基于隐式偏好的推荐系统.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "传统的协同过滤推荐系统一般基于是用户对商品的评分(如5星好评)，这些评分通常可以明确表达用户对商品的喜好,通过分析用户评分，协同过滤推荐算法可以为用户推荐他们可能喜好的其他商品。但是要收集用户评分有时候会非常的困难,大多数情况是我们无法收集到用户的评分，那如果没有用户评分我们还能开发推荐系统吗？其实也是可以的，除了“显示”的用户评分，还有一种“隐式”的用户对物品喜好的表达方式，如用户的点击，点赞，浏览，收藏，评论等行为，隐式的表达方式虽然没有显示的表达方式那样直接表达用户的喜好，但它也可以间接反映出用户对物品的喜好程度。今天我们就来尝试利用用户隐式偏好数据,如点击，点赞，浏览，收藏，评论等行为数据，同时使用python和implicit库来开发一个个性化的推荐系统。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 数据\n",
    "\n",
    "我们的数据来自于kaggle,你可以在这里下载，这里包含了两个数据文件(shared_articles.csv，users_interactions.csv)，请大家将它们下载下来，这些数据来自于Deskdrop是 CI＆T开发的内部通信平台,该平台还允许公司员工与同行分享相关文章，并围绕他们进行协作。数据包中含大约73,000个用户在平台上共享的3k多篇公共文章上的交互信息，更重要的是，它包含丰富的隐式反馈信息，记录了不同的交互类型，我们可以利用这些信息来推断用户对文章的兴趣程度。\n",
    "\n",
    "同时我们将会使用implicit库，它是基于隐式偏好数据的python快速协同过滤推荐算法,它主要通过矩阵分解的方法来预测用户对物品的评分，从而进行个性化的推荐。\n",
    "\n",
    "我们首先加载数据集："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import scipy.sparse as sparse\n",
    "import numpy as np\n",
    "import random\n",
    "import implicit\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn import metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "在shared_articles表的contentId 列中总共有 3057 个唯一值.\n",
      "在users_interactions表的personId 列中总共有 1895 个唯一值.\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>timestamp</th>\n",
       "      <th>eventType</th>\n",
       "      <th>contentId</th>\n",
       "      <th>authorPersonId</th>\n",
       "      <th>authorSessionId</th>\n",
       "      <th>contentType</th>\n",
       "      <th>url</th>\n",
       "      <th>title</th>\n",
       "      <th>text</th>\n",
       "      <th>lang</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1459192779</td>\n",
       "      <td>CONTENT REMOVED</td>\n",
       "      <td>-6451309518266745024</td>\n",
       "      <td>4340306774493623681</td>\n",
       "      <td>8940341205206233829</td>\n",
       "      <td>HTML</td>\n",
       "      <td>http://www.nytimes.com/2016/03/28/business/dea...</td>\n",
       "      <td>Ethereum, a Virtual Currency, Enables Transact...</td>\n",
       "      <td>All of this work is still very early. The firs...</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1459193988</td>\n",
       "      <td>CONTENT SHARED</td>\n",
       "      <td>-4110354420726924665</td>\n",
       "      <td>4340306774493623681</td>\n",
       "      <td>8940341205206233829</td>\n",
       "      <td>HTML</td>\n",
       "      <td>http://www.nytimes.com/2016/03/28/business/dea...</td>\n",
       "      <td>Ethereum, a Virtual Currency, Enables Transact...</td>\n",
       "      <td>All of this work is still very early. The firs...</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1459194146</td>\n",
       "      <td>CONTENT SHARED</td>\n",
       "      <td>-7292285110016212249</td>\n",
       "      <td>4340306774493623681</td>\n",
       "      <td>8940341205206233829</td>\n",
       "      <td>HTML</td>\n",
       "      <td>http://cointelegraph.com/news/bitcoin-future-w...</td>\n",
       "      <td>Bitcoin Future: When GBPcoin of Branson Wins O...</td>\n",
       "      <td>The alarm clock wakes me at 8:00 with stream o...</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1459194474</td>\n",
       "      <td>CONTENT SHARED</td>\n",
       "      <td>-6151852268067518688</td>\n",
       "      <td>3891637997717104548</td>\n",
       "      <td>-1457532940883382585</td>\n",
       "      <td>HTML</td>\n",
       "      <td>https://cloudplatform.googleblog.com/2016/03/G...</td>\n",
       "      <td>Google Data Center 360° Tour</td>\n",
       "      <td>We're excited to share the Google Data Center ...</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1459194497</td>\n",
       "      <td>CONTENT SHARED</td>\n",
       "      <td>2448026894306402386</td>\n",
       "      <td>4340306774493623681</td>\n",
       "      <td>8940341205206233829</td>\n",
       "      <td>HTML</td>\n",
       "      <td>https://bitcoinmagazine.com/articles/ibm-wants...</td>\n",
       "      <td>IBM Wants to \"Evolve the Internet\" With Blockc...</td>\n",
       "      <td>The Aite Group projects the blockchain market ...</td>\n",
       "      <td>en</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    timestamp        eventType            contentId       authorPersonId  \\\n",
       "0  1459192779  CONTENT REMOVED -6451309518266745024  4340306774493623681   \n",
       "1  1459193988   CONTENT SHARED -4110354420726924665  4340306774493623681   \n",
       "2  1459194146   CONTENT SHARED -7292285110016212249  4340306774493623681   \n",
       "3  1459194474   CONTENT SHARED -6151852268067518688  3891637997717104548   \n",
       "4  1459194497   CONTENT SHARED  2448026894306402386  4340306774493623681   \n",
       "\n",
       "       authorSessionId contentType  \\\n",
       "0  8940341205206233829        HTML   \n",
       "1  8940341205206233829        HTML   \n",
       "2  8940341205206233829        HTML   \n",
       "3 -1457532940883382585        HTML   \n",
       "4  8940341205206233829        HTML   \n",
       "\n",
       "                                                 url  \\\n",
       "0  http://www.nytimes.com/2016/03/28/business/dea...   \n",
       "1  http://www.nytimes.com/2016/03/28/business/dea...   \n",
       "2  http://cointelegraph.com/news/bitcoin-future-w...   \n",
       "3  https://cloudplatform.googleblog.com/2016/03/G...   \n",
       "4  https://bitcoinmagazine.com/articles/ibm-wants...   \n",
       "\n",
       "                                               title  \\\n",
       "0  Ethereum, a Virtual Currency, Enables Transact...   \n",
       "1  Ethereum, a Virtual Currency, Enables Transact...   \n",
       "2  Bitcoin Future: When GBPcoin of Branson Wins O...   \n",
       "3                       Google Data Center 360° Tour   \n",
       "4  IBM Wants to \"Evolve the Internet\" With Blockc...   \n",
       "\n",
       "                                                text lang  \n",
       "0  All of this work is still very early. The firs...   en  \n",
       "1  All of this work is still very early. The firs...   en  \n",
       "2  The alarm clock wakes me at 8:00 with stream o...   en  \n",
       "3  We're excited to share the Google Data Center ...   en  \n",
       "4  The Aite Group projects the blockchain market ...   en  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "articles_df = pd.read_csv('./data/shared_articles.csv')\n",
