9.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-1. インボート + データ絞り込み"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20000, 19)"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# 本文とは異なり、オリジナルデータから20000件のデータを取り出し\n",
    "model_df = pd.read_json('data/mag_papers/mag_papers_0.txt', lines=True)\n",
    "model_df = model_df.head(20000)\n",
    "\n",
    "model_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10399, 6)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 英語でない文献の除去および変数の絞り込み\n",
    "model_df = model_df[model_df.lang == 'en']\\\n",
    "    .drop_duplicates(subset = 'title', keep = 'first')\\\n",
    "    .drop(['doc_type', 'doi', 'id', 'issue', 'lang', 'n_citation', 'page_end', 'page_start', 'publisher', \n",
    "              'references', 'url', 'venue', 'volume'], axis=1)\n",
    "\n",
    "model_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-2. 協調フィルタリングステージ１: アイテム, 特徴配列の構築"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of first feature array:  3787893471\n"
     ]
    }
   ],
   "source": [
    "# 本の記述と異なりますが、こちらの方が高速に動作するコードとなっています。\n",
    "\n",
    "def feature_array(x):\n",
    "    df_list = [pd.DataFrame([[1] * len(val)], columns=val, index=[index])\n",
    "                  if isinstance(val, list) else pd.DataFrame(index=[index])\n",
    "                  for val, index in zip(x.values, x.index)]\n",
    "    \n",
    "    feature_df = pd.concat(df_list, axis=1, sort=True)\n",
    "    return feature_df.fillna(0)\n",
    "\n",
    "year_features = pd.get_dummies(model_df['year'].astype('category'))\n",
    "fos_features = feature_array(model_df['fos'])\n",
    "first_features = fos_features.join(year_features).T\n",
    "\n",
    "from sys import getsizeof\n",
    "print('Size of first feature array: ', getsizeof(first_features))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-3. 協調フィルタリングステージ2: 類似アイテムの探索"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>5</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "      <th>...</th>\n",
       "      <th>986</th>\n",
       "      <th>988</th>\n",
       "      <th>990</th>\n",
       "      <th>992</th>\n",
       "      <th>993</th>\n",
       "      <th>994</th>\n",
       "      <th>995</th>\n",
       "      <th>996</th>\n",
       "      <th>998</th>\n",
       "      <th>1000</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.447214</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.447214</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 515 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       0     1         2     5     7     8     9     10    11    12    ...  \\\n",
       "0  1.000000   0.0  0.447214   0.0   0.0   0.0   0.0   0.0   0.0   0.0  ...   \n",
       "1  0.000000   1.0  0.000000   0.0   0.0   0.0   0.0   0.0   0.0   0.0  ...   \n",
       "2  0.447214   0.0  1.000000   0.0   0.0   0.0   0.0   0.0   0.0   0.0  ...   \n",
       "5  0.000000   0.0  0.000000   1.0   0.0   0.0   0.0   0.0   0.0   0.0  ...   \n",
       "7  0.000000   0.0  0.000000   0.0   1.0   0.0   0.0   0.0   0.0   0.0  ...   \n",
       "\n",
       "   986   988   990   992   993   994   995   996   998   1000  \n",
       "0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  \n",
       "1   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  \n",
       "2   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  \n",
       "5   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  \n",
       "7   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  \n",
       "\n",
       "[5 rows x 515 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from scipy.spatial.distance import cosine\n",
    "\n",
    "# 本の記述と異なりますが、こちらの方が簡潔なコードとなっています。\n",
    "def item_collab_filter(features_df):\n",
    "    return pd.DataFrame([[1 - cosine(col_val1, col_val2)\n",
    "                                     for col_val1 in features_df.T.values] for col_val2 in features_df.T.values],\n",
    "                                     index = features_df.columns, columns = features_df.columns)\n",
    "\n",
    "first_items = item_collab_filter(first_features.loc[:, 0:1000])\n",
    "\n",
    "first_items.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-4. ヒートマップによる類似論文の探索"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "\n",
    "sns.set()\n",
