008. simple book recommender system with Keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 블로그 설명\n",
    "\n",
    "해당 자료에 대한 설명은 아래 블로그에 올려두었습니다.\n",
    "- https://lsjsj92.tistory.com/606"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import warnings\n",
    "\n",
    "from tensorflow.keras.layers import Input, Embedding, Flatten, Dot, Dense, Concatenate\n",
    "from tensorflow.keras.models import Model, load_model\n",
    "from tensorflow.keras.utils import plot_model\n",
    "from sklearn.model_selection import train_test_split\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('../data/book_kaggle/ratings.csv')\n",
    "books = pd.read_csv('../data/book_kaggle/books.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(981756, 3)\n"
     ]
    },
    {
     "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>book_id</th>\n",
       "      <th>user_id</th>\n",
       "      <th>rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>314</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>439</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>588</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1169</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1185</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   book_id  user_id  rating\n",
       "0        1      314       5\n",
       "1        1      439       3\n",
       "2        1      588       5\n",
       "3        1     1169       4\n",
       "4        1     1185       4"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 사용자가 어떤 book을 평가했는지에 대한 데이터\n",
    "print(data.shape)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10000, 23)\n"
     ]
    },
    {
     "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>id</th>\n",
       "      <th>book_id</th>\n",
       "      <th>best_book_id</th>\n",
       "      <th>work_id</th>\n",
       "      <th>books_count</th>\n",
       "      <th>isbn</th>\n",
       "      <th>isbn13</th>\n",
       "      <th>authors</th>\n",
       "      <th>original_publication_year</th>\n",
       "      <th>original_title</th>\n",
       "      <th>...</th>\n",
       "      <th>ratings_count</th>\n",
       "      <th>work_ratings_count</th>\n",
       "      <th>work_text_reviews_count</th>\n",
       "      <th>ratings_1</th>\n",
       "      <th>ratings_2</th>\n",
       "      <th>ratings_3</th>\n",
       "      <th>ratings_4</th>\n",
       "      <th>ratings_5</th>\n",
       "      <th>image_url</th>\n",
       "      <th>small_image_url</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2767052</td>\n",
       "      <td>2767052</td>\n",
       "      <td>2792775</td>\n",
       "      <td>272</td>\n",
       "      <td>439023483</td>\n",
       "      <td>9.780439e+12</td>\n",
       "      <td>Suzanne Collins</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>The Hunger Games</td>\n",
       "      <td>...</td>\n",
       "      <td>4780653</td>\n",
       "      <td>4942365</td>\n",
       "      <td>155254</td>\n",
       "      <td>66715</td>\n",
       "      <td>127936</td>\n",
       "      <td>560092</td>\n",
       "      <td>1481305</td>\n",
       "      <td>2706317</td>\n",
       "      <td>https://images.gr-assets.com/books/1447303603m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1447303603s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>4640799</td>\n",
       "      <td>491</td>\n",
       "      <td>439554934</td>\n",
       "      <td>9.780440e+12</td>\n",
       "      <td>J.K. Rowling, Mary GrandPré</td>\n",
       "      <td>1997.0</td>\n",
       "      <td>Harry Potter and the Philosopher's Stone</td>\n",
       "      <td>...</td>\n",
       "      <td>4602479</td>\n",
       "      <td>4800065</td>\n",
       "      <td>75867</td>\n",
       "      <td>75504</td>\n",
       "      <td>101676</td>\n",
       "      <td>455024</td>\n",
       "      <td>1156318</td>\n",
       "      <td>3011543</td>\n",
       "      <td>https://images.gr-assets.com/books/1474154022m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1474154022s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>41865</td>\n",
       "      <td>41865</td>\n",
       "      <td>3212258</td>\n",
       "      <td>226</td>\n",
       "      <td>316015849</td>\n",
       "      <td>9.780316e+12</td>\n",
       "      <td>Stephenie Meyer</td>\n",
       "      <td>2005.0</td>\n",
       "      <td>Twilight</td>\n",
       "      <td>...</td>\n",
       "      <td>3866839</td>\n",
       "      <td>3916824</td>\n",
       "      <td>95009</td>\n",
       "      <td>456191</td>\n",
       "      <td>436802</td>\n",
       "      <td>793319</td>\n",
       "      <td>875073</td>\n",
       "      <td>1355439</td>\n",
       "      <td>https://images.gr-assets.com/books/1361039443m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1361039443s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>2657</td>\n",
       "      <td>2657</td>\n",
       "      <td>3275794</td>\n",
       "      <td>487</td>\n",
       "      <td>61120081</td>\n",
       "      <td>9.780061e+12</td>\n",
       "      <td>Harper Lee</td>\n",
       "      <td>1960.0</td>\n",
       "      <td>To Kill a Mockingbird</td>\n",
       "      <td>...</td>\n",
       "      <td>3198671</td>\n",
       "      <td>3340896</td>\n",
       "      <td>72586</td>\n",
       "      <td>60427</td>\n",
       "      <td>117415</td>\n",
       "      <td>446835</td>\n",
       "      <td>1001952</td>\n",
       "      <td>1714267</td>\n",
       "      <td>https://images.gr-assets.com/books/1361975680m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1361975680s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>4671</td>\n",
       "      <td>4671</td>\n",
       "      <td>245494</td>\n",
       "      <td>1356</td>\n",
       "      <td>743273567</td>\n",
       "      <td>9.780743e+12</td>\n",
       "      <td>F. Scott Fitzgerald</td>\n",
       "      <td>1925.0</td>\n",
       "      <td>The Great Gatsby</td>\n",
       "      <td>...</td>\n",
       "      <td>2683664</td>\n",
       "      <td>2773745</td>\n",
       "      <td>51992</td>\n",
       "      <td>86236</td>\n",
       "      <td>197621</td>\n",
       "      <td>606158</td>\n",
       "      <td>936012</td>\n",
       "      <td>947718</td>\n",
       "      <td>https://images.gr-assets.com/books/1490528560m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1490528560s...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   id  book_id  best_book_id  work_id  books_count       isbn        isbn13  \\\n",
       "0   1  2767052       2767052  2792775          272  439023483  9.780439e+12   \n",
       "1   2        3             3  4640799          491  439554934  9.780440e+12   \n",
       "2   3    41865         41865  3212258          226  316015849  9.780316e+12   \n",
       "3   4     2657          2657  3275794          487   61120081  9.780061e+12   \n",
       "4   5     4671          4671   245494         1356  743273567  9.780743e+12   \n",
       "\n",
       "                       authors  original_publication_year  \\\n",
       "0              Suzanne Collins                     2008.0   \n",
       "1  J.K. Rowling, Mary GrandPré                     1997.0   \n",
       "2              Stephenie Meyer                     2005.0   \n",
       "3                   Harper Lee                     1960.0   \n",
       "4          F. Scott Fitzgerald                     1925.0   \n",
       "\n",
       "                             original_title  ... ratings_count  \\\n",
       "0                          The Hunger Games  ...       4780653   \n",
       "1  Harry Potter and the Philosopher's Stone  ...       4602479   \n",
       "2                                  Twilight  ...       3866839   \n",
       "3                     To Kill a Mockingbird  ...       3198671   \n",
       "4                          The Great Gatsby  ...       2683664   \n",
       "\n",
       "  work_ratings_count  work_text_reviews_count  ratings_1  ratings_2  \\\n",
       "0            4942365                   155254      66715     127936   \n",
       "1            4800065                    75867      75504     101676   \n",
       "2            3916824                    95009     456191     436802   \n",
       "3            3340896                    72586      60427     117415   \n",
       "4            2773745                    51992      86236     197621   \n",
       "\n",
       "   ratings_3  ratings_4  ratings_5  \\\n",
       "0     560092    1481305    2706317   \n",
       "1     455024    1156318    3011543   \n",
       "2     793319     875073    1355439   \n",
       "3     446835    1001952    1714267   \n",
       "4     606158     936012     947718   \n",
       "\n",