    "interactions_df = pd.read_csv('./data/users_interactions.csv')\n",
    "articles_df.drop(['authorUserAgent', 'authorRegion', 'authorCountry'], axis=1, inplace=True)\n",
    "interactions_df.drop(['userAgent', 'userRegion', 'userCountry'], axis=1, inplace=True)\n",
    "articles_df = articles_df.drop_duplicates()\n",
    "interactions_df = interactions_df.drop_duplicates()\n",
    "print(\"在shared_articles表的contentId 列中总共有 %d 个唯一值.\" % articles_df['contentId'].nunique())\n",
    "print(\"在users_interactions表的personId 列中总共有 %d 个唯一值.\" % interactions_df['personId'].nunique())\n",
    "articles_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>timestamp</th>\n",
       "      <th>eventType</th>\n",
       "      <th>contentId</th>\n",
       "      <th>personId</th>\n",
       "      <th>sessionId</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1465413032</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-8845298781299428018</td>\n",
       "      <td>1264196770339959068</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1465412560</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>8890720798209849691</td>\n",
       "      <td>-1032019229384696495</td>\n",
       "      <td>3621737643587579081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1465416190</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>310515487419366995</td>\n",
       "      <td>-1130272294246983140</td>\n",
       "      <td>2631864456530402479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1465413895</td>\n",
       "      <td>FOLLOW</td>\n",
       "      <td>310515487419366995</td>\n",
       "      <td>344280948527967603</td>\n",
       "      <td>-3167637573980064150</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1465412290</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>-7820640624231356730</td>\n",
       "      <td>-445337111692715325</td>\n",
       "      <td>5611481178424124714</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    timestamp eventType            contentId             personId  \\\n",
       "0  1465413032      VIEW -3499919498720038879 -8845298781299428018   \n",
       "1  1465412560      VIEW  8890720798209849691 -1032019229384696495   \n",
       "2  1465416190      VIEW   310515487419366995 -1130272294246983140   \n",
       "3  1465413895    FOLLOW   310515487419366995   344280948527967603   \n",
       "4  1465412290      VIEW -7820640624231356730  -445337111692715325   \n",
       "\n",
       "             sessionId  \n",
       "0  1264196770339959068  \n",
       "1  3621737643587579081  \n",
       "2  2631864456530402479  \n",
       "3 -3167637573980064150  \n",
       "4  5611481178424124714  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interactions_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CONTENT SHARED     3047\n",
       "CONTENT REMOVED      75\n",
       "Name: eventType, dtype: int64"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "articles_df['eventType'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "articles_df = articles_df[articles_df['eventType'] == 'CONTENT SHARED']\n",
    "articles_df.drop('eventType', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 3047 entries, 1 to 3121\n",
      "Data columns (total 9 columns):\n",
      "timestamp          3047 non-null int64\n",
      "contentId          3047 non-null int64\n",
      "authorPersonId     3047 non-null int64\n",
      "authorSessionId    3047 non-null int64\n",
      "contentType        3047 non-null object\n",
      "url                3047 non-null object\n",
      "title              3047 non-null object\n",
      "text               3047 non-null object\n",
      "lang               3047 non-null object\n",
      "dtypes: int64(4), object(5)\n",
      "memory usage: 238.0+ KB\n"
     ]
    }
   ],
   "source": [
    "articles_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 70028 entries, 0 to 72311\n",
      "Data columns (total 5 columns):\n",
      "timestamp    70028 non-null int64\n",
      "eventType    70028 non-null object\n",
      "contentId    70028 non-null int64\n",
      "personId     70028 non-null int64\n",
      "sessionId    70028 non-null int64\n",
      "dtypes: int64(4), object(1)\n",
      "memory usage: 3.2+ MB\n"
     ]
    }
   ],
   "source": [
    "interactions_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 数据预处理 \n",
    "\n",
    " 接下来我们要对shared_articles.csv和users_interactions.csv中的数据进行整理，并删除一些不需要的列,并将两个表进行合并"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>contentId</th>\n",
       "      <th>personId</th>\n",
       "      <th>eventType</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-8845298781299428018</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Hiri wants to fix the workplace email problem</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-8845298781299428018</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Hiri wants to fix the workplace email problem</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-108842214936804958</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Hiri wants to fix the workplace email problem</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-1443636648652872475</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Hiri wants to fix the workplace email problem</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-3499919498720038879</td>\n",
       "      <td>-1443636648652872475</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Hiri wants to fix the workplace email problem</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             contentId             personId eventType  \\\n",
       "0 -3499919498720038879 -8845298781299428018      VIEW   \n",
       "1 -3499919498720038879 -8845298781299428018      VIEW   \n",
       "2 -3499919498720038879  -108842214936804958      VIEW   \n",
       "3 -3499919498720038879 -1443636648652872475      VIEW   \n",
       "4 -3499919498720038879 -1443636648652872475      VIEW   \n",
       "\n",
       "                                           title  \n",
       "0  Hiri wants to fix the workplace email problem  \n",
       "1  Hiri wants to fix the workplace email problem  \n",
       "2  Hiri wants to fix the workplace email problem  \n",
       "3  Hiri wants to fix the workplace email problem  \n",