    "ax = sns.heatmap(first_items.fillna(0), \n",
    "                 vmin=0, vmax=1, \n",
    "                 cmap='YlGnBu', \n",
    "                 xticklabels=250, yticklabels=250)\n",
    "ax.tick_params(labelsize=12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-5. アイテムベース協調フィルタリングによる類似論文のレコメンド"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Based on the paper: \n",
      "index =  2\n",
      "abstract                                                  NaN\n",
      "authors     [{'name': 'Jovana P. Lekovich', 'org': 'Weill ...\n",
      "fos                                                       NaN\n",
      "keywords                                                  NaN\n",
      "title       Should endometriosis be an indication for intr...\n",
      "year                                                     2015\n",
      "Name: 2, dtype: object\n",
      "\n",
      "Top three results: \n",
      "1 . Paper index =  2\n",
      "Similarity score:  1.0\n",
      "abstract                                                  NaN\n",
      "authors     [{'name': 'Jovana P. Lekovich', 'org': 'Weill ...\n",
      "fos                                                       NaN\n",
      "keywords                                                  NaN\n",
      "title       Should endometriosis be an indication for intr...\n",
      "year                                                     2015\n",
      "Name: 2, dtype: object \n",
      "\n",
      "2 . Paper index =  292\n",
      "Similarity score:  1.0\n",
      "abstract                                                  NaN\n",
      "authors     [{'name': 'John C. Newton'}, {'name': 'Beers M...\n",
      "fos         [Wide area multilateration, Maneuvering speed,...\n",
      "keywords                                                  NaN\n",
      "title                    Automatic speed control for aircraft\n",
      "year                                                     1955\n",
      "Name: 561, dtype: object \n",
      "\n",
      "3 . Paper index =  593\n",
      "Similarity score:  1.0\n",
      "abstract    This paper demonstrates that on‐site greywater...\n",
      "authors     [{'name': 'Eran Friedler', 'org': 'Division of...\n",
      "fos         [Public opinion, Environmental Engineering, Wa...\n",
      "keywords    [economic analysis, tratamiento desperdicios, ...\n",
      "title       The water saving potential and the socio-econo...\n",
      "year                                                     2008\n",
      "Name: 1152, dtype: object \n",
      "\n"
     ]
    }
   ],
   "source": [
    "def paper_recommender(paper_ix, items_df):\n",
    "    print('Based on the paper: \\nindex = ', paper_ix)\n",
    "    print(model_df.iloc[paper_ix])\n",
    "    top_results = items_df.loc[paper_ix].sort_values(ascending=False).head(4)\n",
    "    print('\\nTop three results: ')\n",
    "    order = 1\n",
    "    for i in top_results.index.tolist()[-3:]:\n",
    "        print(order,'. Paper index = ', i)\n",
    "        print('Similarity score: ', top_results[i])\n",
    "        print(model_df.iloc[i], '\\n')\n",
    "        if order < 5: order += 1\n",
    "            \n",
    "paper_recommender(2, first_items)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-6. 固定幅のビン分割とダミーコーディング"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Year spread:  1831  -  2017\n",
      "Quantile spread:\n",
      " 0.25    1990.0\n",
      "0.50    2005.0\n",
      "0.75    2012.0\n",
      "Name: year, dtype: float64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Occurrence')"
      ]
     },
     "execution_count": 7,
     "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": [
    "year_min = model_df['year'].min()\n",
    "year_max = model_df['year'].max()\n",
    "\n",
    "print('Year spread: ', year_min, ' - ', year_max)\n",
    "print('Quantile spread:\\n', model_df['year'].quantile([0.25, 0.5, 0.75]))\n",
    "\n",
    "# 分布を確認するために出版年をプロットする\n",
    "fig, ax = plt.subplots()\n",
    "model_df['year'].hist(ax=ax, bins= year_max - year_min)\n",
    "ax.tick_params(labelsize=12)\n",
    "ax.set_xlabel('Year Count', fontsize=12)\n",
