 수 있는 걸 평가할 때 던져 볼 만한 질문들\n",
      "\n",
      "AI는 잘할 수 있는 것도 많지만 여전히 한계도 있다. AI로 이런 것도 할 수 있고 저런 것도 할 수 있다는 메시지가 넘쳐나지만 개중에는 AI로 하기 힘든데, 할 수 있다고 부풀리는 경우들도 있다.\n",
      "\n",
      "저넬 셰인이 쓴 좀 이상하지만 재미있는 녀석들에 따르면 AI로 이런 것도 할 수 있고 저런 것도 할 수 있다는 메시지를 평가할 때 던지면 좋은 몇 가지 질문들이 있다.\n",
      "\n",
      "우선은 문제가 얼마나 광범위한가? 이다.\n",
      "\n",
      "다음은 훈련용 데이터는 어디에서 얻었는가? 다.\n",
      "\n",
      "해당 문제가 높은 기억력을 요구하는가? 도 있다.\n",
      "\n",
      "인간의 편견을 모방만 하고 있는가? 라는 관점에서 볼 필요도 있다.\n",
      "['AI', '책이야기', 'IT']\n",
      "\n",
      "\n",
      "--------------------------------------------------\n",
      "\n",
      "\n",
      "나는 생각한다. 고로 존재한다.\n",
      "\n",
      "인류의 난제 중 하나는, 인간의 뇌가 동작하는 원리를 밝혀내는 것이다. 인간의 뇌를 흉내 낸 컴퓨터는 연산 장치(CPU)와 메모리를 통하여 계산을 하고 기억을 저장한다. 인간의 뇌는 뉴런(neuron)과 시냅스(synapse)의 연결에 의존한다. 그리고, 인간의 뇌는 현재의 기술로는 따라잡지 못할 정도의 최소의 에너지를 사용하여 기억하고 연산을 하고 있다. \n",
      " \n",
      "1,000억 개의 뉴런과 100조 개의 시냅스\n",
      "뇌는 크기가 대략 10cm×10cm×10cm, 한 변의 길이가 10cm인 육면체로 볼 수 있으며, 10세제곱cm에 1억 개의 뉴런을 가지고 있어, 인간의 뇌는 대략 1,000억 개의 뉴런으로 구성된다. 그리고, 뉴런과 뉴런을 화학 물질의 전달에 의하여 연결해 주는 선이 시냅스이다. 시냅스의 가중치에 따라서 뇌는 특정 내용을 더 잘 기억하게 된다. 멋진 풍경을 보았거나 어렸을 적 놀랐던 기억이 잘 떠오는 이유이다. 이론적으로 인간의 뇌를 구성하는 1,000억 개의 뉴런은 주변의 뉴런과 연결하여 약 100조 개의 시냅스를 갖는다 한다.                                                                                                                                 \n",
      "\n",
      "커넥톰, 뇌의 지도\n",
      "뉴런을 연결하는 지도, 뉴런은 노드에 해당하고 시냅스는 링크에 해당하여, 이러한 뇌의 지도를 커넥톰(Connectome)이라고 한다.   \n",
      "인류는 1986년에 처음으로 생명체인 ‘예쁜꼬마선충(Caenorhabditis elegans)’의 커넥톰을 완성했다. 0.5mm에 불과한 예쁜 꼬마선충은 302개의 뉴런과 7,000개의 시냅스를 가지고 있다. 첫 커넥톰을 만드는 게 기여한 Sydney Brenner 박사는 2002년 해당 공로로 노벨상 생리 및 의학분야를 수상했다. 30년이 지난 후 연구 수준은 1억 개의 뉴런을 연결하는 수준의 연구까지 왔을 것이다. 뇌는 부위에 따라 신체의 각 부분을 관장하는 역할을 한다. 인류가 커넥톰을 완성하였을 즈음에는 뇌와 관련한 질환들, 치매, 자폐증, 정신분열증에서 진전이 있을 것이다.  \n",
      "\n",
      "인간은 게놈과 커넥톰의 연결\n",
      "나 자신은 결국 게놈과 커넥톰의  연결이다. 즉, 유전자와 경험이 나를 만드는 것이다. 게놈은 선천적인 것을 결정하여 변하지 않으나, 반면에 커넥톰은 경험에 의하여 그 연결이 변경되는 가소성을 갖는다. 가소성은 4R, 즉, Re-weight, Re-connect, Re-wire, Re-generation으로 일어나게 된다.   \n",
      "\n",
      "어려운 커넥톰\n",
      "책 ‘커넥톰’은 몇 번 읽으려다가 내용이 어려워 포기하였는 데, 다시 집어 들었다. 어려운 뇌과학을 로맨스 소설을 쓰듯이 쉽게 쓰려고 하였다. 여인이 식장 문에 들어올 때, 신체 감각기관과 뇌가 어떻게 반응하는지를 설명하고 첫 키스 뉴런과 재니퍼 애니스톤 뉴런을 도입하기도 하였지만, 뇌과학은 쉽게 접근을 허용하는 분야는 아니다.  \n",
      "\n",
      "Je pense, donc je suis\n",
      "데카르트의 말, “나는 생각한다. 고로 존재한다. (Je pense, donc je suis.)”와 같이 뇌는 지금 이 순간에도 생각을 계속하여 분비하고 있다. 최근의 ChatGPT 열풍을 보다라도 지금 AI의 발전은 지수적으로 상승하고 있다. 쉬운 결정은 AI가 처리하고, 어려운 결정은 인간이 할 거라는 예상이 있으나, 이 말은 조만간에 틀린 것으로 밝혀지게 될 것이다. \n",
      "\n",
      "어려운 결정은 인간과 AI가 협업할 수 밖에 없는 시대가 다가오고 있다. \n",
      "\n",
      "by 웨이브리지, 글모음 https://brunch.co.kr/@waybridge/\n",
      "['AI', '뇌과학', '과학기술']\n",
      "\n",
      "\n",
      "--------------------------------------------------\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import jsonlines\n",
    "\n",
    "with jsonlines.open(\"brunch_dataset_10_sample.jsonl\") as f:\n",
    "    for line in f.iter():\n",
    "        print(line[\"title\"]) # 각 json에 해당하는 \"id\" 출력\n",
    "        print(line[\"body_text\"]) # 각 json에 해당하는 \"title\" 출력\n",
    "        print(line['keyword'])\n",
    "        print(\"\\n\\n--------------------------------------------------\\n\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a41749a7",
   "metadata": {},
   "source": [
    "이제 423967개의 데이터 모두를 jsonl로 저장하자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "id": "741ffcd7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "with open(\"brunch_dataset_10.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe10_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99228296",
   "metadata": {},
   "source": [
    "### 좋아요가 20/30/40/50/100 이상인 글 jsonl로 저장"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13adc891",
   "metadata": {},
   "source": [
    "#### 20개 이상\n",