       "4  Hiri wants to fix the workplace email problem  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.merge(interactions_df[['contentId','personId', 'eventType']], articles_df[['contentId', 'title']], how = 'inner', on = 'contentId')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 69985 entries, 0 to 69984\n",
      "Data columns (total 4 columns):\n",
      "contentId    69985 non-null int64\n",
      "personId     69985 non-null int64\n",
      "eventType    69985 non-null object\n",
      "title        69985 non-null object\n",
      "dtypes: int64(2), object(2)\n",
      "memory usage: 2.7+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\"eventType\"字段记录了用户行为的类别，因此我们需要查看一下用户行为的分类情况："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "VIEW               58759\n",
       "LIKE                5745\n",
       "BOOKMARK            2463\n",
       "COMMENT CREATED     1611\n",
       "FOLLOW              1407\n",
       "Name: eventType, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['eventType'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们看到存在以下5种用户行为分类:\n",
    "\n",
    "* VIEW: 查看，表示用户点击过该篇文章\n",
    "* LIKE:  点赞，表示用户喜好该篇文章。\n",
    "* BOOKMARK: 加入书签，用户已将该文章加入书签，以便将来查看。这是用户对该文章表现出感兴趣的迹象\n",
    "* COMMENT CREATED: 创建评论，说明用户被这篇文章所吸引，同时希望表达自己的观点。\n",
    "* FOLLOW: 关注，用户关注了文章的作者，这应该说明用户对这位作者的文章非常感兴趣  \n",
    "\n",
    "我们可以认为上述的5种用户行为可以表达出用户对文章的不同的喜好程度，为此我们可以为这5种用户行为各打上一个分值，以便量化用户的喜好程度:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>contentId</th>\n",
       "      <th>personId</th>\n",
       "      <th>eventType</th>\n",
       "      <th>title</th>\n",
       "      <th>eventStrength</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>68299</th>\n",
       "      <td>2760335717049716507</td>\n",
       "      <td>-2726721797588771398</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Razorfish, US digital revenues, drag down Publ...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10531</th>\n",
       "      <td>2602738827388230683</td>\n",
       "      <td>-1443636648652872475</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Apple is working on an AI system that wipes th...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34808</th>\n",
       "      <td>-5879360586463363298</td>\n",
       "      <td>7741213197781008422</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Introducing LambCI - a serverless build system</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2511</th>\n",
       "      <td>5964888566269323942</td>\n",
       "      <td>4670267857749552625</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>PwC faz virada digital e já fatura US$ 1 bilhã...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39484</th>\n",
       "      <td>-2447632164766022033</td>\n",
       "      <td>-3643155458357242906</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>O que o GitHub tem a nos dizer sobre os estere...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54306</th>\n",
       "      <td>6031953227014493100</td>\n",
       "      <td>3007673865213720513</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>O chefe é gay. E daí?</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31294</th>\n",
       "      <td>-5014627593450767720</td>\n",
       "      <td>8271619811366074205</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Drupal and ambitious digital experiences</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5329</th>\n",
       "      <td>-3581194288660477595</td>\n",
       "      <td>-1032019229384696495</td>\n",
       "      <td>LIKE</td>\n",
       "      <td>The End Of Apple Man</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14908</th>\n",
       "      <td>880612740433495828</td>\n",
       "      <td>6013226412048763966</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>O que você deve fazer para se tornar um líder ...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52071</th>\n",
       "      <td>3367026768872537336</td>\n",
       "      <td>-9223121837663643404</td>\n",
       "      <td>VIEW</td>\n",
       "      <td>Seja esperto no trabalho: Melhore a comunicaçã...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 contentId             personId eventType  \\\n",
       "68299  2760335717049716507 -2726721797588771398      VIEW   \n",
       "10531  2602738827388230683 -1443636648652872475      VIEW   \n",
       "34808 -5879360586463363298  7741213197781008422      VIEW   \n",
       "2511   5964888566269323942  4670267857749552625      VIEW   \n",
       "39484 -2447632164766022033 -3643155458357242906      VIEW   \n",
       "54306  6031953227014493100  3007673865213720513      VIEW   \n",
       "31294 -5014627593450767720  8271619811366074205      VIEW   \n",
       "5329  -3581194288660477595 -1032019229384696495      LIKE   \n",
       "14908   880612740433495828  6013226412048763966      VIEW   \n",
       "52071  3367026768872537336 -9223121837663643404      VIEW   \n",
       "\n",
       "                                                   title  eventStrength  \n",
       "68299  Razorfish, US digital revenues, drag down Publ...            1.0  \n",
       "10531  Apple is working on an AI system that wipes th...            1.0  \n",
       "34808     Introducing LambCI - a serverless build system            1.0  \n",
       "2511   PwC faz virada digital e já fatura US$ 1 bilhã...            1.0  \n",
       "39484  O que o GitHub tem a nos dizer sobre os estere...            1.0  \n",
       "54306                              O chefe é gay. E daí?            1.0  \n",
       "31294           Drupal and ambitious digital experiences            1.0  \n",
       "5329                                The End Of Apple Man            2.0  \n",
       "14908  O que você deve fazer para se tornar um líder ...            1.0  \n",
       "52071  Seja esperto no trabalho: Melhore a comunicaçã...            1.0  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "event_type_strength = {\n",
    "   'VIEW': 1.0,\n",
    "   'LIKE': 2.0, \n",
    "   'BOOKMARK': 3.0, \n",
    "   'FOLLOW': 4.0,\n",
    "   'COMMENT CREATED': 5.0,  \n",
    "}\n",
    "\n",
    "df['eventStrength'] = df['eventType'].apply(lambda x: event_type_strength[x])\n",
    "df.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>personId</th>\n",
       "      <th>contentId</th>\n",