       "                                           image_url  \\\n",
       "0  https://images.gr-assets.com/books/1447303603m...   \n",
       "1  https://images.gr-assets.com/books/1474154022m...   \n",
       "2  https://images.gr-assets.com/books/1361039443m...   \n",
       "3  https://images.gr-assets.com/books/1361975680m...   \n",
       "4  https://images.gr-assets.com/books/1490528560m...   \n",
       "\n",
       "                                     small_image_url  \n",
       "0  https://images.gr-assets.com/books/1447303603s...  \n",
       "1  https://images.gr-assets.com/books/1474154022s...  \n",
       "2  https://images.gr-assets.com/books/1361039443s...  \n",
       "3  https://images.gr-assets.com/books/1361975680s...  \n",
       "4  https://images.gr-assets.com/books/1490528560s...  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 책에 대한 정보\n",
    "print(books.shape)\n",
    "books.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 981756 entries, 0 to 981755\n",
      "Data columns (total 3 columns):\n",
      " #   Column   Non-Null Count   Dtype\n",
      "---  ------   --------------   -----\n",
      " 0   book_id  981756 non-null  int64\n",
      " 1   user_id  981756 non-null  int64\n",
      " 2   rating   981756 non-null  int64\n",
      "dtypes: int64(3)\n",
      "memory usage: 22.5 MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10000 entries, 0 to 9999\n",
      "Data columns (total 23 columns):\n",
      " #   Column                     Non-Null Count  Dtype  \n",
      "---  ------                     --------------  -----  \n",
      " 0   id                         10000 non-null  int64  \n",
      " 1   book_id                    10000 non-null  int64  \n",
      " 2   best_book_id               10000 non-null  int64  \n",
      " 3   work_id                    10000 non-null  int64  \n",
      " 4   books_count                10000 non-null  int64  \n",
      " 5   isbn                       9300 non-null   object \n",
      " 6   isbn13                     9415 non-null   float64\n",
      " 7   authors                    10000 non-null  object \n",
      " 8   original_publication_year  9979 non-null   float64\n",
      " 9   original_title             9415 non-null   object \n",
      " 10  title                      10000 non-null  object \n",
      " 11  language_code              8916 non-null   object \n",
      " 12  average_rating             10000 non-null  float64\n",
      " 13  ratings_count              10000 non-null  int64  \n",
      " 14  work_ratings_count         10000 non-null  int64  \n",
      " 15  work_text_reviews_count    10000 non-null  int64  \n",
      " 16  ratings_1                  10000 non-null  int64  \n",
      " 17  ratings_2                  10000 non-null  int64  \n",
      " 18  ratings_3                  10000 non-null  int64  \n",
      " 19  ratings_4                  10000 non-null  int64  \n",
      " 20  ratings_5                  10000 non-null  int64  \n",
      " 21  image_url                  10000 non-null  object \n",
      " 22  small_image_url            10000 non-null  object \n",
      "dtypes: float64(3), int64(13), object(7)\n",
      "memory usage: 1.8+ MB\n"
     ]
    }
   ],
   "source": [
    "books.info()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(785404, 3) (196352, 3)\n"
     ]
    },
    {
     "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>book_id</th>\n",
       "      <th>user_id</th>\n",
       "      <th>rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>833746</th>\n",
       "      <td>8425</td>\n",
       "      <td>41968</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>445541</th>\n",
       "      <td>4464</td>\n",
       "      <td>9820</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>64697</th>\n",
       "      <td>648</td>\n",
       "      <td>1136</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>528509</th>\n",
       "      <td>5302</td>\n",
       "      <td>30532</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>809576</th>\n",
       "      <td>8171</td>\n",
       "      <td>43077</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        book_id  user_id  rating\n",
       "833746     8425    41968       4\n",
       "445541     4464     9820       5\n",
       "64697       648     1136       2\n",
       "528509     5302    30532       3\n",
       "809576     8171    43077       3"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train, test = train_test_split(data, test_size = 0.2)\n",
    "print(train.shape, test.shape)\n",
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "53424 10000\n"
     ]
    }
   ],
   "source": [
    "number_of_unique_user = len(data.user_id.unique())\n",
    "number_of_unique_book_id = len(data.book_id.unique())\n",
    "print(number_of_unique_user, number_of_unique_book_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 첫 번째 모델\n",
    "\n",
    "간단한 Dot product 형태의 모델이다.   \n",
    "사용자 embedding과 book embedding을 기반으로 나온 벡터 값을 Dot 연산을 통해 rating을 예측하는 방법"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# layer 만들기\n",
    "book_input = Input(shape=(1,), name='book_input_layer')\n",
    "user_input = Input(shape=(1,), name='user_input_layer')\n",
    "\n",
    "book_embedding_layer = Embedding(number_of_unique_book_id + 1, 8, name='book_embedding_layer')\n",
    "user_embedding_layer = Embedding(number_of_unique_user + 1, 8, name='user_embedding_layer')\n",
    "\n",
    "book_vector_layer = Flatten(name='book_vector_layer')\n",
    "user_vector_layer = Flatten(name='user_vector_layer')\n",
    "\n",
    "dot_result_layer = Dot(name='dot_vector_layer', axes = 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 쌓기\n",
    "book_embedding = book_embedding_layer(book_input)\n",
    "user_embedding = user_embedding_layer(user_input)\n",
    "\n",
    "book_vector = book_vector_layer(book_embedding)\n",
    "user_vector = user_vector_layer(user_embedding)\n",
    "\n",
    "dot_result = dot_result_layer([book_vector, user_vector])\n",
    "model = Model(inputs=[user_input, book_input], outputs=dot_result)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_1\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "book_input_layer (InputLayer)   [(None, 1)]          0                                            \n",
      "__________________________________________________________________________________________________\n",
      "user_input_layer (InputLayer)   [(None, 1)]          0                                            \n",
      "__________________________________________________________________________________________________\n",
      "book_embedding_layer (Embedding (None, 1, 8)         80008       book_input_layer[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "user_embedding_layer (Embedding (None, 1, 8)         427400      user_input_layer[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "book_vector_layer (Flatten)     (None, 8)            0           book_embedding_layer[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "user_vector_layer (Flatten)     (None, 8)            0           user_embedding_layer[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "dot_vector_layer (Dot)          (None, 1)            0           book_vector_layer[0][0]          \n",
      "                                                                 user_vector_layer[0][0]          \n",
      "==================================================================================================\n",
      "Total params: 507,408\n",