    "더 좋은 글을 가려내기 위해 좋아요가 20개 이상인 글만 추출."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "id": "ef37f02f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "127520"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "super_high_quality_essay = total_df['likes'] >= 20\n",
    "super_high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "0ec064e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "shqe_df = total_df[super_high_quality_essay]\n",
    "shqe_df = shqe_df.reset_index()\n",
    "shqe_df = shqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "shqe_df = shqe_df.reset_index()\n",
    "shqe_df = shqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "hqe20_json_data = shqe_df.to_dict('records')\n",
    "\n",
    "for json in hqe20_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_20.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe20_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "386fd7c1",
   "metadata": {},
   "source": [
    "#### 30개 이상"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "id": "0c926bb8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "61442"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "amazing_high_quality_essay = total_df['likes'] >= 30\n",
    "amazing_high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "3c1944e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "ahqe_df = total_df[amazing_high_quality_essay]\n",
    "ahqe_df = ahqe_df.reset_index()\n",
    "ahqe_df = ahqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "ahqe_df = ahqe_df.reset_index()\n",
    "ahqe_df = ahqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "hqe30_json_data = ahqe_df.to_dict('records')\n",
    "\n",
    "for json in hqe30_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_30.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe30_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e47d10e",
   "metadata": {},
   "source": [
    "#### 40개  이상"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "45baa205",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "38121"
      ]
     },
     "execution_count": 137,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fantastic_high_quality_essay = total_df['likes'] >= 40\n",
    "fantastic_high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "248928cb",
   "metadata": {},
   "outputs": [],
   "source": [
    "fhqe_df = total_df[fantastic_high_quality_essay]\n",
    "fhqe_df = fhqe_df.reset_index()\n",
    "fhqe_df = fhqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "fhqe_df = fhqe_df.reset_index()\n",
    "fhqe_df = fhqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "hqe40_json_data = fhqe_df.to_dict('records')\n",
    "\n",
    "for json in hqe40_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_40.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe40_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "449b282d",
   "metadata": {},
   "source": [
    "#### 50개 이상"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "c9588690",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "26190"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "incredible_high_quality_essay = total_df['likes'] >= 50\n",
    "incredible_high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "d974c4c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "ihqe_df = total_df[incredible_high_quality_essay]\n",
    "ihqe_df = ihqe_df.reset_index()\n",
    "ihqe_df = ihqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "ihqe_df = ihqe_df.reset_index()\n",
    "ihqe_df = ihqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "hqe50_json_data = ihqe_df.to_dict('records')\n",
    "\n",
    "for json in hqe50_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_50.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe50_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38643216",
   "metadata": {},
   "source": [
    "#### 100개 이상"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "id": "e1407e75",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7999"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "crazy_high_quality_essay = total_df['likes'] >= 100\n",