       "      <th>title</th>\n",
       "      <th>eventStrength</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1758</th>\n",
       "      <td>-8674958742744576254</td>\n",
       "      <td>-1452340812018195881</td>\n",
       "      <td>Globo G1 lança Bot para o Telegram que te ajud...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39138</th>\n",
       "      <td>8562738815734604501</td>\n",
       "      <td>-4625438649010928449</td>\n",
       "      <td>Top Asset Tracking Software: The 52 Best Tools...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26505</th>\n",
       "      <td>2754566407772265068</td>\n",
       "      <td>8582210754044680874</td>\n",
       "      <td>Reactive log stream processing with RxJava - P...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18400</th>\n",
       "      <td>-830175562779396891</td>\n",
       "      <td>-1178371721262103855</td>\n",
       "      <td>Ultrapar, por meio da Ipiranga, assina compra ...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23205</th>\n",
       "      <td>1120069409160402054</td>\n",
       "      <td>-1897853393557630253</td>\n",
       "      <td>Your health insurance might score you an Apple...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16885</th>\n",
       "      <td>-1393866732742189886</td>\n",
       "      <td>7459643459148487877</td>\n",
       "      <td>Grupo de bancos adere à tecnologia do bitcoin ...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30406</th>\n",
       "      <td>3813842765808990208</td>\n",
       "      <td>5962268176939710955</td>\n",
       "      <td>BigQuery 1.11, now with Standard SQL, IAM, and...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5312</th>\n",
       "      <td>-6874532444478397764</td>\n",
       "      <td>897043351348716074</td>\n",
       "      <td>Mark Zuckerberg compartilha como será o Facebo...</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10020</th>\n",
       "      <td>-4028919343899978105</td>\n",
       "      <td>-912681925384819247</td>\n",
       "      <td>15 Interesting JavaScript and CSS Libraries fo...</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15120</th>\n",
       "      <td>-1799631734242668035</td>\n",
       "      <td>-746073086109727488</td>\n",
       "      <td>How to get the most from your agency relations...</td>\n",
       "      <td>12.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  personId            contentId  \\\n",
       "1758  -8674958742744576254 -1452340812018195881   \n",
       "39138  8562738815734604501 -4625438649010928449   \n",
       "26505  2754566407772265068  8582210754044680874   \n",
       "18400  -830175562779396891 -1178371721262103855   \n",
       "23205  1120069409160402054 -1897853393557630253   \n",
       "16885 -1393866732742189886  7459643459148487877   \n",
       "30406  3813842765808990208  5962268176939710955   \n",
       "5312  -6874532444478397764   897043351348716074   \n",
       "10020 -4028919343899978105  -912681925384819247   \n",
       "15120 -1799631734242668035  -746073086109727488   \n",
       "\n",
       "                                                   title  eventStrength  \n",
       "1758   Globo G1 lança Bot para o Telegram que te ajud...            1.0  \n",
       "39138  Top Asset Tracking Software: The 52 Best Tools...            1.0  \n",
       "26505  Reactive log stream processing with RxJava - P...            1.0  \n",
       "18400  Ultrapar, por meio da Ipiranga, assina compra ...            1.0  \n",
       "23205  Your health insurance might score you an Apple...            1.0  \n",
       "16885  Grupo de bancos adere à tecnologia do bitcoin ...            1.0  \n",
       "30406  BigQuery 1.11, now with Standard SQL, IAM, and...            1.0  \n",
       "5312   Mark Zuckerberg compartilha como será o Facebo...            1.0  \n",
       "10020  15 Interesting JavaScript and CSS Libraries fo...            3.0  \n",
       "15120  How to get the most from your agency relations...           12.0  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.drop_duplicates()\n",
    "grouped_df = df.groupby(['personId', 'contentId', 'title']).sum().reset_index()\n",
    "grouped_df.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "personId           int64\n",
       "contentId          int64\n",
       "title             object\n",
       "eventStrength    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "personId和contentId都是int64的长整型字段，为了避免出现负整数警告信息以及为了节省内存空间，我们可以将他们转换成int16的短整型字段:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>personId</th>\n",
       "      <th>contentId</th>\n",
       "      <th>title</th>\n",
       "      <th>eventStrength</th>\n",
       "      <th>person_id</th>\n",
       "      <th>content_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>16568</th>\n",
       "      <td>-1443636648652872475</td>\n",
       "      <td>5314107418764884230</td>\n",
       "      <td>Deep learning software knows that a rose is a ...</td>\n",
       "      <td>10.0</td>\n",
       "      <td>798</td>\n",
       "      <td>2343</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9002</th>\n",
       "      <td>-4447090897576014728</td>\n",
       "      <td>-4130193294870737688</td>\n",
       "      <td>Game of Loans</td>\n",
       "      <td>1.0</td>\n",
       "      <td>492</td>\n",
       "      <td>829</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5544</th>\n",
       "      <td>-6786856227257648356</td>\n",
       "      <td>9042192299854648021</td>\n",
       "      <td>Google's cloud platform is now a force to be r...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>245</td>\n",
       "      <td>2952</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33662</th>\n",
       "      <td>5321535590192312132</td>\n",
       "      <td>-5813211845057621660</td>\n",
       "      <td>Communication patterns in a Microservice world</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1488</td>\n",
       "      <td>558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1096</th>\n",
       "      <td>-8860671864164757449</td>\n",
       "      <td>3608243634452078989</td>\n",
       "      <td>Carne artificial acabará com o agronegócio bra...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>29</td>\n",