      "Trainable params: 507,408\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAuAAAAGVCAIAAAAwu9OxAAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nOzdeVhTV/4/8BMSCKCogBBlrahItSzW1hEpoFN12hGoCyIwYgtagxanaMu009Fq1WrdCtOKVEZrhcLToGgtPBZ3rQsujxVFQXE3UfZFFotkub8/7u+bySQhhBCSm/B+/dEnOdzlc2POp5/cnJzDoiiKAAAAADCJhbEDAAAAAFCGAgUAAAAYBwUKAAAAMA4KFAAAAGAcjr4OVFxc/PXXX+vraADQq5YvXx4YGGjsKHQ0Z84cY4cAAOrpMbfo7Q6KUCjct2+fvo4GfcqFCxcuXLhg7Ci6TSQSmeh7ft++fUKh0NhR6G7fvn0ikcjYUYAJQG4xMP3mFr3dQaHt3btXvweEvoD+QGxyb568vLy5c+eaXNiEEBaLZewQemrZsmVRUVHGjgKYDrnFwPSbWzAGBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjIMCBUzPpEmTvvrqK2NHoa033ngjNjY2Ly/v2bNndItEIklPT29raysoKDh58uQXX3wRHBx87949zccpLCz09/cvLS2ln5aUlJw8eZJ+fPXq1d27d7/00kvbt2/vvQsBMHvILYRJuQUFCpieAwcOfPTRR3o5VGlpaZe9t+f+9re/RUVFDRw4kBAilUqTkpIiIyOtra1TU1NDQ0NXrVrF4/EyMjI0HKGystLFxeX69evyloCAgNbWVnquybFjx8bHx0+YMIGiqN6+FgAzhtxCmJRbUKCA6bG3t7e0tOz5cZqamqKjo9va2np+KO3l5OSMHDmSx+Ox2Ww7O7va2lpCyIgRI8rLyzXsNXToUD8/P6XGsLCwjIwMefwsFssM5rAHMCLkFhpDcgsKFDAxpaWl8fHxK1euJITk5+cHBQVlZ2dHRUU5ODj89NNPhJDc3NyJEydu27YtNDTU1tZ21apVhJDs7Gwej1dRUSESiaKjo0NDQwkhe/bsKSsrS0tLKygoIIRERERs2LChV4OnKGrNmjWzZ8+mnx48eJDH4xFCrly5Ehsb292jsVgsPz8/zR+PAEBLyC1yDMktKFDAxPj6+j579owu7cPDw2/fvv3777/n5uauXLly/fr1hJCZM2devny5trb2+PHjX3/99dq1a6urq2NjY2tqaiQSiZub2/Tp0+vr6wkhSUlJhJCUlJTw8HBCyIwZM4KCgno1+CdPnjx69MjT01OxsaamRiwW65BECCHu7u5Hjx7VU3QAfRpyiyIm5BYUKGB6HBwc6AdWVlZ2dnZBQUEcDmf8+PH0Mt82NjY2NjZTp07lcDiJiYnOzs6HDx9ms9ny3Tkc9Yt4JyQkhISE9GrkFRUVTk5OSjdLBQJBWlqabndQ3d3dKyoq9BQdQF+H3CLHhNyCAgXMBJvNVjuMKzAw8NatWxp2NOR3q21tbVKpVKnRwsIiICBAtwM6OTk1NDT0OC4A6BRyi7GgQAEzd/fuXW9vbw0bGDKJ+Pj41NXVSSQSxcapU6fqfMCGhgYPD48exwUA3Ybc0ttQoIDpefHiRUdHB/1YKpXSH246OjpkMpl8G/qL5Pr6+qqqqsjISEKIh4dHWVkZIaS8vLylpYUQwmazuVxuQ0NDe3s7IeTQoUMlJSW9GrmXl5e1tXV1dbW8pb6+no5KTigUZmZmisVipX2VUg+tqqpK509IAKAEuUWOCbkFBQqYmIsXLxYXF58+ffrmzZsnTpwQiUQFBQV1dXXZ2dktLS30mHlCyJ49e9LT0/l8fn5+fv/+/QkhSUlJcXFxERERjo6O/fv3LyoqIoQkJCTExMTQe23atCkrK6tXg2ez2cnJyZcvX5a3HDt2LD4+XjH9FRYW8vl8Ojy5mpqaTZs2EUJ27979+PFjefvRo0f5fH6vxgzQRyC3MC63UHoiEAj0eDToUyIjIyMjI/V4QDs7uzNnzlRXV8tkMsX2lpYWiqLEYrFiY2trK/2gvb1daXvNtHzPBwUFFRYWyp82NzdHRUUpbtDW1qa0S2Fh4bVr17o8skgkiouLkz+Njo5OT0/vci+KogghAoFAmy2ZydTjB4NBblHaxbRyC+6ggBmiKEomkzk7Oyt9B0x/3FEaad+vXz/6AZfL7aXvjM+dO3fnzh361rGdnd2KFSsUP07Z2toqblxXV9fc3Kw6dZKS+vr6zZs379y5kxDS0NBw+/bt+/fv90LsAPBfyC2GpP43UQCmSyAQtLa25uTkDB8+3NXV1djhkJ07d7a0tDQ1Nbm5udEtvr6+jo6OjY2N9vb2qtsPHjw4Jiamy8M+ffp0y5YtdEJ8/vx5c3Pztm3blGZBAAA9Qm4xMBQoYG7CwsLocWpcLtfYsRBCiI+Pj2qji4tLDw/r6+srf+zm5ibPUADQS5BbDAwFCpgb+W1VAAA9Qm4xMIxBAQAAAMYxaIHyzTffsNlsnYfb5Obmcjic06dPa7PxpEmTvvrqK91O1C3nz5/38PBISUnp7RMdOXLk1VdfZbFYycnJIpFIj0fOz88fOnSohYXFt99+S/9qHwA0MEx6MVhuIb2WXpBboCcMWqC8//77ir/J7q7Y2FhHR0ctNz5w4MBHH32k87kUlZaW3rt3r7O/Tpw4kV6+srdNmzaNXvNp3bp1evlSUH5ds2fPnjx58siRI5cuXWptbd3zIwOYN8OkF4PlFqLv9ILcAnph0ALFwsKC9Gz2X8VlmTSzt7e3tLTU+URyTU1N0dHR9NSBGqIyzJTG9DegevkeVOm6bG1tbWxsen5YgL7AYOnFYLmF6C+9ILeAvhhhDMrx48cDAwOdnJy+//57uuXSpUtLly5dvXp1eHi4fD5gtY201tbWhQsXLlu2rLPJg0tLS+Pj41euXEkIyc/PDwoKys7OjoqKcnBw+Omnnwghubm5EydO3LZtW2hoqK2t7apVqwgh2dnZPB6voqJCJBJFR0fTn1327NlTVlaWlpYmn0ZQs61btwoEgsWLF9Nnz8rKGjly5Jo1ayQSSXNzc3R0ND2LX1FREZ/PDwkJ+fbbbwkh+/fvf/PNN3NycsaMGUPf0Y2IiNiwYYOGExnmunS7HLVXBGAwat/zhJCtW7f+8MMP4eHhP//8M93SZU9UZaz0otQZia7phSG5RfWKdM6WYLb0NeObNjPf0d9Brl69urW1devWrdbW1m1tbZWVlV5eXvQ0fAKBwNHRsb6+Xm0jRVFDhw49derUqVOnMjIyNJ9r5syZy5YtoyjqxYsXjo6OycnJYrH466+/9vX1pSjq+fPnHA7n888/F4vFGRkZLBarqqqKXo/g5s2bFEVlZWWNGTOGoii6saysTMO53n333ZSUFIqihELhwIED6eNbWlp2dHRQFBUQELBhwwZ6yyVLllAUdffu3cTERIqimpqauFzuzZs3q6qq2Gz2hx9+WFJScubMGYqidu3adfr0aaUTbd++nRBCz0io3+tasGCBv7+/0ul0vhyKolSvSAN6PQswJJOeibXL+NW+5+/cuTNnzhyKopqamrKzsynteqJaBksvmnMLpVN6UZtbKIX0YoDc0tkV6ZYtNUBuMTw95hYj/Mx44cKF/fr1S05OXrVq1YkTJ0pLS729velp+MLCwmJjYwUCQVNTk2rj4sWLCSGHDx+2t7fvsnB2cHCgH1hZWdnZ2QUFBXE4nPHjx69Zs4YQYmNjY2NjM3XqVA6Hk5iYuHr16sOHD8+fP1++u9KEgFpydXWlx/AWFxdLpVKhUOjl5bV06dK1a9f+4x//qKmpoReHzMzMfPbs2fr16wkhEyZMuHDhQkJCwuDBg8PDw/39/elDJSQkaD6XAa5L58sZPXo0j8dTuiLNJkyYsGzZMu1jY4Li4uK0tDS6NDctc+fONXYIvUvxu2D5e37AgAEFBQXffPNNUlLSrFmziHY9US3Dpxe1nZEQokN6eeONNzSfyzAXpcdsqRlyiyHpN7cYoUChv1K1sLAYN27c1atXhUIhPTaFEGJra+vr6/vw4cPGxkbVRvrpsWPHLC0tly9frv14FDk2m01RlGp7YGDgrVu3uoy5SywWq7m5+Z///OfChQtZLJZUKiWExMTEpKSkFBUVPX78OCoqihBy9+7d8PDw9957jxDy2Wef0ftaWFjoVhWRXruunlwO6eYVubm50UczLWlpaaYYttkXKGo5OztnZmYuWrQoPz9///79tra2+u2JvZpe1HZG0vvppfcuymDZErnFkPSbW4w5D0p7e/uIESNcXV0vXbokb7S0tHR1dVXbSD9es2ZNU1PTl19+qcdI7t696+3trWEDLQuUmzdvLliwYO3atcOHD5c32tjYLFiwYPv27UKhcNiwYYQQf3//8+fPyzfQ8BOhHurJdVVWVu7fv59RlwPQEzU1NfPmzbty5UpjY+PSpUuJod66ekkvanMLMV5/7GFuuXbtGtOyJTCQEQoUuh5vbm6ura2dOXNmdHR0U1PTjRs3CCFisfjOnTuzZ89W20jva2Njk52dvWHDhpMnT2o4y4sXL+j1kwghUqmUPmlHR4fi75zpceb19fVVVVX0V5UeHh5lZWWEkPLycnpKYzabzeVyGxoaNPyIXyqV0uX/b7/9Ro+buXXrllQqFYvF9AZLliw5cuSIfAbit99+Oysr67vvvhOLxefOnbt+/TohRCaTybcnhBw6dEh1CHBra6v8v/q9LnrRS/kRZDLZhx9+eO/ePZ0vR/WKAAxJ9T1fUlJy8eLF0aNHf//99w0NDUS7nqiWwdJLl7mFdD+9qM0t5H/TS2/nFhcXFz1mSzBb+hrMos0gWZlMFh8fP2fOnNTU1HfeeefGjRt0e3Z2dkhIyKFDh+bNm5efn99ZY15eHiEkMTGxtrY2LCxs0KBB27dvV3uiCxcuDB8+/JVXXrlx48bx48dZLFZcXFxtbe3ChQsJIb/88gtFUXZ2djExMdu2bZs9e7Z8yNimTZusra3Dw8PT0tJGjx7966+/UhS1ePFiT0/PvLy8zs7l6ek5duzYsrKyBw8euLi4vPbaazt27AgMDJw2bVpdXR292VtvvVVZWSl/Hei7mp6enp999plMJvvxxx8JIXPmzHn48CG9TWhoKD0KT+7IkSNjx44lhCQnJwuFQj1eV35+/pAhQ7hc7sKFCxMTE2NjY0eMGOHn56fz5VAUpXpFGuh9SXTD0HJJdAYi5j5IllL3nj9y5MjkyZOLioq2bNly8uRJSrueqMpg6UXL3EJ1M72o5hbqf9PLnj17eju3UBSlx2ypAXKLgek3txi0QKF1dHQ0NjYqNba3t5eXl7948aLLRn2xs7M7c+ZMdXU1/T9UObqoF4vFio2tra1aHlYsFkskEoqiJBKJ4pHpEemKGhsbNVxae3u7UmBa0u916etyNEMSMbC+UKBQKu95+q5AbW2t0mY9eet2pjfSS2edkepmf2RIbqEMkl6QWwxMv7nFCINkLS0tBw0apNTI5XJVF2ZU26hkxowZSi0sFuvAgQNdhkFRlEwmc3Z2VmqnfzqkNAKLnrxIm3PJd6TH8IpEooqKirKyspkzZyrtq/oiKNJ5tUwdrksDfV0OgOEpvefpcfeDBw9W2kzDW9dg6UWH3EJ07Y8MyS0E6QW6YvKrGcsnXOoWgUDQ2tqak5MzfPhw+fDb3jjXL7/88umnn6akpEyZMqW7++pAt+vSnoEvB8C4DJZedDuRIftjb+cWgvQCKky+QNFNWFgYPZ5L5w8TWlqyZEliYqL8J9O9rbevy8CXA2CKzDK9GOCikF5ASR8tUPSynI2WDNnfDHBdSB8AmpllejHMRSG9gCK8GwB61xtvvBEbG5uXl/fs2TO6RSKRpKent7W1FRQUnDx58osvvggODu5ygofCwkJ/f//S0lL6aUlJifyX9levXt29e/dLL71Ez1YOAH2B2ecWFChghjQsYa/3vbTxt7/9LSoqauDAgYQQqVSalJQUGRlpbW2dmpoaGhq6atUqHo+XkZGh4QiVlZUuLi7yOWYIIQEBAa2trXv37iWEjB07Nj4+fsKECZS6ST8BQF+QWwwJBQqYmy6XsNfjXjrIyckZOXIkj8djs9l2dna1tbWEkBEjRpSXl2vYa+jQoX5+fkqNYWFhGRkZ8phZLJaWUx4DgA6QWwwMBQqYjEuXLi1dunT16tXh4eH0VJhdrvau2xrxXa5HrzOKotasWUNPi0wIOXjwII/HI4RcuXIlNja2u0djsVh+fn6aPx4BQJeQW5QwJbfoa0IV051YBoxOm8mUKisrvby86CmhBAKBo6NjfX19l6u967ZGfGfr0SvR8j0fFBRUWFhIPxYKhRwOR2meq+rq6tDQ0C7nzqLn9r5+/bpi45YtW6ZNm0Y/jo6OTk9P7zIeqs9M1AaA3GLSuQV3UMA07N6929vbm54SKiwsrKmpSSAQKK5orXZ1U6U14p2dnQ8fPtzlXgkJCSEhIb1wEaSiosLJyUnpZqlAIEhLS9PtDqq7u3tFRYWeogPoi5Bb1GJCbkGBAqbhwYMH8p8g2tra+vr6Pnz4UMP2avukXha+74m2tjZ67TdFFhYWAQEBuh3QycmJXvcOAHSD3KIWE3ILChQwDa6urpcuXZI/tbS01Dydpdp0oJeF73vCx8enrq6OvuUrN3XqVJ0P2NDQ4OHh0eO4APou5Ba1mJBbUKCAaYiOjm5qarpx4wYhRCwW37lzhx4Ops0S9t1dI76z9eh7zsvLy9raurq6Wt5SX19PRyInFAozMzNVV5NXSj20qqoqnT8hAQBBbmFwbkGBAqZh1KhRu3fv/uCDD3799deEhIRdu3bRn3KSkpLi4uIiIiIcHR379+9fVFRECElISIiJiSkoKKD33bNnT3p6Op/Pz8/Pp79p1rzXpk2bsrKyeuMq2Gx2cnLy5cuX5S3Hjh2Lj4+XyWTylsLCQj6fT4ckV1NTs2nTJkLI7t27Hz9+LG8/evQon8/vjVAB+gjkFubmFn2NtsWveEBn2i+J3t7eXl5errTwuubV3nVYI17L9eh1GGlPUVRzc3NUVJTiBm1tbUq7FBYWXrt2rcsji0SiuLg4+VP8igdACXKL0i6mlVtwBwVMCZfL9fHxsbKyUmzUvNo79X9rxCt9B6xhLy6Xq98vjM+dO3fnzp2Ojg5CiJ2d3YoVKxQ/Rdna2ipuXFdX19zcrDp1kpL6+vrNmzfv3LmTENLQ0HD79u379+/rMWaAPgW5RY45uaWPLhYIfYQB1ojv0s6dO1taWpqamtzc3OgWX19fR0fHxsZGe3t71e0HDx4cExPT5WGfPn26ZcsWOgk+f/68ubl527Ztnp6e+g0eANRCbjEAFChgzgy28L0GPj4+qo0uLi49PKyvr6/8sZubmzxDAYABILcYAAoUMGeGXPgeAPoO5BYDwBgUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOPoeZBsXl6efg8IfYFIJCIm+OYpLi4mJhi2eaBffADNkFtMm75mfKNnvgMAk2DSM7Ea+8UDgE7pMbew0NtBg6ioKIJCHgB6DYvFEggEdKoBUIQxKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjIMCBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjIMCBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjIMCBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjIMCBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4LIqijB0DMEhOTs6uXbtkMhn99MGDB4SQYcOG0U8tLCwWLFjwt7/9zWjxAYCJ4/P5t2/flj/9/fffhw0bZm9vTz9ls9l79uxxc3MzUnTAIBxjBwDM4uvre/LkSaXGx48fyx+npaUZNiIAMCs8Hi8zM1Ox5fr16/LHXl5eqE6Ahq944H/4+fmNGjWqs7+OGDHCz8/PkPEAgJmJjY3t7E9WVlbvvfeeAWMBRkOBAsri4uIsLS1V2y0tLePj4w0fDwCYEx8fnzFjxrBYLNU/dXR0zJ071/AhATOhQAFlMTExEolEtV0sFkdFRRk+HgAwM/Pnz2ez2UqNLBbL39/f29vbKCEBA6FAAWVeXl6vvvqq0ucbFov12muvjRgxwlhRAYDZiImJkUqlSo1sNvvdd981SjzATChQQA3VzzdsNnv+/PnGigcAzIm7u/uECRMsLP7nf0BSqRT3aEERChRQIzo6Wv5LY5pMJkPuAAB9iYuLU7xNa2Fh8cYbb7i6uhoxJGAaFCighrOzc2hoqPwmCpvNnjRpEo/HM25UAGA25syZo/iUxWLhHi0oQYEC6sXFxSlO4hcXF2fEYADAzAwePPjNN9+UfwpisVgzZ840bkjANChQQL3Zs2dzOP9/Hj8LC4sZM2YYNx4AMDPz5s2jPwWx2ey//OUvjo6Oxo4ImAUFCqg3YMCAt956i8PhcDicv/71r4MGDTJ2RABgVmbNmmVlZUUIoShq3rx5xg4HGAcFCnRq3rx5UqlUKpVi8R0A0Lt+/fqFhYURQqysrMLDw40dDjAOChToVHh4uK2trY2NDZ1EAAD0i/7wM3PmzH79+hk7FmAeSoFAIDB2OADAdAKBgOoxY18EADCOUm5Rs5oxyhSQKykpoeefNnYg/yM1NZUQsmzZMmMH0j3FxcVpaWmm3r/0uFRKcnJyYGCgvo4GJurHH3+Mjo6WD8k3LuQWI1LNLWreE5iPC+RmzZpFCGFI7pDbu3cvMc03alpamimGrUiPBUpgYKCpvxrQcxEREdbW1saO4v9DbjEirQoUADmmlSYAYGaYU50A02CQLAAAADAOChQAAABgHBQoAAAAwDgoUAAAAIBxUKAAAAAA46BAAQAAAMZBgQIAAACMgwIFzN+kSZO++uorY0fRUxKJJD09va2traCg4OTJk1988UVwcPC9e/c071VYWOjv719aWko/LSkpOXnyZO8HC9AnILf0am5BgQLm78CBAx999JFeDlVaWtplv+0NUqk0KSkpMjLS2to6NTU1NDR01apVPB4vIyNDw16VlZUuLi7Xr1+XtwQEBLS2ttLTZQJADyG3yFt6I7egQAHzZ29vb2lp2fPjNDU1RUdHt7W19fxQ3ZWTkzNy5Egej8dms+3s7GprawkhI0aMKC8v17DX0KFD/fz8lBrDwsIyMjKMchUAZga5RZHecwsKFDBzpaWl8fHxK1euJITk5+cHBQVlZ2dHRUU5ODj89NNPhJDc3NyJEydu27YtNDTU1tZ21apVhJDs7Gwej1dRUSESiaKjo0NDQwkhe/bsKSsrS0tLKygoIIRERERs2LDBAJdAUdSaNWtmz55NPz148CCPxyOEXLlyJTY2trtHY7FYfn5+mj8eAUCXkFuU6D23oEABM+fr6/vs2TO6qA8PD799+/bvv/+em5u7cuXK9evXE0Jmzpx5+fLl2tra48ePf/3112vXrq2uro6Nja2pqZFIJG5ubtOnT6+vryeEJCUlEUJSUlLCw8MJITNmzAgKCjLAJTx58uTRo0eenp6KjTU1NWKxWIckQghxd3c/evSonqID6KOQW1TpN7egQAHz5+DgQD+wsrKys7MLCgricDjjx48XCoWEEBsbGxsbm6lTp3I4nMTERGdn58OHD7PZbPnuna2YmJCQEBISYoD4KyoqnJycWCyWYqNAIEhLS1Nq1JK7u3tFRYWeogPou5BblOg3t6BAgT6KzWZTFKXaHhgYeOvWLQ076tZve6KtrU0qlSo1WlhYBAQE6HZAJyenhoaGHscFAGogt/Q4rv+LRF8HAjAPd+/e9fb21rCB4ZOIj49PXV2dRCJRbJw6darOB2xoaPDw8OhxXADQDcgt3YUCBczfixcvOjo66MdSqZT+cNPR0SGTyeTb0F8k19fXV1VVRUZGEkI8PDzKysoIIeXl5S0tLYQQNpvN5XIbGhra29sJIYcOHSopKTFA/