    "crazy_high_quality_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "id": "178a877e",
   "metadata": {},
   "outputs": [],
   "source": [
    "chqe_df = total_df[crazy_high_quality_essay]\n",
    "chqe_df = chqe_df.reset_index()\n",
    "chqe_df = chqe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "chqe_df = chqe_df.reset_index()\n",
    "chqe_df = chqe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "hqe100_json_data = ihqe_df.to_dict('records')\n",
    "\n",
    "for json in hqe100_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_100.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in hqe100_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "207ff679",
   "metadata": {},
   "source": [
    "### 좋아요가 0개인 글 jsonl로 저장\n",
    "\n",
    "좋아요가 0개인 글은 상대적으로 낮은 퀄리티일 가능성이 높다.(혹은 대중들이 좋아하지 않는 글)\n",
    "\n",
    "좋아요가 0개인 글과 좋아요가 50개 이상인 글을 에세이 자동 채점 모델을 통해 채점하고 평균 점수를 기록해본다면 에세이 자동 채점 모델의 성능을 어느정도 검증할 수 있을 것으로 예상한다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "id": "69d8faf2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "48957"
      ]
     },
     "execution_count": 151,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unpopular_essay = total_df['likes'] == 0\n",
    "unpopular_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "id": "a1d1def9",
   "metadata": {},
   "outputs": [],
   "source": [
    "upe_df = total_df[unpopular_essay]\n",
    "upe_df = upe_df.reset_index()\n",
    "upe_df = upe_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "upe_df = upe_df.reset_index()\n",
    "upe_df = upe_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "upe_json_data = upe_df.to_dict('records')\n",
    "\n",
    "for json in upe_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_0.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in upe_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6eac903b",
   "metadata": {},
   "source": [
    "### 너무 짧은 글을 제외한 데이터 jsonl로 저장\n",
    "\n",
    "AI hub의 에세이 평가 데이터셋의 평균 음절수는 대략 300 ~ 1000이다. \n",
    "\n",
    "띄어쓰기를 감안하여 body_text가 500이하인 글을 제거한다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "e978436e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "739213"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "long_essay = total_df['body_text'].str.len() > 500\n",
    "long_essay.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "7ffff065",
   "metadata": {},
   "outputs": [],
   "source": [
    "long_df = total_df[long_essay]\n",
    "long_df = long_df.reset_index()\n",
    "long_df = long_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "long_df = long_df.reset_index()\n",
    "long_df = long_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "long_json_data = long_df.to_dict('records')\n",
    "\n",
    "for json in long_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset_long.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in long_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56d867ff",
   "metadata": {},
   "source": [
    "### 전체(857807개)도 jsonl로 저장"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "id": "26b47a0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "total_df = total_df[['title', 'sub_title', 'body_text', 'keyword', 'likes', 'post_date', 'post_url', 'author_id']]\n",
    "total_df = total_df.reset_index()\n",
    "total_df = total_df.drop(['index'], axis=1) # 기존 인덱스 제거\n",
    "\n",
    "total_json_data = total_df.to_dict('records')\n",
    "\n",
    "for json in total_json_data:\n",
    "    keyword_list = str_to_list(json['keyword'])\n",
    "    json['keyword'] = keyword_list\n",
    "\n",
    "import json    \n",
    "    \n",
    "with open(\"brunch_dataset.jsonl\",\"w\", encoding=\"UTF-8\") as f:\n",
    "    for line in total_json_data:\n",
    "        f.writelines(json.dumps(line,ensure_ascii=False))\n",
    "        f.write(\"\\n\")"
   ]
  }
 ],
 "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": "3.11.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}