    "ax.set_ylabel('Occurrence', fontsize=12)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-7. 固定幅のビン分割とダミーコーディング: part2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "IntervalIndex([(1831.0, 1841.0], (1841.0, 1851.0], (1851.0, 1860.0], (1860.0, 1870.0], (1870.0, 1880.0] ... (1968.0, 1978.0], (1978.0, 1988.0], (1988.0, 1997.0], (1997.0, 2007.0], (2007.0, 2017.0]],\n",
       "              closed='right',\n",
       "              dtype='interval[float64]')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# ビンの幅を10年区切りに修正し、特徴空間を156から19に削減\n",
    "bins = int(round((year_max - year_min) / 10))\n",
    "\n",
    "temp_df = pd.DataFrame(index = model_df.index)\n",
    "temp_df['yearBinned'] = pd.cut(model_df['year'].tolist(), bins, precision = 0)\n",
    "\n",
    "X_yrs = pd.get_dummies(temp_df['yearBinned'])\n",
    "X_yrs.columns.categories"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Counts')"
      ]
     },
     "execution_count": 9,
     "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": [
    "# 新しい分布をプロットする\n",
    "fig, ax = plt.subplots()\n",
    "X_yrs.sum().plot.bar(ax = ax)\n",
    "ax.tick_params(labelsize=8)\n",
    "ax.set_xlabel('Binned Years', fontsize=12)\n",
    "ax.set_ylabel('Counts', fontsize=12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-8. 「bag-of-phrases」データをpd.SeriesからNumPy sparse arrayに変更"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Our pandas Series, in bytes:  3772086600\n",
      "Our hashed numpy array, in bytes:  112\n"
     ]
    }
   ],
   "source": [
    "X_fos = fos_features.values\n",
    "\n",
    "print('Our pandas Series, in bytes: ', getsizeof(fos_features))\n",
    "print('Our pandas Series, in bytes: ', X_fos.nbytes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-9. 協調フィルタリング stage1 + 2: 類似アイテム検索のためのアイテム特徴量行列作成"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The power of feature engineering saves us, in bytes:  14637007\n"
     ]
    }
   ],
   "source": [
    "second_features = np.append(X_fos, X_yrs, axis = 1)\n",
    "\n",
    "print('The power of feature engineering saves us, in bytes: ',\n",
    "         getsizeof(first_features) - getsizeof(second_features))\n",
    "\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "\n",
    "def piped_collab_filter(features_matrix, index, top_n):\n",
    "    item_similarities = \\\n",
    "        cosine_similarity(features_matrix[index:index+1],\n",
    "                              features_matrix).flatten()\n",
    "    related_indices = \\\n",
    "        [i for i in item_similarities.argsort()[::-1] if i != index]\n",
    "    return [(index, item_similarities[index]) for index in related_indices][0:top_n]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-10. アイテムベース協調フィルタリングによる類似論文のレコメンド Take2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Based on the paper:\n",
      "Paper index =  2\n",
      "Title : Should endometriosis be an indication for intracytoplasmic sperm injection (ICSI) in fresh IVF cycles\n",
      "FOS : nan\n",
      "Year : 2015\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Jovana P. Lekovich', 'org': 'Weill Cornell Medical College, New York, NY'}, {'name': 'G.D. Palermo', 'org': 'Weill Medical College of Cornell University, New York, NY'}, {'name': 'Nigel Pereira', 'org': 'The Ronald O. Perelman and Claudia Cohen Center, New York, NY'}, {'name': 'Zev Rosenwaks', 'org': 'Weill Cornell Medical College, New York, NY'}] \n",
      "\n",
      "\n",
      "Top 3 results: \n",
      "1 . Paper index =  12673\n",
      "Similarity score:  1.0\n",
      "Title : Correlations of Portal Pressure In Benign Biliary Stricture\n",
      "FOS : nan\n",
      "Year : 2014\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Hirdaya Nag', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Asit Arora', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Ila Tyagi', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'D. Ramaswamy', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Nilesh Patil', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Puja Sakhuja', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Renuka Saha', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}, {'name': 'Anil K. Agarwal', 'org': 'G B Pant Hospital and MAM College, New Delhi, India'}] \n",
      "\n",
      "2 . Paper index =  18108\n",
      "Similarity score:  1.0\n",
      "Title : PET imaging of angiogenesis in murine hindlimb ischemia with a 64Cu-labeled anti-CD105 antibody\n",