F5eXtbW1tXV1fKW+vp6OjY5oVCYmZkpFouV9lVKPbSqqiqdPyEBgBxyixL95hYUKGDmLl68WFxcfPr06Zs3b544cUIkEhUUFNTV1WVnZ7e0tNBj5gkhe/bsSU9P5/P5+fn5/fv3J4QkJSXFxcVFREQ4Ojr279+/qKiIEJKQkBATE0PvtWnTpqysLANcApvNTk5Ovnz5srzl2LFj8fHxikmwsLCQz+fTQcrV1NRs2rSJELJ79+7Hjx/L248ePcrn83s/cABzhtxCeju3UAoEAoFSCwADRUZGRkZG6vGAdnZ2Z86cqa6ulslkiu0tLS0URYnFYsXG1tZW+kF7e7vS9pr1pH81NzdHRUUptrS1tSltU1hYeO3atS4PJRKJ4uLidAuDoihCiEAg0Hl3vR8HQI+QWygm5RbcQQEgFEXJZDJnZ2el74DpjztKI+379etHP+ByuQb7ztjOzm7FihWKH6psbW0VN6irq2tubladOklJfX395s2bd+7c2StRAsD/Qm7pCRQo0NcJBILW1tacnJwnT54YOxZNfH19p0yZ0tjYqPavgwcPjomJ6fIgT58+3bJli5WVlb6jAwBlyC09pP5H2AB9R1hYGD1OjcvlGjuWLri4uPTwCL6+vnqJBAC6hNzSQyhQoK+T31YFANAj5JYe6vZXPN988w2bzb5//75u58vNzeVwOKdPn9Ztd826G1tra+vSpUvppQc6a9TXatrnz5/38PBISUnp+aG6lJ+fP3To0AEDBixbtuzjjz/++9//HhQU9P7773e5I5NfQADjYnLvMFh60Tm3EGa/gMBQiiNmtRkJ/Pz5c0LIvXv3dB6p6+zsfOrUKZ1310CH2E6dOuXi4qKhsaGhoaOjQy/hzZs37+OPP9bLoboUExPj7+8vf/r8+fPk5OQu92L4Cyin9x6ct8YAACAASURBVJH2hmEev5IjffVXPAzvHQZLL7rlForxLyANucWIVHNCt7/isbCwID2b8E5xJQL96m5sMpnMwsJCaXulRnt7e32Fx2azDTYwW2kYNpvN/vLLL7vci+EvIIARMbx3GCy96JZbCONfQGAgHX/Fc/z48cDAQCcnp++//55uuXTp0tKlS1evXh0eHi6fAk9tI621tXXhwoXLli3TMF9eUVERn88PCQn59ttvCSGnTp2aNWtWdnY2n893d3dPT0+/cOFCZGSku7v74cOHNcSmdBxCyOXLl5OTk9PS0rZt2yZ/r6s2drmaNiHk9u3ba9asWbt2bVBQ0Icffnjw4EFtXsCtW7cKBILFixfTB8/Kyho5cuSaNWskEklzc3N0dDQ9K45q5Pv373/zzTdzcnLGjBmTkpKi5arcX3zxha2trTm9gGD21K5KTwjZunXrDz/8EB4e/vPPP9MtSu9PpT7S2fGRXrp83bRJL3RuMbMXEBhCx0GyT548OXbs2I4dOz744IPo6Ojm5uaYmJhr1671798/Ly9vypQpFRUVHR0dqo3ytR+vXLny2muvJSYmdnaKe/fuHTx4cMeOHc+ePePxeG+++WZgYGB8fLyDg8OOHTv2798fFxe3a9euvXv3/utf//r3v//9l7/8RW1slZWVSsdxdXWdO3furVu3rKysNm7ceP78eULIs2fPVBvp1bTpIj08PJzP59OraX/77bfr16+Pjo4mhHz88cfr1q3z9/e/ePEiIWTatGldvnoikWjt2rVNTU0REREDBw78/PPP58+fn5qaamVlxeFwBgwY4Ojo+NZbb6m+AqNHjw4KCoqKivL19c3NzW1paamoqBgxYoTaswiFQnpGv3v37tXW1n755Zdm8wLSr2FeXp42WzJHcXExIcTkwjaW2NjY+fPny1el37hxIyHk7t27Fy9ezMvLmzlzJj3npmo3Ueojag+O9KLN6zZjxgy16UU1txBCzOkFNLlOaq65RccCZeHChf369UtOTl61atWJEydKS0u9vb3pmWfCwsJiY2MFAkFTU5Nq4+LFiwkhhw8ftre31zykKzMz89mzZ+vXryeETJgw4cKFC6NHj3ZycgoNDWWz2VOnTn3x4sXkyZNZLNakSZN++eWXzmI7c+aM0nFaWlpGjRpF/1w7MDAwPT2dEPLDDz+oNpLOV9Nes2YNIUQmkx0/fvyTTz6h97p7966NjU2Xr56rqys9TLi4uFgqlQqFQi8vr6VLl65du/Yf//hHTU0NvdiS2leAx+MNHjw4PDzc39+fEPLGG290dhY3N7fvvvuOEHLv3r1PP/2UEMLlcs3jBSSEXLhwYe7cudpsyTQmGrbhqV2VfsCAAQUFBd98801SUtKsWbOIum6SkJCg2EfUQnrR5nVLSEhQewrV3ELMKL0gtzCHjgUKfYfNwsJi3LhxV69eFQqF9PeLhBBbW1tfX9+HDx82NjaqNtJPjx07ZmlpuXz5cg3jUe7evRseHv7ee+8RQj777DOlvw4YMED+2NLS8o8//ugsNtXjLFq0yNXVVb4xvf3NmzdVGzsjX03bwsIiJCRk3759b7zxxqNHj+bNm6dhL8UIm5ub//nPfy5cuJDFYtGrXcfExKSkpBQVFT1+/DgqKkrDK2BhYaE0/2BnZ6GvYsSIEf/617+U/mrSLyAhJDIycu/evVpuzBB5eXlz586l1K3DbkIMv+CqImdn58zMzEWLFuXn5+/fv9/W1lZtN+myjyC96Pa6KV2C2txCTPwFRG4xFtV/1p7OJNve3j5ixAhXV9dLly7JGy0tLV1dXdU20o/XrFnT1NSkeWiVv78/fR+Pdu/ePd1iUz3OsGHDLly4oLSx2kZt5ObmXrt2LTMzMzEx8c0339Rml5s3by5YsGDt2rXDhw+XN9rY2CxYsGD79u1CoXDYsGFEH6+A3NixY58+faq0RmWXGPsCQp9VU1Mzb968K1euNDY2Ll26lOjaTZBe9JVedMsthMEvIDCHjgUKXak1NzfX1tbOnDkzOjq6qanpxo0bhBCxWHznzp3Zs2erbaT3tbGxyc7O3rBhw8mTJzs7xdtvv52VlfXdd9+JxeJz585dv36dEEKvn0T+bwFregFoxXUXVWNTPc7MmTPLysquXbtGCHny5Am9MJLaRqLFatorVqwQCASLFi169dVXNb9oUqmU/jTz22+/0StF3bp1SyqVypexXrJkyZEjR+Qz+ql9Bejrle/S2arcbW1t9Lnkuyxbtsze3t6kX0Doa1RXpS8pKbl48eLo0aO///77hoYG0nmiUF0dXhHSizavm9r00lluMfUXEJhI8TfH2vyWWiaTxcfHz5kzJzU19Z133rlx4wbdnp2dHRIScujQoXnz5uXn53fWSI/iSUxMrK2tDQsLGzRo0Pbt2zs7EX2L0tPT87PPPpPJZMXFxVwuNzo6uq6ubvPmzYSQTz/9tK6ubuHChWw2+/Dhw2pjUz0ORVHLly93cHCYNWvWggULvLy8du7cqbbxwoULw4cPf+WVV27cuHH8+HEWixUXF1dbW7tw4UJCyC+//EJRVHBwMIvFGjhwoLu7e1xc3LNnz9RezoULFzw9PceOHVtWVvbgwQMXF5fXXnttx44dgYGB06ZNq6urozd76623KisrO3sFKIr68ccfCSFz5sx5+PAhRVGhoaHLli1TOld+fv6QIUMsLCzi4+MXL1789ttvu7u7e3t7m/QLqAhzFRgRMeA8KJs2bbK2tg4PD09LSxs9evSvv/565MiRyZMnFxUVbdmy5eTJk5S696dSH1EL6UWb1001vXSWWyiKMukXUA65xYhUc0K3CxRaR0dHY2OjUmN7e3t5efmLFy+6bNReY2Njd/dVG5vqcVpbW+nqW3FVa7WNGrS2tq5bt66ysvLatWtnz55NTU3Nzs7WZkexWCyRSCiKkkgkiudasmRJl5Er6u6q3Npg/guIJGJEhixQKJVV6elPybW1tUqb6ZAodNuL+b2DQnpR2L67LyByixGp5gQdB8laWloOGjRIqZHL5fr4+GjTqGTGjBlKLSwW68CBA4QQ1bPoFptqi9pVErq7dMKmTZuePn06ZMiQIUOGUBR1//79sWPHargcOflINHqYsEgkqqioKCsrmzlzZpeRK+qNNaiM/gJ26whg3pRWpafH3Q8ePFhpMw3dBOkF6QXpxUQxYrFA+YRLJmfRokXvv/++n59fQECAi4tLZGTkmDFjdLicX3755dNPP01JSZkyZUpvxMlYal9AYwcFZgXpBekF6cVEMaJAMV2urq6HDh2iZ1/uyXGWLFmSmJjYw4OYIn29gADmB+mlh5BeTB3+2fRAL+/+vtyF+vK1A2iG9NJDffnaTR3+5QAAAIBxUKAAMI5EIklPT6cnkygoKDh58uQXX3wRHByseTatwsJCf3//0tJSzQc/cODAtm3b9u3bt2TJEnqOrJKSEg0zEgGA2TCt3IICBYCUlpbqMJOmbnt1SSqVJiUlRUZG9uvXTyqVpqamhoaGrlq1isfjZWRkdLZXZWWli4uLfDa/zojF4hUrViQmJkZGRsbHx69evZoQEhAQ0NraanLTewMwH3JLT3ILChTo65qamqKjo+mPFL29lzZycnJGjhzJ4/EIIWw2287Orra2lhAyYsSI8vLyzvYaOnSon59flwdva2u7c+fOzZs3CSENDQ2Ojo50e1hYWEZGRm9cDkCfhdxCepZbUKCA2bp06dLSpUtXr14dHh5Oz9idnZ3N4/EqKipEIlF0dHRoaCghZM+ePWVlZWlpaQUFBbm5uRMnTty2bVtoaKitre2qVau02SsiImLDhg16iZmiqDVr1tCLQtAOHjxIJ5QrV67Exsb28PiDBg2aPn16ZGRkSUnJzp075Su9sVgsPz8/DZ+iAEAOuUVVr+QWxVnbzGM2OjB72sz2WFlZ6eXlRc9DKhAIHB0d6+vrJRIJIeTmzZsURWVlZY0ZM4aiKLqxrKyMoqjnz59zOJzPP/9cLBZnZGSwWKyqqqou99q1a9fp06e7DFub/iUUCjkcjupEmdXV1aGhoZon0KSXPrl+/brmUzx79mzcuHGEkJ9//lmxfcuWLdOmTdO8L2XwmWQBDAm5RS1j5RbcQQHztHv3bm9vb3oe0rCwsKamJoFAQE+sSVO7rLyNjY2Njc3UqVM5HE5iYqKzs/Phw4e73CshISEkJEQvYVdUVDg5OakuOy4QCNLS0jSvMq+l+/fv+/n5RURExMXFnTt3Tt7u7u5eUVHR8+MDmDfkls7oPbdgojYwTw8ePJDPf2Bra+vr6/vw4UMN26vtn4GBgbdu3eruXj2htFSsnIWFRUBAgF5OERsbe/bsWXt7ez6fz+fz6fXGCSFOTk70+sAAoAFyS2f0nltwBwXMk6ur66VLl+RPLS0tXV1dNWyvNh3cvXvX29u7u3v1hI+PT11dHX2PV9HUqVP1cvyKioo//vjDwcGBxWKtW7euvLy8urqa/lNDQ4OHh4dezgJgxpBb1OqN3IICBcxTdHR0U1MTXcKLxeI7d+7Qo8M8PDzKysoIIeXl5S0tLYQQNpvN5XIbGhra29vpfekB5/X19VVVVZGRkV3udejQIXqgXM95eXlZW1vLOzatvr6ePjtNKBRmZmbS3worUk09qlsOGzZMKpU+ffqUEGJtbT127Fh6lBwhpKqqSl8fpADMGHKL2i17I7egQAHzNGrUqN27d3/wwQe//vprQkLCrl276E85SUlJcXFxERERjo6O/fv3LyoqIoQkJCTExMQUFBTQ++7Zsyc9PZ3P5+fn59PfNGvea9OmTVlZWXoJm81mJycnX758WbHx2LFj8fHxMpmMflpYWMjn8+kY5GpqajZt2kQI2b179+PHjzvb0tLSMjU1dd26dQKB4Lvvvlu/fr38T0ePHuXz+Xq5CgAzhtyidsteyS2KI2bxKx4wCdqMtKe1t7eXl5e/ePFCsZEefi8WixUbW1tb6Qd2dnZnzpyprq5WGtauYa/29nbNY+BpWvav5ubmqKgopca2tjbFp4WFhdeuXevyUBq2bGxsVHwqEoni4uK0OSDBr3jAfCG3MCq34A4KmDMul+vj42NlZaXYSH9wURoz369fP/oBRVEymczZ2VnpO2ANe3G5XD1+YWxnZ7dixQqlj022trbyx3V1dc3NzdpMnaRhy0GDBskf19fXb968eefOnT2IGqBvQW4xQG5BgQLwXwKBoLW1NScn58mTJ0YMw9fXd8qUKY2NjWr/Onjw4JiYGG2Oo+WWT58+3bJli1KqBQA9Qm7RAX5mDPBfYWFh9Dg1Lpdr3EhcXFwMdi5fX1+DnQugb0Ju0QEKFID/kt+MBQDQI+QWHeArHgAAAGAcFCgAAADAOChQAAAAgHFQoAAAAADjqBkkO2fOHMPHAaC9CxcuEBN8o4pEImKCYfee1NTUvXv3GjsKgP9CbmEUFj19G624uPjrr782YjRgEtrb20+dOjVhwgTFCXmg71i+fHlgYGAPD2J+yRS0V1xc7OjoqHm1POiDlHLL/xQoANqgKMrJyWnFihXJycnGjgUATI+np2dSUlJKSoqxAwFGwxgU6DYWixUUFHTmzBljBwIApkcikTx9+tTT09PYgQDToUABXQQHB585cwa33wCgu548eSKRSFCgQJdQoIAuQkJCamtrb926ZexAAMDEPH78mBDi4eFh7ECA6VCggC5effVVeulwYwcCACbm0aNHXC6Xx+MZOxBgOhQooAsOhzNhwgQUKADQXY8ePXJ3d7ewwP99oAt4i4COgoODT506ZewoAMDEPHr0CANQQBsoUEBHwcHBIpHo0aNHxg4EAEwJChTQEgoU0NGECROsra1/++03YwcCAKbk8ePHKFBAGyhQQEfW1tbjxo3DMBQA6BahUIif8IA2UKCA7kJCQnAHBQC0V1tb29bWhjsooA0UKKC74ODg27dvV1ZWGjsQADAN9Kg1FCigDRQooLugoCA2m33u3DljBwIApuHRo0cWFhZubm7GDgRMAAoU0N2AAQP8/f0xDAUAtPTo0aOhQ4daWVkZOxAwAShQoEdCQkJQoACAlvAbY9AeChTokeDg4GvXrjU1NRk7EAAwAfiNMWgPBQr0SHBwMEVR58+fN3YgAGACcAcFtIcCBXrEycnJx8cH3/IAgDYePXqESVBASyhQoKeCg4MxGwoAdKm1tbWhoQF3UEBLKFCgp4KDgy9fvvz8+XNjBwIAjIZJUKBbUKBAT02aNEksFl+8eNHYgQAAo9EFCr7iAS2hQIGecnNz8/T0xLc8AKDZo0ePHBwc7OzsjB0ImAYUKKAHmA0FALqE3xhDt6BAAT0IDg4uLi7u6OgwdiAAwFz4CQ90CwoU0IOQkJDnz5///vvvxg4EAJgLk6BAt6BAAT0YNWrU0KFD8S0PAGiAAgW6BQUK6EdQUBAKFADojFgsrqqqQoEC2kOBAvoRHBx89uxZmUxm7EAAgImEQqFUKkWBAtpDgQL6ERwc3NjYeOPGDWMHAgBMhFnaoLtQoIB++Pv7Dxo0CLOhAIBajx8/trGxGTx4sLEDAZOBAgX0w8LCYuLEiRiGAgBq0b8xZrFYxg4ETAYKFNAbrBoIAJ3BT3igu1CggN6EhIRUVVXduXPH2IEAAOOgQIHu4hg7ADAfr7/+uq2t7W+//TZkyJBz586dOXPm9OnTK1aseOutt4wdGgAY2rZt29rb2z09PT08PDw8PB49ejR58mRjBwWmhEVRlLFjAHNQW1t79uzZZcuWvXjxoqamRiaTWVpaisXiS5cuvf7668aODgAMbenSpenp6YQQ+v8ybDbb2dl51KhRw4cP9/DweOmll8LCwhwcHIwdJjAXChTokWvXrmVkZBw/fvzevXuEEC6X297erriBSCRydXU1UnQAYDS5ublxcXGqcyNxOByZTObk5PTw4UNra2ujxAYmAV/xQI+4urrm5eU1NjbST5WqExaLxePxjBEXABjZhAkT1M7cKJFIOBzOp59+iuoENMMdFOip3NzcefPmqX0jOTo61tXVGT4kAGACR0fHhoYG1XYHBwehUGhra2v4kMCE4Fc80FOxsbHTp0+3tLRU/dOQIUMMHw8AMMTEiRMtLJT/L0PfPkF1Al1CgQJ6sGPHDi6Xq9ru4eFh+GAAgCEmTpzI4SgPJOjXr19iYqJR4gHTggIF9MDFxWXjxo1KH5U4HI67u7uxQgIAowsMDOzo6FBs4XA4n3zyiZ2dnbFCAhOCAgX0IzEx8U9/+pPiFz1sNnvo0KFGDAkAjOv1119ns9mKLdbW1osXLzZWPGBaUKCAflhYWPzwww+KLVKp1MXFxUjhAIDx9evXb9SoUfKnHA4nJSVl0KBBRgwJTAgKFNAbb2/v1atXy7/okUgkuIMC0MeFhITIb6xaWVktXbrUuPGACUGBAvr0j3/8Y8yYMfJhcbiDAtDHTZgwQSqVEkIsLS0/+ugje3t7Y0cEJgPzoICeXblyZfz48fQETU+ePEGNAtCXVVRU0N/y2NjYiEQizG0P2sMdFNCzcePGLV++nBBiYWGBaWQB+riRI0cOHDiQEJKcnIzqBLrFxKa6z8vLM3YI0DU/P7/Bgwe/ePEiPz/f2LGA/k2cONHNzc0w5youLhYKhYY5F/SSl156qayszMvLCwncVBiyj2tCmRRjv1oAQAQCgcG6fGRkpLEvF6DPMWQf18DE7qAQQgQCQVRUlLGjgK4dPHjwnXfe6Y0jz5kzhxCyd+/e3jh478nLy5s7dy5l4nU2i8Uy8BkjIyNN7t8aFJ07d2748OHdWvgCfdyIDN/HO2N6BQqYil6qTgDAtAQFBRk7BDBJGCQLAAAAjIMCBQAAABgHBQoAAAAwDgoUAAAAYBwUKAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYEC5m/SpElfffWVsaPoBolEkp6e3tbWRggpKCg4efLkF198ERwcfO/ePQ17FRYW+vv7l5aWaj74gQMHtm3btm/fviVLlpw/f54QUlJScvLkST3GD2Bg6OOKzKaPo0AB83fgwIGPPvpIL4cqLS3VnEF6TiqVJiUlRUZG9uvXTyqVpqamhoaGrlq1isfjZWRkdLZXZWWli4vL9evXNR9cLBavWLEiMTExMjIyPj5+9erVhJCAgIDW1laTm1YcQA59XM6c+jgKFDB/9vb2lpaWPT9OU1NTdHQ0/aGn9+Tk5IwcOZLH4xFC2Gy2nZ1dbW0tIWTEiBHl5eWd7TV06FA/P78uD97W1nbnzp2bN28SQhoaGhwdHen2sLCwjIyM3r40gF6CPi5nTn0cBQqYudLS0vj4+JUrVxJC8vPzg4KCsrOzo6KiHBwcfvrpJ0JIbm7uxIkTt23bFhoaamtru2rVKkJIdnY2j8erqKgQiUTR0dGhoaGEkD179pSVlaWlpRUUFBBCIiIiNmzYoN9oKYpas2bN7Nmz5S0HDx6kE9mVK1diY2N7ePxBgwZNnz49MjKypKRk586d//rXv+h2Fovl5+en4dMbAGOhjysypz6OAgXMnK+v77Nnz+jPDeHh4bdv3/79999zc3NXrly5fv16QsjMmTMvX75cW1t7/Pjxr7/+eu3atdXV1bGxsTU1NRKJxM3Nbfr06fX19YSQpKQkQkhKSkp4eDghZMaMGXpfBe3JkyePHj3y9PRUaq+pqRGLxT1PXoSQPXv2DBw4cOzYsfPmzXvllVfk7e7u7kePHu358QEMDH1cidn0cRQoYP4cHBzoB1ZWVnZ2dkFBQRwOZ/z48UKhkBBiY2NjY2MzdepUDoeTmJjo7Ox8+PBhNpst353DUb/od0JCQkhIiH5DraiocHJyUl3uXCAQpKWl6WUZ9Pv37/v5+UVERMTFxZ07d07e7u7uXlFR0fPjAxge+rgis+nj6v9VAMwem82mKEq1PTAw8NatWxp21EsG6UxbW5tUKlVtt7CwCAgI0MspYmNjz549a29vz+fz+Xz+jRs36HYnJ6eGhga9nAKACdDHTb2P4w4KwP+4e/eut7e3hg16NXn5+PjU1dVJJBKl9qlTp+rl+BUVFX/88YeDgwOLxVq3bl15eXl1dTX9p4aGBg8PD72cBYDJ0MdNBQoUMH8vXrzo6OigH0ulUvpDVUdHh0wmk29Df4FdX19fVVUVGRlJCPHw8CgrKyOElJeXt7S0EELYbDaXy21oaGhvbyeEHDp0qKSkRL+henl5WVtbyxMKrb6+no6EJhQKMzMzxWKx0r6qKU91y2HDhkml0qdPnxJCrK2tx44dS4/OI4RUVVXp6wMcgIGhj8tbzKmPo0ABM3fx4sXi4uLTp0/fvHnzxIkTIpGooKCgrq4uOzu7paWFHqtPCNmzZ096ejqfz8/Pz+/fvz8hJCkpKS4uLiIiwtHRsX///kVFRYSQhISEmJgYeq9NmzZlZWXpN1o2m52cnHz58mXFxmPHjsXHx8tTbWFhIZ/Pp+ORq6mp2bRpEyFk9+7djx8/7mxLS0vL1NTUdevWCQSC7777jh5CSDt69Cifz9fv5QAYAPq42fZxyqQQQgQCgbGjACOLjIyMjIzU4wHt7OzOnDlTXV0tk8kU21taWiiKEovFio2tra30g/b2dqXtNRMIBNr0uObm5qioKKXGtrY2xaeFhYXXrl3T5qSdbdnY2Kj4VCQSxcXFaXNAA/dBvf9bg0lAH+87fVwD3EEBIBRFyWQyZ2dnpe+e6Y9ZSiP8+/XrRz/gcrm98V21nZ3dihUrlD632drayh/X1dU1NzdrM2WThi0HDRokf1xfX7958+adO3f2IGoARkMfN8U+jgIF+jqBQNDa2pqTk/PkyRNjx/L/+fr6TpkypbGxUe1fBw8eHBMTo81xtNzy6dOnW7ZssbKy6l6UACYCfdxE+zh+Zgx9XVhYGD0+jsvlGjuW/3JxcTHYuXx9fQ12LgDDQx830T6OAgX6OvntXAAwS+jjJsrcvuL55ptv2Gz2/fv3dds9NzeXw+GcPn1av1Hp3fnz5z08PFJSUgxwrvz8/KFDhw4YMGD58uUff/zx+++/HxAQ4O3tbbAAAPomg3VztX38vffeM2SeAVBlbndQ3n///Q8//FDn3WNjY5ctW6bHeDpTWlpqa2s7fPhw3XafOHEivbSVAcyePTs/P7+srOzrr7+mW9ra2lJSUuhbphrIr7GHFwtginr+tjdYN++sj3cZgOI1opuD3pnbHRQLCwvSs3kAFRdo6CV6WdGbzWb36nSHihSHlxNCLC0tN2/erDkA+TUaZvlyAEbR19veYN1cbR/XHIDiNaKbQ28wtwKFdvz48cDAQCcnp++//55uuXTp0tKlS1evXh0eHi6fGVBtI621tXXhwoXLli3rbBrBn3/+2dHRkb7d0tDQMHnyZHpin6KiIj6fHxIS8u2338o3vnr1akJCwsaNG995552mpialFb1Vw9i/f/+bb76Zk5MzZswYbe6vbt26VSAQLF68mF5wPCsra+TIkWvWrJFIJM3NzdHR0fQ0PqqxKZ1Iy7XFv/jiC6XvdJUCIAqrlkdGRiperGoMapdHB9BednY2j8erqKgQiUTR0dHyD/1bt2794YcfwsPDf/75Z6Ly3tOml2nfzXu7jxNdu7nqibTp5qp9XDUAxWtUul6lGNDHQTfm9hUP7cmTJ8eOHduxY8cHH3wQHR3d3NwcExNz7dq1/v375+XlTZkypaKioqOjQ7VRviTmlStXXnvttcTExM5OMWPGjKysLHrNJwcHh4CAgPDw8Hv37h08eHDHjh3Pnj3j8Xhvvvnm6NGj6+vrExMTi4uLLSws/vrXv+7cuXPZsmXJyckpKSkvv/xyVVWVahhBQUFRUVG+vr65ubldfpMiEonWrl3b1NQUERExcODAzz//fP78+ampqVZWVhwOZ8CAAY6Ojm+99Zba2JROVFFRMWLECLVnEQqFfD6foqiysjLFn9erDcDS0jIpKYm+Rm9vbw6Hpu1UmwAAIABJREFUQ1+s2hjCw8P5fD69PPq33367fv366OjoLv+JL1y4MGfOnC43YxSRSEQIMbmwmS82Nnb+/PkSicTNzW369OkbN24khNy9e/fixYt5eXkzZ84sKChQfe9p08u07OY8Hq9X+zjpQTdXPdGMGTPUdnMNfVxtAPI+/vLLL0ulUvlj1RjQx0E35nkHZeHChf369UtOTuZwOCdOnNi9e7e3tzc9IU9YWFhTU5NAIFDbSO9++PDhS5cuaahOaIsWLfrpp5/EYvGDBw9efvllQkhmZuazZ8/Wr1+fnp4+YcKECxcuEEJ27tw5fvx4+runvLy8v//974oHURsGj8cbPHhweHi4v7//G2+8oTkMV1dXelRvcXGxVCqllxdfunTpjh07ZDJZVVUVvTqU2tiUTqRhbXE3N7fv/o+9vX2XAailNga1y6MDaE/xa1n5jFsDBgwoKCj45ptv7OzsZs2apfre07KXadPNe7uPkx50c9UTddbNNfTxzgJQSzUG9HHQjXneQaG/NLWwsBg3btzVq1eFQiGdOwghtra2vr6+Dx8+bGxsVG2knx47dszS0nL58uWax6NMmzaNy+UWFRXdvn17wYIFhJC7d++Gh4e/9957hJDPPvuM3qysrMzNzY1+TCcp+gMZHeSDBw/UhmFhYaE0uaGGi21ubv7nP/+5cOFCFotFHzwmJiYlJaWoqOjx48dRUVGdxab9iVj/55VXXlm+fHmXAcj/pPigsxjkOlseXdWECRP27t2rzZbMkZeXN3fuXJMLW4nBRj71kLOzc2Zm5qJFi/Lz8/fv36/2vafNm1+bbv7uu+/2ah8nPevmPe/jnQVA/vf9oE03Rx9nPub0cfO8gyLX3t4+YsQIV1fXS5cuyRstLS1dXV3VNtKP16xZ09TU9OWXX2o+uIWFxbvvvrtnz55nz57RHzj8/f3Pnz8v3+DevXuEkEGDBh06dEjeWFlZST+g3wQawtDSzZs3FyxYsHbtWsXx8zY2NgsWLNi+fbtQKBw2bFhnselm7NixXQZAUypQ9BgDgGY1NTXz5s27cuVKY2Pj0qVLdX7vadPNe7uPE4N3c6U+3lkARF2Bgm4O+mKeBQpdoTc3N9fW1s6cOTM6OrqpqenGjRuEELFYfOfOndmzZ6ttpPe1sbHJzs7esGHDyZMnNZ8oPj7+559/Hj9+PP307bffzsrK+u6778Ri8blz565fv04ImT59eklJyX/+85/29vb9+/eXlpYqrujdWRgymUx1rW1FUqmU/hDz22+/0etd3bp1SyqVyvdasmTJkSNH5JMVqo1N6USdrS3e3Nzc2trarQDk1ygWi+UX21kMnS2PDqAlDw8PerX68vJyeqRFSUnJxYsXR48e/f333zc0NKh973XZy2hddvNe6uNET91c6URqu7naPq45AMVrVHysNgb0cdCFMVYo1B3papVFmUwWHx8/Z86c1NTUd95558aNG3R7dnZ2SEjIoUOH5s2bl5+f31ljXl4eISQxMbG2tjYsLGzQoEHbt2/XHNKcOXM6OjrkZ6fvf3p6en722WfyZTDpcaNcLvezzz6jWxYvXuzp6ZmXl6c2jB9//JEQMmfOnIcPH6o96YULFzw9PceOHVtWVvbgwQMXF5fXXnttx44dgYGB06ZNq6urozd76623KisrNcSmdKLQ0NBly5YpnWvfvn1DhgxhsVj//ve/5VeqTQDya5Q/UBvD8ePHWSxWXFxcbW3twoULCSG//PKL5tfcRFe41XKlU4brsg/ql5b/1ps2bbK2tg4PD09LSxs9evSvv/565MiRyZMnFxUVbdmy5eTJk6rvvS57maIuu7ne+zilp26ueiLVbq62j2sTgOI1aujm6OOmxcB9XAMTeym1fOE6OjqUVpqmKKq9vb28vPzFixddNnaLYn+mNTY2qh7wjz/+UFpNW76id8/DEIvFEomEoiiJRKK4OPiSJUu0iU0xjG6tLd5lAPJrVLxYzTFoA8nLiJhZoFAURX++F4vF9FP6I3ttba3iNjq/97Tp5r3axymmdnPFa9RjN0cfNyLmFCjmOUjW0tJS9WdyXC7Xx8dHm0YlM2bMUGphsVgHDhyQn0vpr6qnJoRYW1srtShOM6AhDM1np8lHwNGjekUiUUVFRVlZ2cyZM7WJTTEMDX/VQCkAOfk1Kl6s5hgAdEOPTpW/FelxqYMHD1bcRsN7r+fdXOc+3uXZaczs5orXiG4O+mWeBYp+0bM8mdDZf/nll08//TQlJWXKlCm9ERKA+UE3B2AaFChmaMmSJYmJifJfNgKA+UE3B7OHN7d5QtoCMHvo5mDecAcFwDRIJJIdO3a89957J06c6N+//2+//Xbs2LEffvhBw/qxBw4cePLkyZAhQ06cODFv3ryJEyeWlJQ0NjZOnjzZkJEDgDbQx5WgAAcgpaWlOkwnpdteupFKpUlJSZGRkdbW1qmpqaGhoatWreLxeBkZGZ3tIhaLV6xYkZiYGBkZGR8fv3r1akJIQEBAa2urqc90CdBd6OOmCAUK9HW6rRRv4PXlc3JyRo4cyePx2Gy2nZ1dbW0tIWTEiBHl5eWd7dLW1nbnzp2bN28SQhoaGhwdHen2sLCwjIwMg0UOYHTo4yYKBQqYLdU17rOzs3k8XkVFhUgkio6ODg0NJf+7anxubu7EiRO3bdsWGhpqa2u7atUqbfbSZv36nqAoas2aNfT0o4SQgwcP8ng8QsiVK1diY2M722vQoEHTp0+PjIwsKSnZuXPnv/71L7qdxWL5+flp+FgGYCrQx828jxt7IpbuIYyZQAaMSJtJnCorK728vOjJuwQCgaOjY319vUQiIYTcvHmToqisrKwxY8ZQFEU3lpWVURT1/PlzDofz+eefi8XijIwMFotVVVXV5V67du06ffp0l2HrPImTUCjkcDhKk2tVV1eHhoZqnnHr2bNn48aNI4T8/PPPiu1btmyZNm2aDpFQDJ6oDcwJ+jjVZ/q4BriDAuZJ7Rr3ihNMqV3f1cbGxsbGZurUqRwOJzEx0dnZ+fDhw13u1dn69fpSUVHh5OSktMSoQCBIS0vTvO7o/fv3/fz8IiIi4uLizp07J293d3evqKjorXABDAJ9nJh7H8eveMA8dbbGfWfUZoHAwMBbt251dy+9a2trk69uL2dhYREQEKB5x9jY2LNnz9rb2/P5fD6fT69XRwhxcnJqaGjolVgBDAV9nJh7H8cdFDBP3V3jXm0aunv3rre3d3f30jsfH5+6ujr6hrPc1KlTNe9VUVHxxx9/ODg4sFisdevWlZeXV1dX039qaGjw8PDorXABDAJ93Oz7OAoUME+drXHv4eFRVlZGCCkvL29paSGEKK4UT+9Lj36vr6+vqqqKjIzsci+169frkZeXl7W1tTz10LHR8cgJhcLMzEyxWCxvGTZsmFQqffr0KSHE2tp67Nix9LA7QkhVVVWXn8wAGA593Oz7OAoUME+jRo3avXv3Bx988OuvvyYkJOzatYv+dJWUlBQXFxcREeHo6Ni/f/+ioiJCSEJCQkxMTEFBAb3vnj170tPT+Xx+fn4+/Q235r02bdqUlZXVe9fCZrOTk5MvX74sbzl27Fh8fLxMJpO3FBYW8vl8OjCapaVlamrqunXrBALBd999t379evmfjh49yufzey9gAANAHzf/Pm7sUbrdQxgzuhiMSPtfdqhd454e9i8WixUb5SvF29nZnTlzprq6WmnwvIa9tFy/vidLsTc3N0dFRSm2tLW1KW1TWFh47do11X0bGxsVn4pEori4ON3CoPArHjAI9HGqz/RxDXAHBcwZvca9lZWVYiP9gUlprL58pXiKomQymbOzs9J3zxr24nK5vf1FtZ2d3YoVKxQ/w9na2ipuUFdX19zc7Ofnp7qv4sL39fX1mzdv3rlzZ++FCmBI6OPEfPs4ChSA/xIIBK2trTk5OU+ePDF2LMp8fX2nTJnS2Nio9q+DBw+OiYnp8iBPnz7dsmWLUjYH6DvQx00IfmYM8F9hYWH0+Dgul2vsWNRwcXHp4RF8fX31EgmAiUIfNyEoUAD+S34TGADMEvq4CcFXPAAAAMA4KFAAAACAcVCgAAAAAOOgQAEAAADGQYECAAAAjGN6v+KZO3fu3LlzjR0FGJ9hFvHSOxMN24j27duHF61vMtF/dxMNm4FYFEUZO4ZuyMvLM3YIYPIkEsnnn38uEonmzZs3depUZJPumjhxopubm2HOVVxcLBQKDXMuU/THH3/89NNPhw8fHjduXEpKirHDATNhyD6ugYkVKAB6IZFItm7dunLlykmTJn3//fdM6IoA3VVYWPjBBx+0tLR89dVX77//PkptMDMYgwJ9EYfD+eSTT86ePfv48WNfX98ff/zR2BEBdENVVdX8+fPDw8P/9Kc/3b59e9GiRahOwPygQIG+a/z48VevXuXz+e+++25UVFRDQ4OxIwLoAkVRWVlZY8aMOXfu3OHDh/Py8pycnIwdFECvwFc8AOTo0aPx8fFSqXTnzp3Tp083djgA6t25c4fP5585c2bJkiXr16/HrO1g3nAHBYBMnTr1xo0bU6ZMCQ8P5/P5bW1txo4I4H+IxeKNGzf6+vo2NTUVFxf/+9//RnUCZg93UAD+a+/evYmJiYMGDcrKygoKCjJ2OACEEHL27NlFixY9evTo888///jjj9lstrEjAjAE3EEB+K85c+bcvHnz5ZdfnjRp0qefftrR0WHsiKBPa2pq+vDDD0NDQ728vMrKyj755BNUJ9B34A4KgDKKov7zn/8sX7589OjRWVlZPj4+xo4I+qKCgoLFixdLpdKNGzfOnz/f2OEAGBruoAAoY7FYixYtun79upWV1auvvrpx40aZTGbsoKAPefr06axZs955550///nPN27cQHUCfRMKFAD1vLy8Tp06tWrVqpUrV/7lL38RiUTGjgjMn0wmy8zM9PHxKS0tPXbsWFZWlqOjo7GDAjAOFCgAnaLnczt37pxQKHzllVcyMzONHRGYs+vXrwcGBiYlJS1ZsuTGjRt//vOfjR0RgDGhQAHowuuvv3716tXExMTFixdHRUXV19cbOyIwN3/88cfq1atff/11S0vLkpKSr776isvlGjsoACPDIFkAbR07diw+Pv7/tXenUVGcWR/An6ZpGhBwQWVGUQSj4gLijBIQFYmo4xEQBpDFLWLGBoXjSkYdF8ScYNQYnIAGNwYJTho0UUDQuCAaAYEMyKqgoAGVtUUWBXqp90Odt08fwAZRqGr9/z41j/XUvV3Hbi5VT92SSCQnTpxwcHBgOh34QKSmpq5du7a6unrv3r0BAQFqavi7EYAQnEEB6Dl7e/uCggInJycnJyeBQNDc3Mx0RqDaRCKRQCCws7MbP358QUHBhg0bUJ0AyOEMCsBbi4uL8/PzGzhwYFRU1KxZs5hOB1RSXFzc+vXreTzev//9b1dXV6bTAWAdVOsAb83d3b2goGDSpEl2dnbo5wZvq6ysbMGCBZ6eni4uLvfv30d1AtAlFCgAvfGnP/0pPj4+PDw8LCxsxowZeXl5TGcEKkAikRw5csTc3LyqqurOnTsRERG6urpMJwXAUihQAHpJ3s9NT0/v008/RT83UC4nJ8fKymr79u1bt27Nzs62srJiOiMAVkOBAvBOTExMUlJSgoKCdu/evWDBgoqKCqYzAtZ59erVtm3bZsyYMWDAgJycnKCgIA0NDaaTAmA7FCgA74ru55aVlVVbW2tmZoZ+bqDo0qVLEydOPH78+NGjR2/evDlhwgSmMwJQDShQAN4Pc3Pzu3fv0v3c3N3d0c8NqqqqVq5c6eDg8Omnnz548GDt2rUcDofppABUBm4zBnjPrl+/vnr16vb29hMnTjg6