       "      <td>2058</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9329</th>\n",
       "      <td>-4298177694576275482</td>\n",
       "      <td>2301338403202434417</td>\n",
       "      <td>Top 10 GCP sessions from Google I/O 2016</td>\n",
       "      <td>1.0</td>\n",
       "      <td>509</td>\n",
       "      <td>1833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7095</th>\n",
       "      <td>-5706287032724665714</td>\n",
       "      <td>-4707863452603278075</td>\n",
       "      <td>PSA: iOS 9.3.5 fixes serious security exploit ...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>348</td>\n",
       "      <td>741</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14147</th>\n",
       "      <td>-2511855597392146401</td>\n",
       "      <td>-8909927294729358771</td>\n",
       "      <td>Android N Developer Preview 2, out today! | An...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>693</td>\n",
       "      <td>75</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25494</th>\n",
       "      <td>2279740393166882579</td>\n",
       "      <td>-6810702648836067504</td>\n",
       "      <td>diagramas para entender, criar, inovar e empre...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1182</td>\n",
       "      <td>401</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29014</th>\n",
       "      <td>3609194402293569455</td>\n",
       "      <td>-3095494965842381869</td>\n",
       "      <td>Fintechs can help incumbents, not just disrupt...</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1319</td>\n",
       "      <td>998</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  personId            contentId  \\\n",
       "16568 -1443636648652872475  5314107418764884230   \n",
       "9002  -4447090897576014728 -4130193294870737688   \n",
       "5544  -6786856227257648356  9042192299854648021   \n",
       "33662  5321535590192312132 -5813211845057621660   \n",
       "1096  -8860671864164757449  3608243634452078989   \n",
       "9329  -4298177694576275482  2301338403202434417   \n",
       "7095  -5706287032724665714 -4707863452603278075   \n",
       "14147 -2511855597392146401 -8909927294729358771   \n",
       "25494  2279740393166882579 -6810702648836067504   \n",
       "29014  3609194402293569455 -3095494965842381869   \n",
       "\n",
       "                                                   title  eventStrength  \\\n",
       "16568  Deep learning software knows that a rose is a ...           10.0   \n",
       "9002                                       Game of Loans            1.0   \n",
       "5544   Google's cloud platform is now a force to be r...            1.0   \n",
       "33662     Communication patterns in a Microservice world            1.0   \n",
       "1096   Carne artificial acabará com o agronegócio bra...            1.0   \n",
       "9329            Top 10 GCP sessions from Google I/O 2016            1.0   \n",
       "7095   PSA: iOS 9.3.5 fixes serious security exploit ...            1.0   \n",
       "14147  Android N Developer Preview 2, out today! | An...            3.0   \n",
       "25494  diagramas para entender, criar, inovar e empre...            1.0   \n",
       "29014  Fintechs can help incumbents, not just disrupt...            1.0   \n",
       "\n",
       "       person_id  content_id  \n",
       "16568        798        2343  \n",
       "9002         492         829  \n",
       "5544         245        2952  \n",
       "33662       1488         558  \n",
       "1096          29        2058  \n",
       "9329         509        1833  \n",
       "7095         348         741  \n",
       "14147        693          75  \n",
       "25494       1182         401  \n",
       "29014       1319         998  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df['title'] = grouped_df['title'].astype(\"category\")\n",
    "grouped_df['personId'] = grouped_df['personId'].astype(\"category\")\n",
    "grouped_df['contentId'] = grouped_df['contentId'].astype(\"category\")\n",
    "grouped_df['person_id'] = grouped_df['personId'].cat.codes\n",
    "grouped_df['content_id'] = grouped_df['contentId'].cat.codes\n",
    "grouped_df.sample(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "personId         category\n",
       "contentId        category\n",
       "title            category\n",
       "eventStrength     float64\n",
       "person_id           int16\n",
       "content_id          int16\n",
       "dtype: object"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这样我们就可以不再需要personId和contentId这两个字段了。"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 交替最小二乘法（Alternating Least Squares）\n",
    "\n",
    "在矩阵分解（matrix factorization）中使用的一种算法。有一个稀疏矩阵，假设这个矩阵是低阶的，可以分解成两个小矩阵相乘。然后交替对两个小矩阵使用最小二乘法，算出这两个小矩阵，就可以估算出稀疏矩阵缺失的值：\n",
    "\n",
    "![image.png](attachment:image.png)\n",
    "\n",
    "假设用户评分矩阵可以分解成两个矩阵相乘的形式,这两个矩阵相乘的结果就可以近似等于原来的矩阵，为了寻找出最优的两个矩阵以便它们相乘以后的结果与原来的矩阵误差最小，我们需要使用交替最小二乘法，算法在迭代的过程中交替固定两个矩阵中的一个矩阵然后使用最小二乘法求出另外一个矩阵的最优矩阵，最终使两个矩阵都是最优矩阵并且它们乘积的结果与原来的矩阵误差最小。\n",
    "\n",
    "\n",
    "在这里我们将使用implicit库中的als.AlternatingLeastSquares方法来实现交替最小二乘法。不过，这里我们首先要创建两个稀疏矩阵:内容-人-评分矩阵，人-内容-评分矩阵。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2979, 1895)\n",
      "(1895, 2979)\n"
     ]
    }
   ],
   "source": [
    "sparse_content_person = sparse.csr_matrix((grouped_df['eventStrength'].astype(float), (grouped_df['content_id'], grouped_df['person_id'])))\n",
    "sparse_person_content = sparse.csr_matrix((grouped_df['eventStrength'].astype(float), (grouped_df['person_id'], grouped_df['content_id'])))\n",
    "\n",
    "print(sparse_content_person.shape)\n",
    "print(sparse_person_content.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们看到文章总数为2979，用户总数为1895 \n",
    "\n",
    "内容-人-评分矩阵(sparse_content_person)的每一行代表一篇文章，每一列代表一个用户。\n",
    "\n",
    "人-内容-评分矩阵(sparse_person_content)的每一行代表一个用户，每一列代表一篇文章。\n",
    "\n",
    "下面我们开始训练模型，并设置20个特征因子。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 训练模型\n",
    "\n",
    "下面我们开始训练模型，并设置20个特征因子。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Intel MKL BLAS detected. Its highly recommend to set the environment variable 'export MKL_NUM_THREADS=1' to disable its internal multithreading\n",
      "100%|████████████████████████████████████████| 50.0/50 [00:01<00:00, 32.34it/s]\n"
     ]
    }
   ],
   "source": [
    "alpha = 15\n",