      "FOS : nan\n",
      "Year : 2013\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Hakan Orbay'}, {'name': 'Yin Zhang'}, {'name': 'Timothy A. Hacker'}, {'name': 'Charles P. Theuer'}, {'name': 'Todd E. Barnhart'}, {'name': 'Weibo Cai'}] \n",
      "\n",
      "3 . Paper index =  14439\n",
      "Similarity score:  1.0\n",
      "Title : Innovation Framework for Generating Biogas and Electricity from Biogas\n",
      "FOS : nan\n",
      "Year : 2011\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Elke Bruns', 'org': 'Technische Universität Berlin (Berlin Institute of Technology)'}, {'name': 'Dörte Ohlhorst', 'org': 'Technische Universität Berlin (Berlin Institute of Technology)'}, {'name': 'Bernd Wenzel', 'org': 'Ingenieurbüro für neue Energien (IfnE)'}, {'name': 'Johann Köppel', 'org': 'Technische Universität Berlin (Berlin Institute of Technology)'}] \n",
      "\n"
     ]
    }
   ],
   "source": [
    "def paper_recommender(items_df, paper_ix, top_n):\n",
    "    if paper_ix in model_df.index:\n",
    "        print('Based on the paper:')\n",
    "        print('Paper index = ', model_df.loc[paper_ix].name)\n",
    "        print('Title :', model_df.loc[paper_ix]['title'])\n",
    "        print('FOS :', model_df.loc[paper_ix]['fos'])\n",
    "        print('Year :', model_df.loc[paper_ix]['year'])\n",
    "        print('Abstract :', model_df.loc[paper_ix]['abstract'])\n",
    "        print('Authors :', model_df.loc[paper_ix]['authors'], '\\n')\n",
    "\n",
    "        array_ix = model_df.index.get_loc(paper_ix)\n",
    "        top_results = piped_collab_filter(items_df, array_ix, top_n)\n",
    "        print('\\nTop',top_n,'results: ')\n",
    "\n",
    "        order = 1\n",
    "        for i in range(len(top_results)):\n",
    "            print(order,'. Paper index = ', model_df.iloc[top_results[i][0]].name)\n",
    "            print('Similarity score: ', top_results[i][1])\n",
    "            print('Title :', model_df.iloc[top_results[i][0]]['title'])\n",
    "            print('FOS :', model_df.iloc[top_results[i][0]]['fos'])\n",
    "            print('Year :', model_df.iloc[top_results[i][0]]['year'])\n",
    "            print('Abstract :', model_df.iloc[top_results[i][0]]['abstract'])\n",
    "            print('Authors :', model_df.iloc[top_results[i][0]]['authors'], '\\n')\n",
    "            if order < top_n: order += 1\n",
    "                    \n",
    "    else:\n",
    "        print('Whoops! Choose another paper. Try something from here: \\n', model_df.index[100:200])\n",
    "\n",
    "paper_recommender(second_features, 2, 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-11. loc, iloc, get_locの使い分け(explain_loc_iloc_get_loc.ipynbにより詳しいサンプルコードを追加)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "abstract    A microprocessor includes hardware registers t...\n",
       "authors                      [{'name': 'Mark John Ebersole'}]\n",
       "fos         [Embedded system, Parallel computing, Computer...\n",
       "keywords                                                  NaN\n",
       "title       Microprocessor that enables ARM ISA program to...\n",
       "year                                                     2013\n",
       "Name: 21, dtype: object"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_df.loc[21]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "abstract                                                  NaN\n",
       "authors     [{'name': 'Nicola M. Heller'}, {'name': 'Steph...\n",
       "fos         [Biology, Medicine, Post-transcriptional regul...\n",
       "keywords    [glucocorticoids, post transcriptional regulat...\n",
       "title       Post-transcriptional regulation of eotaxin by ...\n",
       "year                                                     2002\n",
       "Name: 30, dtype: object"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_df.iloc[21]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "21"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_df.index.get_loc(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-12. ストップワード + TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "filled_df = model_df.fillna('None')\n",
    "\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "vectorizer = TfidfVectorizer(sublinear_tf=True, max_df=0.5, stop_words='english')\n",