OjKdDjCAoqjo6OjNmzfr6ekdO3Zs4cKFTGcEoHpwBgXgPZs3b15+fv6SJUucnJxWrlyJfm4fm9LSUnt7+zVr1ixbtiwvLw/VCUDvoEABeP8GDhwYERERFxeXlJRkbm5++/ZtpjOC/iAWi7/55hszMzORSJSenn7kyBEdHR2mkwJQVShQAPqKm5tbYWHhlClT0M/tY/Dbb79ZWFgEBwfv3bs3Ozt7+vTpTGcEoNpQoAD0IQMDg/j4+NOnT4eHh0+fPv3evXtMZwTvX0NDw4YNG2xtbceMGVNUVPTPf/6Ty+UynRSAykOBAtDnVq5cmZeXN2jQICsrK/Rz+8AkJCSYmZkJhcLIyMhLly4ZGRkxnRHABwIFCkB/MDY2vnHjBt3Pbc6cOWVlZUxnBO/q2bNnrq6uS5YssbOzKywsXLlyJdMZAXxQUKAA9BO6n1t2dnZzc/O0adPQz011yWSy48ePm5qa5uXlXb169cyZM/r6+kwnBfChQYEC0K/MzMwyMjL8/Pz8/Pzc3Nzq6uqYzgjeTl5e3syZM/39/detW5efnz9v3jymMwL4MKFAAehvmpqa+/fvv3XrVk5OzuTJk+Pj45nOCHrk9evXQUFBM2bM4HK5OTk5+/fv19TUZDopgA8WChQAZtjY2OTk5Dg7Oy9ZsgT93NgvNTV12rRpoaGhBw4cuH379uTJk5nOCOADhwIFgDF6enoRERHnzp1LTk42MzO7desW0xlBF168eCEQCOzs7MaNG1dQULBhwwY1NXxzAvQ5fMwAGObq6lpQUGBubm5nZ7dhwwb0c2OVuLi4CRMmJCQkxMXFJSQkGBoaMp0RwMcCBQoA8wwMDC5evBgZGXn69Om//vWvubm5TGcEpKysbOHChR4eHi4uLvfv33d1dWU6I4CPCwoUALag+7kNGTLE2tq6y35ubW1tvr6+EomEkfQ+PNevX+9yXCKRHDlyZOrUqc+fP09LS4uIiNDT0+vn3AAABQoAixgbG6ekpOzfv3/37t2zZ89+9OiR4r/u3LkzIiIiODiYqfQ+JKdOnVq4cGF2dnaH8ZycHGtr623btm3ZsiU7O9vKyoqR9ACAQ1EU0zkAQEcFBQUrVqwoKys7ePDg2rVrCSGpqal2dnYURampqV2/fn3u3LlM56jCsrKybGxsJBLJ5MmTc3Jy1NXVCSGvXr0KDg4+dOiQjY1NRESEqakp02kCfNRwBgWAjaZMmZKenk73c1u8eHFJScmyZcvom0c4HI6Hhwc6vPVafX29i4sLRVEURRUXF4eGhhJCLl26NGnSpOPHjx89evTmzZuoTgAYhzMoAKx28+bNzz//vLW1tb6+Xr76hMfjffbZZ8nJyRwOh9n0VI5EIvnss88yMjLEYjE9wufz7e3tL126tHz58sOHDw8bNozZDAGAhjMoAKw2d+7ckJCQ6upqxbWxYrH4119/DQ8PZzAxFfXll1+mpaXJqxNCiEwmKywsTE5Ojo6ORnUCwB44gwLAarW1taampg0NDZ1v6uHxeJmZmRYWFowkpor++9//ent7dx7ncDjR0dHLli3r/5QA4E1QoACwmrOzc1JSkuJf/HLq6urGxsa5ubna2tr9n5jKyc/Pt7S0bGtr6/ylx+Fw9PT0SktLcQYFgD1wiQeAvU6dOnXx4sUuqxNCiEQiKS8vDwgI6OesVNGLFy8cHR2lUmmXf5JRFPXq1astW7b0f2IA8CYoUADYS0dHZ+nSpQMHDiSE8Pn8zo+AkUgkp0+fjouLYyI7lSGTyby8vJ49e/amUk9DQ0Mikfz4448ZGRn9nBsAvAku8QCwnUwmy8nJuXbtWnJy8p07dyQSCY/Hk/+u5XA4AwYMyM/PHzNmDKNpsteuXbu+/vprxUU8HA5HQ0Ojra1NTU1t7NixdnZ2NjY2tra2RkZGDOYJAIpQoACoksbGxpSUlCtXriQlJT158kRdXZ2iKKlUOmPGjLS0NLrhGCiKj493dnamKIrL5RJCpFKptrb2zJkzbW1tZ82aZWlpiRU8AOyEAgVYwd3dnekUVE9zc3N1dXVVVVVNTY1UKp04ceLkyZOZTopdmpqabty4IRaLNTU1hw8frq+vP3ToUD09PfSP6dbmzZutra2ZzgI+aihQgBU4HI6VlRWeZd87FEXV19fX1NQYGxtraWn1T1B6uQbLH1VTVlbG4/GGDh0qPyyVlZUZGRlubm7MJsZy586dEwqFS5cuZToR+KjhhDCwxaZNm/CFqELok14qtz43NjbWw8ND5dLuZzjDBGyAu3gAAACAdVCgAAAAAOugQAEAAADWQYECAAAArIMCBQAAAFgHBQoAAACwDgoUAAAAYB0UKAAAAMA6KFAAoD/MnTt3//79TGfRU7NmzfL29o6NjX358iU9IpFIwsPDW1paEhISUlJS9u7dO3v27EePHinZyS+//BIWFnbu3Ll169alpaURQnJzc1NSUrqNLo9FCOlhuG5j5eTkREZGjhkz5ujRoz07BgBMowBYgBAiFAqZzgLegpubm5ubW8+3F4lE7e3t7yV0Xl7ew4cPezdXKBT25HvPxsYmMTFR/qNEIhEIBFVVVRKJxM7OTiqVUhTl6uq6ZcuWN+2hvb190qRJYrGYoqjMzMz58+fT4/Hx8bGxsUpCy2PRr3sSruexPDw8wsLCun37+DwCG+AMCgD0h8GDB/N4vHffT0NDg6enJ312od/ExMSMGzfOwMCAy+Xq6urW1tYSQj755JPi4uI3TWlpaSktLS0sLCSEiEQifX19etzBweHYsWNK8pfHIoT0MFzPY3E4HLSxB1WBAgUA+lx+fv7q1at37dpFCDl//ryNjU10dPTSpUuHDBny008/EULOnj07c+bMsLAwW1tbbW3tPXv2EEKio6MNDAxKSkoqKys9PT1tbW0JIVFRUUVFRaGhoQkJCYQQJyenkJCQPk2eoqjg4GBXV1f6x4sXL9LVw++//+7t7f2mWYMGDVq8eLGbm1tubu7Jkyf/9a9/0eMcDsfc3PzYsWM9idXDcL2LBcByKFAAoM+ZmZm9fPmS/lPe0dHxwYMH//vf/86ePbtr166vv/6aEOLi4pKVlVVbW3v9+vXDhw/v27evurra29u7pqZGIpEYGhouXry4vr6eEOLv708ICQwMdHR0JIQ4Ozvb2Nj0afJPnz598uSJkZGR4mBNTY1YLFZSoBBCoqKiBg4cOG3atOXLl0+ZMkU+PmrUqKtXr/Y8Vk/C9SIWAMuhQAGA/jBkyBD6hYaGhq6uro2Njbq6uqWlZUVFBSFES0tLS0tr/vz56urqvr6+w4cPv3LlCpfLlU9XV+/60es+Pj5z5szp08xLSkqGDRvW4cqIUCgMDQ1VfrmkrKzM3NzcyclpxYoVd+7ckY+PGjWqpKSk57F6Eq4XsQBYruvPPABAP+ByuRRFdR63tra+f/++kon9uZCipaVFKpV2GFRTU7OwsFA+0dvb+7fffhs8eLBAIBAIBAUFBfT4sGHDRCJRz2P1JFwvYgGwHM6gAADrPHz4cPz48Uo26M8CxdTUtK6uTiKRKA7Onz9f+aySkpLXr18PGTKEw+F89dVXxcXF1dXV9D+JRKLRo0f3PFa34XoXC4DlUKAAQH9oa2trb2+nX9P3zRJC2tvbZTKZfBt6kUp9fX1VVZWbmxshZPTo0UVFRYSQ4uLipqYmQgiXy+Xz+SKRqLW1lRCSlJSUm5vbp5mbmJhoamrKf+XTGdJZyVVUVBw/flwsFstHjI2NpVLps2fPCCGamprTpk2j17oSQqqqqujTIZ1ndY7VOVzvYgGoHBQoANDn7t69m56enpqaWlhYeOPGjcrKyoSEhLq6uujo6KamJvp+HEJIVFRUeHi4QCA4f/68jo4OIcTf33/FihVOTk76+vo6OjqXL18mhPj4+Hh5edGzDhw4cObMmT5Nnsvlbty4MSsrSz5y7dq11atXK5ZWiYmJAoGATo/G4/G+++67r776SigU/vDDD/RaYNrVq1cFAkGXszrH6hyud7EAVA+zbVgAaASNoVTN2zZq65auru7t27erq6tlMpnieFNTE0VRdBcyuebmZvpFa2trh+2V612jtsbGxqVLlypu0NLS0mFKYmLivXv3Ou/qxYsXij9WVlauWLFCyazOsTqH610siqI8PT3Dw8M7T+wAn0dgA5xBAQBWoChKJpMNHz68w/oS+lRKh7t4BgwYQL/g8/l9tB7lzp07paWl9GUpXV3dnTt3Kp6q0dbWVty4rq6usbHR3Ny8834GDRokf11fX3/w4MGTJ08qmdU5VodwvYslEokePHhQVlbWzdsGYA3cxQMAzBMKhc3NzTExMWPHjh05ciTT6ZCTJ082NTU1NDQYGhrSI2ZmZvr6+i9evBg8eHDn7YcOHerl5dXtbp89e3bo0CF5sfWmWX0R69WrV42NjWFhYZ2brACwEwoUAGCeg4MDvQaWz+cznQshhJiamnYeHDFixDvu1szMrIdbvvdYhoaG8mILQCWgQAEA5skv2QAA0LAGBQAAAFgHBQqomObm5oCAAHmbB5WTlpY2evTowMDAfoiVnJw8duxYLpcbEBAQGBi4devW4OBgND4HAJWAAgVUjI6Ojpub25uezELLz89/9OhRHyXwjjufOXMm/VTefrBo0aJFixYZGhp+//33Bw8ePHDggL6+vo2NTXZ2tpJZfXr0AAB6CAUKqBiZTKampqbkztKGhgZPT0+6J+l79152zuVy+61Tu5aWljyWmpra+vXr3d3d582b19jY2OX2fXr0AAB6DgUKqIysrKyNGzeGhoaGhYXJf+lmZmYGBAQEBQU5OjrSLc+joqKKiopCQ0Pl/Uk7uHDhgr6+/qZNmwghIpHIzs6O3vLy5csCgWDOnDnff/89vWVOTo6Pj88333yzZMmShoaGzjvvHP3nn3+eN29eTEzM5MmTe3Id59tvvxUKhX5+frt27SKEnDlzZty4ccHBwRKJpLGx0dPTk24Y2jm3DoGcnJxCQkJ6chh9fX0bGxvv3r3bu6MHANBPmO4UB0BRPehc2dDQYGxs3NbWRlHU/v37DQ0NKYp6/vy5iYkJ3WlUKBTq6+vX19fTD1orKir0RKLHAAAKp0lEQVRSsjcXF5eAgAD69caNGymKevjwoa+vLx2Iz+cXFhbW1dVZWlrST41ZtGjRwYMHKYpS3HmX0auqqrhc7oYNG3Jzc2/fvt1l9FWrVgUGBlIUVVFRMXDgQIqiXr16xePx2tvbKYqysLAICQmht1y3bl2XuVEU1SHQqVOnUlNTO8faunWrkZGR4khrayuXy927d2+vjx6NflYOfKjQSRYYh9uMQTX85z//mTBhgoaGBiHE2to6PDycEBIZGTl+/Hi606iDg4O3t7dQKFy7dm23e1u7du3KlSu//fbbysrKiRMnEkKOHz/+8uVL+iEmVlZWGRkZtbW1lpaWampqhJDY2Fg6tKIuo/v5+Q0dOtTR0XHq1KndpjFy5MjU1FRCSHp6ulQqraioMDExCQgI2Ldv35dffllTU0M/h7ZzbpMmTTIwMFAMNGvWrB4eSfqRLjKZrNdHT87Kyoo+EaVC0tPTQ0ND6Yb38CYeHh5MpwCAPiigIgoLC+UNRjkcDn2Jp7y8nC4gCCHa2tpmZmaPHz+Wb6NkbwsWLODz+ZcvX37w4MGaNWsIIQ8fPnR0dPz8888JITt27CCErFq1St7Yiv4tLqc8upqamvI1vIr7aWxs3L59+xdffMHhcKRSKSHEy8srMDDw8uXLf/zxx9KlS7vMjdbzQIqKioqkUulf/vKXxMTE3h09OUNDQzpD1RIaGqqKafcnFCjABliDAqrB2Ng4IyOjw+DIkSMzMzPlP/J4PMUiRsne1NTUVq1aFRUV9fLlS7qb+NSpU9PS0uQbPHr0aNCgQUlJSfKR58+fy1/TO1cSvYcKCwvXrFmzb9++sWPHyge1tLTWrFlz9OjRiooKY2PjLnN7qygdxMTEGBsb29nZ9froAQD0AxQooBpcXFyKioru3btHCHn69Cn9cFdPT8+GhoaCggJCiFgsLi0tdXV15XK5fD5fJBK1trYq2eHq1asvXLhgaWlJ/7ho0aIzZ8788MMPYrH4zp07eXl5ixcvzs3NPXHiRGtr688//5yfn08IUdx5l9EJITKZTCwWKwktlUrpkyW3bt2il4Dcv39fKpXKZ61bt+7XX3+VNzvvnBs9rhgoKSmJXuXaAX2g6NevX78ODw8PDw//6aefdHV13+XoAQD0OYbXwABQFNWzx7tv3rx5yJAhf//739esWWNiYnLy5EmKoqKjo+fMmZOUlLR8+fLz58/TW/r5+RkZGcXGxirfobu7O70ulaIomUxGX2cxMjLasWOHTCajKGr37t08Ho/P5+/YsUM+S3HnnaP/+OOPhBB3d/fHjx93GTQjI8PIyGjatGlFRUXl5eUjRoyYPn16RESEtbX1ggUL6urq6M3+9re/PX/+XEluHQLZ2tpu2rSpQ6ykpKQxY8ZwOJx//OMfzs7Oc+bM8fHxUVwA+y5Hz83Nzc3NTfk2LESvPmE6C7bryecRoK9xqP//6wqAQRwORygUdrsyoKWlhcfjcTgcdXV1+WWItra28vJyExMTxXWsLS0t3T7eRSwW83g8xZGGhgZtbW3F/bS2tspkMsWH3XfYeZfRe04ikXA4HC6XK5VKFfu7rF+/nl4IrCQ3RW1tbRoaGr24NNPro+fu7k4IiYuLe9uIzIqNjfXw8MD3nnI9/DwC9CkskgVV0uVvTT6f3/nZs/SWzs7OHcY5HM4vv/xCv+5QnRBCBg0a1GFEU1NTeRpdRqcpj06Tr3LlcrmEkMrKypKSkqKiIhcXl25zU9TrhwArOXoAAAxCgQIfsgsXLqhW9Pj4+G3btgUGBtrb2/dFSgAAqgIFCgCLrFu3ztfXV373LwDARwvfgwDsguoEAICgQAEA6DsSiSQ8PLylpSUhISElJWXv3r2zZ8/utpNNYmLi1KlT6TvbCSG5ubkpKSl9nywAu6BAAQBWyM/P70UPut7N6h9SqdTf39/NzU1TU/O7776ztbXds2ePgYHBsWPHlMx6/vz5iBEj5N1uCCEWFhbNzc0qd8MUwDtCgQIAzGtoaPD09GxpaemHWf0mJiZm3LhxBgYGXC5XV1e3traWEPLJJ58UFxcrmfXnP//Z3Ny8w6CDg8OxY8dY+04B+gIKFADoQ5mZmQEBAUFBQY6OjnSv2+joaAMDg5KSksrKSk9PT1tbW0JIVFRUUVFRaGhoQkLC2bNnZ86cGRYWZmtrq62tvWfPnp7McnJyCgkJYfbNKqIoKjg4mG4uTAi5ePGigYEBIeT333/39vZ+271xOBxzc3Plp14APjAoUACgr1RVVXl5eYWEhAQFBa1YscLe3l4kEnl7e9fU1EgkEkNDw8WLF9fX1xNC/P39CSGBgYGOjo4uLi5ZWVm1tbXXr18/fPjwvn37qquru53l7OxsY2PD7PtV9PTp0ydPnhgZGSkO1tTUiMXiXhQohJBRo0ZdvXr1PWUHoAJQoABAX4mMjBw/fjz9LGgHB4eGhgahUEi3pKN1+TRmLS0tLS2t+fPnq6ur+/r6Dh8+/MqVK93O8vHxmTNnTh+8iV4qKSkZNmxYh96+QqEwNDS0d89iHDVqVElJyXvKDkAFoEABgL5SXl4uv2taW1vbzMzs8ePHSrbv8je3tbX1/fv333YW41paWuhHQipSU1OzsLDo3Q6HDRsmEoneOS8AlYECBQD6ysiRIzMzM+U/8ni8kSNHKtm+y1Lj4cOH48ePf9tZjDM1Na2rq5NIJIqD8+fP7/UORSLR6NGj3zkvAJWBAgUA+oqnp2dDQ0NBQQEhRCwWl5aW0otGR48eXVRURAgpLi5uamoihHC5XD6fLxKJWltb6bn0HSv19fVVVVVubm7dzkpKSqIX4bKEiYmJpqZmdXW1fKS+vp7OX66iouL48eNisbjD3A5lDa2qqqrXZ18AVBEKFADoKxMmTIiMjFy/fn1ycrKPj8+pU6foMyj+/v4rVqxwcnLS19fX0dG5fPkyIcTHx8fLyyshIYGeGxUVFR4eLhAIzp8/T69iUT7rwIEDZ86cYe69dsTlcjdu3JiVlSUfuXbt2urVq2UymXwkMTFRIBDQb0SupqbmwIEDhJDIyMg//vhDPn716lWBQND3iQOwBQePHQc2wOPdVY67uzshpCfdw9ra2srLy01MTDQ0NOSDzc3NOjo6EolEccVrS0sL/SBlPT29pKSk8ePHd1hnqmRWW1ubhoZGt5d7YmNjPTw8+ud7r6mp6YsvvhAKhfKRV69eaWtrK25z6dKlUaNGdW580sHTp0+3b9/ebxUYPo/ABjiDAgB9i8/nm5qaKlYnhBD6pEiH+3Ho6oQQQlGUTCYbPnx4h4JDySw+n8+2xSi6uro7d+5UrCo6VCd1dXWNjY3dVif19fUHDx48efJkn2QJwFYoUACAXYRCYXNzc0xMzNOnT5nO5V2ZmZnZ29u/ePGiy38dOnSol5dXtzt59uzZoUOHOlR4AB+8LtoJAAAwyMHBgV4Dy+fzmc7lPRgxYsQ77sHMzOy9ZAKgWlCgAAC7yC/0AMDHDJd4AAAAgHVQoAAAAADroEABAAAA1kGBAgAAAKyDRbLAFunp6UynAG+hsrKSEBIbG8t0Im+H/m+mcmkDfITQSRZYgW0ttgA+cugkC4xDgQIAAACsgzUoAAAAwDooUAAAAIB1UKAAAAAA66BAAQAAANb5P3K4uwfx1624AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "plot_model(model, to_file='./dot_model.png', show_shapes=True, show_layer_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss = 'mse', optimizer='adam', metrics=['mse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/8\n",