    "data = (sparse_content_person * alpha).astype('double')\n",
    "\n",
    "model = implicit.als.AlternatingLeastSquares(factors=20, regularization=0.1, iterations=50)\n",
    "model.fit(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 推荐内容相似的文章\n",
    "\n",
    "模型训练完成以后，我们可以计算文章之间的相似度，并根据相似度来进行推荐:我们以content_id = 235的文章为例,这篇文章的标题是\"Artificial intelligence is hard to see\"，这是一篇有关人工智能的文章，我们的目标是在所有的文章的标题中找出和它最相似的10篇文章标题。这里我们主要通过计算文章之间的相似度，在计算相似度的过程中有以下几个步骤:\n",
    "\n",
    "从模型中获取人和内容的向量\n",
    "计算内容的向量的范数\n",
    "计算相似度得分\n",
    "获取相似度得分最大的10篇文章\n",
    "生成这10篇文章的content_id和title的元组"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Artificial intelligence is hard to see\n"
     ]
    }
   ],
   "source": [
    "content_id = 235\n",
    "n_similar = 10\n",
    "print(grouped_df.title.loc[grouped_df.content_id == content_id][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1895, 20)\n",
      "(2979, 20)\n"
     ]
    }
   ],
   "source": [
    "#获取用户矩阵\n",
    "person_vecs = model.user_factors\n",
    "#获取内容矩阵\n",
    "content_vecs = model.item_factors\n",
    "#计算内容的向量的范数\n",
    "content_norms = np.sqrt((content_vecs * content_vecs).sum(axis=1))\n",
    "#计算指定的content_id 与其他所有文章的相似度\n",
    "scores = content_vecs.dot(content_vecs[content_id]) / content_norms\n",
    "#获取相似度最大的10篇文章\n",
    "top_idx = np.argpartition(scores, -n_similar)[-n_similar:]\n",
    "#组成content_id和title的元组\n",
    "similar = sorted(zip(top_idx, scores[top_idx] / content_norms[content_id]), key=lambda x: -x[1])\n",
    "\n",
    "print(person_vecs.shape)\n",
    "print(content_vecs.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "我们注意到我们的用户矩阵和我们的内容矩阵都只有20列，这是因为我们在定义模型时设置了factors=20。这里的用户矩阵和内容矩阵就类似与前面交替最小二乘法示意图中介绍的User Feature Matrix和Movie Feature Matrix.\n",
    "\n",
    "下面我们展示这10篇最相似的文章title:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Artificial intelligence is hard to see | 1.0000001\n",
      "Machine Learning Is No Longer Just for Experts | 0.75046736\n",
      "Clarifying the uses of artificial intelligence in the enterprise | 0.7082239\n",
      "Google Calendar for Android can now find meeting times that work for everyone | 0.6599234\n",
      "Money 20/20 Panel: Artificial Intelligence and Machine Learning | 0.65292907\n",
      "Leonardo Dicaprio, Barry Sternlicht back Qloo, a cultural recommendation engine | 0.6389775\n",
      "A trip to Google's New York City Pop-Up Shop | 0.62346613\n",
      "Stop Coding Machine Learning Algorithms From Scratch - Machine Learning Mastery | 0.6212184\n",
      "Larry Page's dream of building a 'Google Island' could be coming true | 0.6211126\n",
      "10 documentários para empreendedores | 0.617943\n"
     ]
    }
   ],
   "source": [
    "for content in similar:\n",
    "    idx, score = content\n",
    "    print(grouped_df.title.loc[grouped_df.content_id == idx].iloc[0],\"|\",score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "这些文章似乎都和人工智能有关。其中第一篇文章就是content_id = 235的文章本身，正如我们前面说content_id = 235要和所有文章计算相似度，这当然也包含它自己本身，所以content_id = 235的文章和自己的相似度应该是最大的，因此应该排第一位，其余9篇文章标题则是按照相似度由高到低排列的。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 为用户推荐他可能感兴趣的文章\n",
    "\n",
    "接下来我们要为用户推荐他们没有看过的(即没有发生过交互行为),但是他们可能会敢兴趣的文章。我们首先要定义一个推荐函数，在推荐函数中我们主要要做以下3件事:\n",
    "\n",
    "* 将指定用户的用户向量乘以内容矩阵，得到该用户对所有文章的评分向量\n",
    "* 从评分向量中过滤掉用户已经评分过的文章(将其评分值置为0),因为用户已经发生过交互行为的文章不应该被推荐\n",
    "* 将余下的评分分数排序，并输出分数最大的10篇文章。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def recommend(person_id, sparse_person_content, person_vecs, content_vecs, num_contents=10):\n",
    "    #*****************得到指定用户对所有文章的评分向量******************************\n",
    "    # 将该用户向量乘以内容矩阵(做点积),得到该用户对所有文章的评价分数向量\n",
    "    rec_vector = person_vecs[person_id,:].dot(content_vecs.T).toarray()\n",
    "    \n",
    "    #**********过滤掉用户已经评分过的文章(将其评分值置为0),因为用户已经发生过交互行为的文章不应该被推荐*******\n",
    "    # 从稀疏矩阵sparse_person_content中获取指定用户对所有文章的评价分数\n",
    "    person_interactions = sparse_person_content[person_id,:].toarray()\n",
    "    # 为该用户的对所有文章的评价分数+1，那些没有被该用户看过(view)的文章的分数就会等于1(原来是0)\n",
    "    person_interactions = person_interactions.reshape(-1) + 1\n",
    "    # 将那些已经被该用户看过的文章的分数置为0\n",
    "    person_interactions[person_interactions > 1] = 0        \n",
    "    # 将该用户的评分向量做标准化处理,将其值缩放到0到1之间。\n",
    "    min_max = MinMaxScaler()\n",
    "    rec_vector_scaled = min_max.fit_transform(rec_vector.reshape(-1,1))[:,0]\n",
    "    # 过滤掉和该用户已经交互过的文章，这些文章的评分会和0相乘。\n",
    "    recommend_vector = person_interactions * rec_vector_scaled\n",
    "    \n",
    "    #*************将余下的评分分数排序，并输出分数最大的10篇文章************************\n",
    "    # 根据评分值进行排序,并获取指定数量的评分值最高的文章\n",
    "    content_idx = np.argsort(recommend_vector)[::-1][:num_contents]\n",
    "    \n",
    "    # 定义两个list用于存储文章的title和推荐分数。\n",
    "    titles = []\n",
    "    scores = []\n",
    "\n",
    "    for idx in content_idx:\n",
    "        # 将title和分数添加到list中\n",
    "        titles.append(grouped_df.title.loc[grouped_df.content_id == idx].iloc[0])\n",
    "        scores.append(recommend_vector[idx])\n",
    "\n",
    "    recommendations = pd.DataFrame({'title': titles, 'score': scores})\n",
    "\n",
    "    return recommendations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "在上面的函数中，我们首先从稀疏矩阵sparse_person_content中获取指定用户对所有文章的评分，这里需要注意一点的是，如果某个用户对某篇文章没有发生过交互行为(VIEW,LIKE,BOOKMARK,COMMENT CREATED,FOLLOW),那么在原始的数据集中是不存在这条交互记录的，但是当我们使用了稀疏矩阵的.toarray()方法后，那些用户没有发生过交互的所有文章都会被展示出来，只不过对那些文章的评分值都会被置为0，因此toarray()方法展现的是所有用户对所有文章的评分结果。\n",
    "\n",
    "下面我们要为特定的用户推荐他们没有看过的，但可能会感兴趣的10篇文章："
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                               title     score\n",
      "0  Custo do Erro - Cinco motivos para investir em...  1.000000\n",
      "1  Ray Kurzweil: The world isn't getting worse - ...  0.859316\n",
      "2  How I built an app with 500,000 users in 5 day...  0.827874\n",
      "3           Drupal and ambitious digital experiences  0.815111\n",
      "4  Do You Suffer From Deployment Anxiety? - DZone...  0.812248\n",
      "5  Former Google career coach shares a visual tri...  0.788370\n",
      "6  'The Simpsons' celebrates 600 episodes with a ...  0.781911\n",
      "7  GitLab launches Issue Boards, an open-source t...  0.768058\n",
      "8                            15 minutos sobre Docker  0.757384\n",
      "9  BDD Best Practices and Guidelines - Testing Ex...  0.753251\n"
     ]
    }
   ],
   "source": [
    "# 从model中获取经过训练的用户和内容矩阵,并将它们存储为稀疏矩阵\n",
    "person_vecs = sparse.csr_matrix(model.user_factors)\n",
    "content_vecs = sparse.csr_matrix(model.item_factors)\n",
    "\n",
    "# 为指定用户推荐文章。\n",
    "person_id = 50\n",
    "recommendations = recommend(person_id, sparse_person_content, person_vecs, content_vecs)\n",
    "print(recommendations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>title</th>\n",
       "      <th>person_id</th>\n",
       "      <th>eventStrength</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1727</th>\n",
       "      <td>Acquia Engage 2016: Day One</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1791</th>\n",
       "      <td>Um bilhão de arquivos mostram quem vence a dis...</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1781</th>\n",
       "      <td>Acquia Engage Awards Finalists Announced</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1778</th>\n",
       "      <td>Sharing innovation with your competitors - Dri...</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1769</th>\n",
       "      <td>Don't document your code. Code your documentat...</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1747</th>\n",
       "      <td>Who sponsors Drupal development? | Dries Buytaert</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1768</th>\n",
       "      <td>Johnson &amp; Johnson comprará grupo suíço por US$...</td>\n",
       "      <td>50</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1767</th>\n",
       "      <td>Slack and Google announce partnership focused ...</td>\n",
       "      <td>50</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1770</th>\n",
       "      <td>Rating the English Proficiency of Countries an...</td>\n",
       "      <td>50</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1766</th>\n",
       "      <td>Infográfico: Algoritmos para Aprendizado de Má...</td>\n",
       "      <td>50</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  title  person_id  \\\n",
       "1727                        Acquia Engage 2016: Day One         50   \n",
       "1791  Um bilhão de arquivos mostram quem vence a dis...         50   \n",
       "1781           Acquia Engage Awards Finalists Announced         50   \n",
       "1778  Sharing innovation with your competitors - Dri...         50   \n",
       "1769  Don't document your code. Code your documentat...         50   \n",
       "1747  Who sponsors Drupal development? | Dries Buytaert         50   \n",
       "1768  Johnson & Johnson comprará grupo suíço por US$...         50   \n",
       "1767  Slack and Google announce partnership focused ...         50   \n",
       "1770  Rating the English Proficiency of Countries an...         50   \n",
       "1766  Infográfico: Algoritmos para Aprendizado de Má...         50   \n",
       "\n",
       "      eventStrength  \n",
       "1727            3.0  \n",
       "1791            3.0  \n",
       "1781            3.0  \n",
       "1778            3.0  \n",
       "1769            3.0  \n",
       "1747            3.0  \n",
       "1768            1.0  \n",
       "1767            1.0  \n",
       "1770            1.0  \n",
       "1766            1.0  "
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df.loc[grouped_df['person_id'] == 50].sort_values(by=['eventStrength'], \n",
    "                                                          ascending=False)[['title', 'person_id', 'eventStrength']].head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                               title     score\n",
      "0  Novo workaholic trabalha, pratica esportes e t...  0.814111\n",
      "1                   Livro: Retrospectivas Divertidas  0.808701\n",
      "2  'The Simpsons' celebrates 600 episodes with a ...  0.765099\n",
      "3               Why Programmers Want Private Offices  0.682038\n",
      "4           Drupal and ambitious digital experiences  0.663015\n",
      "5          Speeding up ReSharper (and Visual Studio)  0.656806\n",
      "6                    40 Basic Japanese conversations  0.650549\n",
      "7  How to Improve 8 Major Problem Areas for Japan...  0.641954\n",
      "8                               Japanese for dummies  0.627957\n",
      "9  Ganhe 6 meses de acesso ao Pluralsight, maior ...  0.605869\n"
     ]
    }
   ],
   "source": [
    "person_id = 2\n",
    "recommendations = recommend(person_id, sparse_person_content, person_vecs, content_vecs)\n",
    "print(recommendations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>title</th>\n",
       "      <th>eventStrength</th>\n",
       "      <th>person_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>Former Google career coach shares a visual tri...</td>\n",
       "      <td>6.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>Request lesson : How and when to use はず(=hazu)...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>Aposta na inovação</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>The Algorithm March, Japan's Strangely Enterta...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>Como são escrita as risadas em japonês? - Suki...</td>\n",
       "      <td>3.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>A minha viagem à Maternidade #tetodomundo</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>Learn Hiragana: The Ultimate Guide</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                title  eventStrength  \\\n",
       "51  Former Google career coach shares a visual tri...            6.0   \n",
       "48  Request lesson : How and when to use はず(=hazu)...            3.0   \n",
       "49                                 Aposta na inovação            3.0   \n",
       "50  The Algorithm March, Japan's Strangely Enterta...            3.0   \n",
       "54  Como são escrita as risadas em japonês? - Suki...            3.0   \n",
       "52          A minha viagem à Maternidade #tetodomundo            1.0   \n",
       "53                 Learn Hiragana: The Ultimate Guide            1.0   \n",
       "\n",
       "    person_id  \n",
       "51          2  \n",
       "48          2  \n",
       "49          2  \n",
       "50          2  \n",
       "54          2  \n",
       "52          2  \n",
       "53          2  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df.loc[grouped_df['person_id'] == 2].sort_values(by=['eventStrength'], \n",