    "X_abstract = vectorizer.fit_transform(filled_df['abstract'])\n",
    "\n",
    "third_features = np.append(second_features, X_abstract.toarray(), axis = 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-13. scikit-learnのDictVectorizerをつかったOne-Hotエンコーディング"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'None': 0},\n",
       " {'Ahmed M. Alluwaimi': 1},\n",
       " {'Jovana P. Lekovich': 2, 'Weill Cornell Medical College, New York, NY': 2},\n",
       " {'George C. Sponsler': 5},\n",
       " {'M. T. Richards': 7}]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "authors_list = []\n",
    "\n",
    "for row in filled_df.itertuples():\n",
    "    # それぞれのSeriesインデックスからdict型オブジェクトを作成する\n",
    "    if isinstance(row.authors, str):\n",
    "        y = {'None': row.Index}\n",
    "    if isinstance(row.authors, list):\n",
    "        # これらのキー、値をdict型オブジェクトに追加する\n",
    "        y = dict.fromkeys(row.authors[0].values(), row.Index)\n",
    "    authors_list.append(y)\n",
    "\n",
    "authors_list[0:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction import DictVectorizer\n",
    "v = DictVectorizer(sparse=False)\n",
    "D = authors_list\n",
    "X_authors = v.fit_transform(D)\n",
    "fourth_features = np.append(third_features, X_authors, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 例9-14. アイテムベース協調フィルタリングによる類似論文のレコメンド Take3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Based on the paper:\n",
      "Paper index =  2\n",
      "Title : Should endometriosis be an indication for intracytoplasmic sperm injection (ICSI) in fresh IVF cycles\n",
      "FOS : nan\n",
      "Year : 2015\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Jovana P. Lekovich', 'org': 'Weill Cornell Medical College, New York, NY'}, {'name': 'G.D. Palermo', 'org': 'Weill Medical College of Cornell University, New York, NY'}, {'name': 'Nigel Pereira', 'org': 'The Ronald O. Perelman and Claudia Cohen Center, New York, NY'}, {'name': 'Zev Rosenwaks', 'org': 'Weill Cornell Medical College, New York, NY'}] \n",
      "\n",
      "\n",
      "Top 3 results: \n",
      "1 . Paper index =  0\n",
      "Similarity score:  0.13608276348795434\n",
      "Title : System and Method for Maskless Direct Write Lithography\n",
      "FOS : ['Electronic engineering', 'Computer hardware', 'Engineering', 'Engineering drawing']\n",
      "Year : 2015\n",
      "Abstract : A system and method for maskless direct write lithography are disclosed. The method includes receiving a plurality of pixels that represent an integrated circuit (IC) layout; identifying a first subset of the pixels that are suitable for a first compression method; and identifying a second subset of the pixels that are suitable for a second compression method. The method further includes compressing the first and second subset using the first and second compression method respectively, resulting in compressed data. The method further includes delivering the compressed data to a maskless direct writer for manufacturing a substrate. In embodiments, the first compression method uses a run-length encoding and the second compression method uses a dictionary-based encoding. Due to the hybrid compression method, the compressed data can be decompressed with a data rate expansion ratio sufficient for high-volume IC manufacturing.\n",
      "Authors : nan \n",
      "\n",
      "2 . Paper index =  1\n",
      "Similarity score:  0.13608276348795434\n",
      "Title : The dilemma of the Mycobacterium avium subspecies paratuberculosis infection: In pursue for effective vaccine\n",
      "FOS : ['Biology', 'Virology', 'Immunology', 'Microbiology']\n",
      "Year : 2016\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Ahmed M. Alluwaimi'}] \n",
      "\n",
      "3 . Paper index =  9\n",
      "Similarity score:  0.03681050869161551\n",
      "Title : Lasing from Top-Down Gallium Nitride Nanowires.\n",
      "FOS : nan\n",
      "Year : 2012\n",
      "Abstract : nan\n",
      "Authors : [{'name': 'Jeremy B. Wright'}, {'name': 'George T. Wang'}, {'name': 'Weng Wah Chow'}, {'name': 'Qiming Li'}, {'name': 'Ting Shan Luk'}, {'name': 'Igal Brener'}, {'name': 'Luke F. Lester'}, {'name': 'Huiwen Xu'}] \n",
      "\n"
     ]
    }
   ],
   "source": [
    "paper_recommender(fourth_features, 2, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}