      "24544/24544 [==============================] - 43s 2ms/step - loss: 13.0896 - mse: 13.0896\n",
      "Epoch 2/8\n",
      "24544/24544 [==============================] - 43s 2ms/step - loss: 3.4429 - mse: 3.4429\n",
      "Epoch 3/8\n",
      "24544/24544 [==============================] - 41s 2ms/step - loss: 1.4427 - mse: 1.4427\n",
      "Epoch 4/8\n",
      "24544/24544 [==============================] - 42s 2ms/step - loss: 1.0189 - mse: 1.0189\n",
      "Epoch 5/8\n",
      "24544/24544 [==============================] - 39s 2ms/step - loss: 0.8485 - mse: 0.8485\n",
      "Epoch 6/8\n",
      "24544/24544 [==============================] - 38s 2ms/step - loss: 0.7596 - mse: 0.7596\n",
      "Epoch 7/8\n",
      "24544/24544 [==============================] - 37s 2ms/step - loss: 0.7084 - mse: 0.7084\n",
      "Epoch 8/8\n",
      "24544/24544 [==============================] - 36s 1ms/step - loss: 0.6741 - mse: 0.6741\n"
     ]
    }
   ],
   "source": [
    "history = model.fit([train.user_id, train.book_id], train.rating, epochs=8, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'training error')"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(history.history['loss'])\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('training error')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6136/6136 [==============================] - 2s 303us/step - loss: 0.9524 - mse: 0.9524\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.9524081945419312, 0.9524081945419312]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate([test.user_id, test.book_id], test.rating)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3.0390196] 4\n",
      "[3.4663684] 5\n",
      "[4.4353533] 4\n",
      "[5.022065] 5\n",
      "[1.3333156] 2\n",
      "[3.702732] 4\n",
      "[3.8910027] 5\n",
      "[4.9327035] 4\n",
      "[4.0983667] 4\n",
      "[3.0378323] 3\n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict([test.user_id.head(10), test.book_id.head(10)])\n",
    "\n",
    "for p, t in zip(predictions, test.rating.values[:10]):\n",
    "    print(p, t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Dense layer (Neural Network로 만들기)\n",
    "\n",
    "위에서는 단순하게 Dot을 이용해서 rating을 예측하는 형태로 만들었다.   \n",
    "이번에는 Dense layer를 이용해서 layer를 쌓고 rating을 예측하는 방법을 적용해본다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# layer 쌓기\n",
    "\n",
    "book_input = Input(shape=(1, ), name='book_input_layer')\n",
    "user_input = Input(shape=(1, ), name='user_input_layer')\n",
    "\n",
    "book_embedding_layer = Embedding(number_of_unique_book_id + 1, 8, name='book_embedding_layer')\n",
    "user_embedding_layer = Embedding(number_of_unique_user + 1, 8, name='user_embedding_layer')\n",
    "\n",
    "book_vector_layer = Flatten(name='book_vector_layer')\n",
    "user_vector_layer = Flatten(name='user_vector_layer')\n",
    "\n",
    "concate_layer = Concatenate()\n",
    "\n",
    "dense_layer1 = Dense(128, activation='relu')\n",
    "dense_layer2 = Dense(32, activation='relu')\n",
    "\n",
    "result_layer = Dense(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 쌓기\n",
    "book_embedding = book_embedding_layer(book_input)\n",
    "user_embedding = user_embedding_layer(user_input)\n",
    "\n",
    "book_vector = book_vector_layer(book_embedding)\n",
    "user_vector = user_vector_layer(user_embedding)\n",
    "\n",
    "concat = concate_layer([book_vector, user_vector])\n",
    "dense1 = dense_layer1(concat)\n",
    "dense2 = dense_layer2(dense1)\n",
    "\n",
    "result = result_layer(dense2)\n",
    "\n",
    "model = Model(inputs=[user_input, book_input], outputs=result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "book_input_layer (InputLayer)   [(None, 1)]          0                                            \n",
      "__________________________________________________________________________________________________\n",
      "user_input_layer (InputLayer)   [(None, 1)]          0                                            \n",
      "__________________________________________________________________________________________________\n",
      "book_embedding_layer (Embedding (None, 1, 8)         80008       book_input_layer[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "user_embedding_layer (Embedding (None, 1, 8)         427400      user_input_layer[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "book_vector_layer (Flatten)     (None, 8)            0           book_embedding_layer[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "user_vector_layer (Flatten)     (None, 8)            0           user_embedding_layer[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 16)           0           book_vector_layer[0][0]          \n",
      "                                                                 user_vector_layer[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 128)          2176        concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dense_4 (Dense)                 (None, 32)           4128        dense_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_5 (Dense)                 (None, 1)            33          dense_4[0][0]                    \n",
      "==================================================================================================\n",
      "Total params: 513,745\n",
      "Trainable params: 513,745\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "plot_model(model, to_file='./dense_predict_model.png', show_shapes=True, show_layer_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss = 'mse', optimizer='adam', metrics=['mse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/8\n",
      "24544/24544 [==============================] - 43s 2ms/step - loss: 0.7982 - mse: 0.7982\n",
      "Epoch 2/8\n",
      "24544/24544 [==============================] - 37s 2ms/step - loss: 0.6871 - mse: 0.6871\n",
      "Epoch 3/8\n",
      "24544/24544 [==============================] - 37s 2ms/step - loss: 0.6526 - mse: 0.6526\n",
      "Epoch 4/8\n",
      "24544/24544 [==============================] - 38s 2ms/step - loss: 0.6243 - mse: 0.6243\n",
      "Epoch 5/8\n",
      "24544/24544 [==============================] - 38s 2ms/step - loss: 0.5966 - mse: 0.5966\n",
      "Epoch 6/8\n",
      "24544/24544 [==============================] - 44s 2ms/step - loss: 0.5686 - mse: 0.5686\n",
      "Epoch 7/8\n",
      "24544/24544 [==============================] - 40s 2ms/step - loss: 0.5433 - mse: 0.5433\n",
      "Epoch 8/8\n",
      "24544/24544 [==============================] - 42s 2ms/step - loss: 0.5200 - mse: 0.5200\n"
     ]
    }
   ],
   "source": [
    "history = model.fit([train.user_id, train.book_id], train.rating, epochs=8, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'training error')"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(history.history['loss'])\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('training error')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6136/6136 [==============================] - 2s 349us/step - loss: 0.7414 - mse: 0.7414\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.7414111495018005, 0.7414111495018005]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate([test.user_id, test.book_id], test.rating)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.7931433] 4\n",
      "[4.2889414] 5\n",
      "[4.764618] 4\n",
      "[4.2071877] 5\n",
      "[3.5028908] 2\n",
      "[3.7568953] 4\n",
      "[4.081534] 5\n",
      "[4.465719] 4\n",
      "[3.6455593] 4\n",
      "[3.799435] 3\n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict([test.user_id.head(10), test.book_id.head(10)])\n",
    "\n",
    "for p, t in zip(predictions, test.rating.values[:10]):\n",
    "    print(p, t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "tmp_book_data = np.array(list(set(data.book_id)))[:1000]\n",
    "tmp_user = np.array([1 for i in range(1000)])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 1  2  3  4  5  6  7  8  9 10]\n",
      "[1 1 1 1 1 1 1 1 1 1]\n"
     ]
    }
   ],
   "source": [
    "print(tmp_book_data[:10])\n",
    "print(tmp_user[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[4.232527  4.4912014 2.74577   4.705388  3.6804   ]\n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict([tmp_user, tmp_book_data])\n",
    "predictions = np.array([p[0] for p in predictions])\n",
    "print(predictions[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[862 610 780 950   4]\n"
     ]
    }
   ],
   "source": [
    "# id는 1부터 시작인데 argsort를 하면 0부터 되므로 1을 더한다.\n",
    "recommended_book_ids = (-predictions).argsort()[:5] + 1\n",