    "                                                         ascending=False)[['title', 'eventStrength', 'person_id']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                               title     score\n",
      "0  Como são escrita as risadas em japonês? - Suki...  0.849396\n",
      "1  Former Google career coach shares a visual tri...  0.848883\n",
      "2  Ganhe 6 meses de acesso ao Pluralsight, maior ...  0.784309\n",
      "3          Speeding up ReSharper (and Visual Studio)  0.784002\n",
      "4             The Continuous Delivery Maturity Model  0.764710\n",
      "5  'The Simpsons' celebrates 600 episodes with a ...  0.735431\n",
      "6  Node.js Examples - How Enterprises use Node in...  0.724938\n",
      "7  Request lesson : How and when to use はず(=hazu)...  0.715854\n",
      "8  Desenvolvimento Efetivo na Plataforma Microsof...  0.696465\n",
      "9                               Jenkins 2.0 is here!  0.685920\n"
     ]
    }
   ],
   "source": [
    "person_id = 1\n",
    "recommendations = recommend(person_id, sparse_person_content, person_vecs, content_vecs)\n",
    "print(recommendations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>title</th>\n",
       "      <th>eventStrength</th>\n",
       "      <th>person_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>Learn Hiragana: The Ultimate Guide</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>Firebase Test Lab for Android</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>Fresco, sim! - Android Dev BR</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>Japanese for dummies</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>Firebase and Google Cloud: better together</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                         title  eventStrength  person_id\n",
       "44          Learn Hiragana: The Ultimate Guide            3.0          1\n",
       "43               Firebase Test Lab for Android            1.0          1\n",
       "45               Fresco, sim! - Android Dev BR            1.0          1\n",
       "46                        Japanese for dummies            1.0          1\n",
       "47  Firebase and Google Cloud: better together            1.0          1"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_df.loc[grouped_df['person_id'] == 1].sort_values(by=['eventStrength'], \n",
    "                                                         ascending=False)[['title', 'eventStrength', 'person_id']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 推荐系统的评估 \n",
    "\n",
    "在实际应用中要检验推荐系统的表现，最重要的还是要看推荐系统是否为企业带来了更多的利润，或者是否为网站带来了更多的流量等商业价值。\n",
    "\n",
    "在这里我们只能通过理论的方式来检验推荐系统的有效性，大家可以参考这篇文章，也可以参考我之前写的一篇关于评估推荐系统表现的博客，还有这篇关于ROC和AUC的博客。\n",
    "\n",
    "大体上来说评估推荐系统的表现主要是通过计算推荐结果的“命中率”来考察推荐算法表现，主要思想是这样的，我们从现有的评分矩阵中分离出少部分评分数据(如20%左右),将剩余的80%的推荐数据用来训练推荐算法模型，然后让推荐模型对用户未评分过的文章进行推荐，在推荐结果中我们考察其中是否包含了之前被分离出来的那20%的文章，同时我们计算“命中率”或AUC作为评价指标。\n",
    "\n",
    "现在我们要做的是从评分数据中创建训练集和测试集，在测试集中我们删除了20%的有过交互行为的评分数据，在测试集中我们将所有的有过交互行为评分置为1，这样就测试集变成了一个二分类数据的集合。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "def make_train(ratings, pct_test = 0.2):\n",
    "    test_set = ratings.copy() # 拷贝一份评分数据当作测试集\n",
    "    test_set[test_set != 0] = 1 # 将有评分数据置为1，我们要模拟成二分类数据集\n",
    "    \n",
    "    training_set = ratings.copy() # 拷贝一份评分数据当作训练集 \n",
    "    \n",
    "    nonzero_inds = training_set.nonzero() # 找到有过评分(有交互行为，评分数不为0)的数据的索引。\n",
    "    nonzero_pairs = list(zip(nonzero_inds[0], nonzero_inds[1])) # 将它们组成元组并存放在list中\n",
    "\n",
    "    \n",
    "    random.seed(0) #设置随机数种子\n",
    "    \n",
    "    num_samples = int(np.ceil(pct_test*len(nonzero_pairs))) # 获取20%的非0评价的数量\n",
    "    samples = random.sample(nonzero_pairs, num_samples) # 随机从非零评价的索引对中抽样20%\n",
    "\n",
    "    content_inds = [index[0] for index in samples] # 从样本中得到文章列(第一列)索引值\n",
    "\n",
    "    person_inds = [index[1] for index in samples] # 从样本中得到文章列(第二列)索引值\n",
    "\n",
    "    \n",
    "    training_set[content_inds, person_inds] = 0 # 在训练集中将这20%的随机样本的评分值置为0\n",
    "    training_set.eliminate_zeros() # 在测试集中删除这0元素\n",
    "    \n",
    "    return training_set, test_set, list(set(person_inds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "content_train, content_test, content_persons_altered = make_train(sparse_content_person, pct_test = 0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "#计算AUC分数\n",
    "def auc_score(predictions, actual):\n",
    "    fpr, tpr, thresholds = metrics.roc_curve(actual, predictions)\n",
    "    return metrics.auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "#计算评价AUC分数\n",
    "def calc_mean_auc(training_set, altered_persons, predictions, test_set):\n",
    "    store_auc = [] # 用来存储那些在训练集中被删除评分的用户的AUC\n",
    "    popularity_auc = [] # 用来存储最受欢迎的文章的AUC\n",
    "    pop_contents = np.array(test_set.sum(axis = 1)).reshape(-1) # 在测试集中按列合计所有评价分数，以便找出最受欢迎的文章。\n",
    "    content_vecs = predictions[1]\n",
    "    for person in altered_persons: # 迭代那些在训练集中被删除评分的那20%的用户\n",
    "        training_column = training_set[:,person].toarray().reshape(-1) # 在训练集中找到对应用户的那一列\n",
    "        zero_inds = np.where(training_column == 0) # 找出所有没有发生过交互行为的评分的索引,这其中也包括被删除评分的索引\n",
    "        \n",
    "        # 对用户没有交互过的文章预测用户对它们的评分\n",
    "        person_vec = predictions[0][person,:]\n",
    "        pred = person_vec.dot(content_vecs).toarray()[0,zero_inds].reshape(-1)\n",
    "        \n",
    "        #获取预测的评分，预测评分包含用户交互过的文章的评分(原评分为0)和那20%被强制置为0的实际评分\n",
    "        actual = test_set[:,person].toarray()[zero_inds,0].reshape(-1)\n",
    "        \n",
    "        #从所有文章评价总和中过滤出过滤出那么没有评价过的文章的合计总分(每篇文章各自的合计总分)\n",
    "        pop = pop_contents[zero_inds] \n",
    "        \n",
    "        store_auc.append(auc_score(pred, actual)) # 计算当前用户的预测和实际评分的AUC\n",
    "        \n",
    "        popularity_auc.append(auc_score(pop, actual)) # 计算合计总分和实际评分的AUC\n",
    "    \n",
    "    return float('%.3f'%np.mean(store_auc)), float('%.3f'%np.mean(popularity_auc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.98, 0.819)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calc_mean_auc(content_train, content_persons_altered,\n",
    "              [person_vecs, content_vecs.T], content_test)"
   ]
  },
  {
   "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": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}