    "print(recommended_book_ids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "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>id</th>\n",
       "      <th>book_id</th>\n",
       "      <th>best_book_id</th>\n",
       "      <th>work_id</th>\n",
       "      <th>books_count</th>\n",
       "      <th>isbn</th>\n",
       "      <th>isbn13</th>\n",
       "      <th>authors</th>\n",
       "      <th>original_publication_year</th>\n",
       "      <th>original_title</th>\n",
       "      <th>...</th>\n",
       "      <th>ratings_count</th>\n",
       "      <th>work_ratings_count</th>\n",
       "      <th>work_text_reviews_count</th>\n",
       "      <th>ratings_1</th>\n",
       "      <th>ratings_2</th>\n",
       "      <th>ratings_3</th>\n",
       "      <th>ratings_4</th>\n",
       "      <th>ratings_5</th>\n",
       "      <th>image_url</th>\n",
       "      <th>small_image_url</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>2657</td>\n",
       "      <td>2657</td>\n",
       "      <td>3275794</td>\n",
       "      <td>487</td>\n",
       "      <td>61120081</td>\n",
       "      <td>9.780061e+12</td>\n",
       "      <td>Harper Lee</td>\n",
       "      <td>1960.0</td>\n",
       "      <td>To Kill a Mockingbird</td>\n",
       "      <td>...</td>\n",
       "      <td>3198671</td>\n",
       "      <td>3340896</td>\n",
       "      <td>72586</td>\n",
       "      <td>60427</td>\n",
       "      <td>117415</td>\n",
       "      <td>446835</td>\n",
       "      <td>1001952</td>\n",
       "      <td>1714267</td>\n",
       "      <td>https://images.gr-assets.com/books/1361975680m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1361975680s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>609</th>\n",
       "      <td>610</td>\n",
       "      <td>16158542</td>\n",
       "      <td>16158542</td>\n",
       "      <td>21998925</td>\n",
       "      <td>33</td>\n",
       "      <td>067002581X</td>\n",
       "      <td>9.780670e+12</td>\n",
       "      <td>Daniel James Brown</td>\n",
       "      <td>2013.0</td>\n",
       "      <td>The Boys in the Boat: Nine Americans and Their...</td>\n",
       "      <td>...</td>\n",
       "      <td>113565</td>\n",
       "      <td>139128</td>\n",
       "      <td>16287</td>\n",
       "      <td>2292</td>\n",
       "      <td>3772</td>\n",
       "      <td>15002</td>\n",
       "      <td>43457</td>\n",
       "      <td>74605</td>\n",
       "      <td>https://images.gr-assets.com/books/1354683116m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1354683116s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>779</th>\n",
       "      <td>780</td>\n",
       "      <td>77727</td>\n",
       "      <td>77727</td>\n",
       "      <td>2110010</td>\n",
       "      <td>24</td>\n",
       "      <td>836220889</td>\n",
       "      <td>9.780836e+12</td>\n",
       "      <td>Bill Watterson, G.B. Trudeau</td>\n",
       "      <td>1987.0</td>\n",
       "      <td>Calvin and Hobbes</td>\n",
       "      <td>...</td>\n",
       "      <td>117788</td>\n",
       "      <td>118294</td>\n",
       "      <td>684</td>\n",
       "      <td>980</td>\n",
       "      <td>1435</td>\n",
       "      <td>7684</td>\n",
       "      <td>22365</td>\n",
       "      <td>85830</td>\n",
       "      <td>https://s.gr-assets.com/assets/nophoto/book/11...</td>\n",
       "      <td>https://s.gr-assets.com/assets/nophoto/book/50...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>861</th>\n",
       "      <td>862</td>\n",
       "      <td>17332218</td>\n",
       "      <td>17332218</td>\n",
       "      <td>16482835</td>\n",
       "      <td>34</td>\n",
       "      <td>765326361</td>\n",
       "      <td>9.780765e+12</td>\n",
       "      <td>Brandon Sanderson</td>\n",
       "      <td>2014.0</td>\n",
       "      <td>Words of Radiance</td>\n",
       "      <td>...</td>\n",
       "      <td>73572</td>\n",
       "      <td>108176</td>\n",
       "      <td>7261</td>\n",
       "      <td>154</td>\n",
       "      <td>404</td>\n",
       "      <td>2653</td>\n",
       "      <td>18272</td>\n",
       "      <td>86693</td>\n",
       "      <td>https://images.gr-assets.com/books/1391535251m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1391535251s...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>949</th>\n",
       "      <td>950</td>\n",
       "      <td>20820994</td>\n",
       "      <td>20820994</td>\n",
       "      <td>11409817</td>\n",
       "      <td>71</td>\n",
       "      <td>803734964</td>\n",
       "      <td>9.780804e+12</td>\n",
       "      <td>Jandy Nelson</td>\n",
       "      <td>2014.0</td>\n",
       "      <td>I'll Give You the Sun</td>\n",
       "      <td>...</td>\n",
       "      <td>140937</td>\n",
       "      <td>153584</td>\n",
       "      <td>16701</td>\n",
       "      <td>7025</td>\n",
       "      <td>7517</td>\n",
       "      <td>20793</td>\n",
       "      <td>40241</td>\n",
       "      <td>78008</td>\n",
       "      <td>https://images.gr-assets.com/books/1496659336m...</td>\n",
       "      <td>https://images.gr-assets.com/books/1496659336s...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      id   book_id  best_book_id   work_id  books_count        isbn  \\\n",
       "3      4      2657          2657   3275794          487    61120081   \n",
       "609  610  16158542      16158542  21998925           33  067002581X   \n",
       "779  780     77727         77727   2110010           24   836220889   \n",
       "861  862  17332218      17332218  16482835           34   765326361   \n",
       "949  950  20820994      20820994  11409817           71   803734964   \n",
       "\n",
       "           isbn13                       authors  original_publication_year  \\\n",
       "3    9.780061e+12                    Harper Lee                     1960.0   \n",
       "609  9.780670e+12            Daniel James Brown                     2013.0   \n",
       "779  9.780836e+12  Bill Watterson, G.B. Trudeau                     1987.0   \n",
       "861  9.780765e+12             Brandon Sanderson                     2014.0   \n",
       "949  9.780804e+12                  Jandy Nelson                     2014.0   \n",
       "\n",
       "                                        original_title  ... ratings_count  \\\n",
       "3                                To Kill a Mockingbird  ...       3198671   \n",
       "609  The Boys in the Boat: Nine Americans and Their...  ...        113565   \n",
       "779                                  Calvin and Hobbes  ...        117788   \n",
       "861                                  Words of Radiance  ...         73572   \n",
       "949                              I'll Give You the Sun  ...        140937   \n",
       "\n",
       "    work_ratings_count  work_text_reviews_count  ratings_1  ratings_2  \\\n",
       "3              3340896                    72586      60427     117415   \n",
       "609             139128                    16287       2292       3772   \n",
       "779             118294                      684        980       1435   \n",
       "861             108176                     7261        154        404   \n",
       "949             153584                    16701       7025       7517   \n",
       "\n",
       "     ratings_3  ratings_4  ratings_5  \\\n",
       "3       446835    1001952    1714267   \n",
       "609      15002      43457      74605   \n",
       "779       7684      22365      85830   \n",
       "861       2653      18272      86693   \n",
       "949      20793      40241      78008   \n",
       "\n",
       "                                             image_url  \\\n",
       "3    https://images.gr-assets.com/books/1361975680m...   \n",
       "609  https://images.gr-assets.com/books/1354683116m...   \n",
       "779  https://s.gr-assets.com/assets/nophoto/book/11...   \n",
       "861  https://images.gr-assets.com/books/1391535251m...   \n",
       "949  https://images.gr-assets.com/books/1496659336m...   \n",
       "\n",
       "                                       small_image_url  \n",
       "3    https://images.gr-assets.com/books/1361975680s...  \n",
       "609  https://images.gr-assets.com/books/1354683116s...  \n",
       "779  https://s.gr-assets.com/assets/nophoto/book/50...  \n",
       "861  https://images.gr-assets.com/books/1391535251s...  \n",
       "949  https://images.gr-assets.com/books/1496659336s...  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "books[books['id'].isin(recommended_book_ids)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.5 64-bit ('soojin': conda)",
   "language": "python",
   "name": "python38564bitsoojinconda500ccdc350d846ea9d3ed9887713270b"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}