natural_language_processing/example_text_summarization.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Natural Language Processing - Text Summarization example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Links: [CNN DailyMail (HuggingFace)](https://huggingface.co/datasets/cnn_dailymail)  |  [Article (TowardsDataScience)](https://towardsdatascience.com/text-summarization-with-nlp-textrank-vs-seq2seq-vs-bart-474943efeb09)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 0 - Setup"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Import pckgs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "from nlp_utils import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Import data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#import datasets\n",
    "#dataset = datasets.load_dataset(\"cnn_dailymail\", '3.0.0')\n",
    "\n",
    "#lst_dics = [dic for dic in dataset[\"train\"]]\n",
    "#dtf = pd.DataFrame(lst_dics).rename(columns={\"article\":\"text\", \"highlights\":\"y\"})[[\"text\",\"y\"]].head(20000)\n",
    "#dtf.to_csv(\"data_summary.csv\", index=False)\n",
    "#dtf.head()"
   ]
  },
  {
   "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>Unnamed: 0</th>\n",
       "      <th>text</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>It's official: U.S. President Barack Obama wan...</td>\n",
       "      <td>Syrian official: Obama climbed to the top of t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>(CNN) -- Usain Bolt rounded off the world cham...</td>\n",
       "      <td>Usain Bolt wins third gold of world championsh...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>Kansas City, Missouri (CNN) -- The General Ser...</td>\n",
       "      <td>The employee in agency's Kansas City office is...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>Los Angeles (CNN) -- A medical doctor in Vanco...</td>\n",
       "      <td>NEW: A Canadian doctor says she was part of a ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>(CNN) -- Police arrested another teen Thursday...</td>\n",
       "      <td>Another arrest made in gang rape outside Calif...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0                                               text  \\\n",
       "0           0  It's official: U.S. President Barack Obama wan...   \n",
       "1           1  (CNN) -- Usain Bolt rounded off the world cham...   \n",
       "2           2  Kansas City, Missouri (CNN) -- The General Ser...   \n",
       "3           3  Los Angeles (CNN) -- A medical doctor in Vanco...   \n",
       "4           4  (CNN) -- Police arrested another teen Thursday...   \n",
       "\n",
       "                                                   y  \n",
       "0  Syrian official: Obama climbed to the top of t...  \n",
       "1  Usain Bolt wins third gold of world championsh...  \n",
       "2  The employee in agency's Kansas City office is...  \n",
       "3  NEW: A Canadian doctor says she was part of a ...  \n",
       "4  Another arrest made in gang rape outside Calif...  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dtf = pd.read_csv(\"data_summary.csv\")\n",
    "dtf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 20000 entries, 0 to 19999\n",
      "Data columns (total 3 columns):\n",
      " #   Column      Non-Null Count  Dtype \n",
      "---  ------      --------------  ----- \n",
      " 0   Unnamed: 0  20000 non-null  int64 \n",
      " 1   text        20000 non-null  object\n",
      " 2   y           20000 non-null  object\n",
      "dtypes: int64(1), object(2)\n",
      "memory usage: 468.9+ KB\n"
     ]
    }
   ],
   "source": [
    "dtf.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Full text ---\n",
      "(CNN) -- Usain Bolt rounded off the world championships Sunday by claiming his third gold in Moscow as he anchored Jamaica to victory in the men's 4x100m relay. The fastest man in the world charged clear of United States rival Justin Gatlin as the Jamaican quartet of Nesta Carter, Kemar Bailey-Cole, Nickel Ashmeade and Bolt won in 37.36 seconds. The U.S finished second in 37.56 seconds with Canada taking the bronze after Britain were disqualified for a faulty handover. The 26-year-old Bolt has now collected eight gold medals at world championships, equaling the record held by American trio Carl Lewis, Michael Johnson and Allyson Felix, not to mention the small matter of six Olympic titles. The relay triumph followed individual successes in the 100 and 200 meters in the Russian capital. \"I'm proud of myself and I'll continue to work to dominate for as long as possible,\" Bolt said, having previously expressed his intention to carry on until the 2016 Rio Olympics. Victory was never seriously in doubt once he got the baton safely in hand from Ashmeade, while Gatlin and the United States third leg runner Rakieem Salaam had problems. Gatlin strayed out of his lane as he struggled to get full control of their baton and was never able to get on terms with Bolt. Earlier, Jamaica's women underlined their dominance in the sprint events by winning the 4x100m relay gold, anchored by Shelly-Ann Fraser-Pryce, who like Bolt was completing a triple. Their quartet recorded a championship record of 41.29 seconds, well clear of France, who crossed the line in second place in 42.73 seconds. Defending champions, the United States, were initially back in the bronze medal position after losing time on the second handover between Alexandria Anderson and English Gardner, but promoted to silver when France were subsequently disqualified for an illegal handover. The British quartet, who were initially fourth, were promoted to the bronze which eluded their men's team. Fraser-Pryce, like Bolt aged 26, became the first woman to achieve three golds in the 100-200 and the relay. In other final action on the last day of the championships, France's Teddy Tamgho became the third man to leap over 18m in the triple jump, exceeding the mark by four centimeters to take gold. Germany's Christina Obergfoll finally took gold at global level in the women's javelin after five previous silvers, while Kenya's Asbel Kiprop easily won a tactical men's 1500m final. Kiprop's compatriot Eunice Jepkoech Sum was a surprise winner of the women's 800m. Bolt's final dash for golden glory brought the eight-day championship to a rousing finale, but while the hosts topped the medal table from the United States there was criticism of the poor attendances in the Luzhniki Stadium. There was further concern when their pole vault gold medalist Yelena Isinbayeva made controversial remarks in support of Russia's new laws, which make \"the propagandizing of non-traditional sexual relations among minors\" a criminal offense. She later attempted to clarify her comments, but there were renewed calls by gay rights groups for a boycott of the 2014 Winter Games in Sochi, the next major sports event in Russia.\n",
      " \n",
      "--- Summary ---\n",
      "Usain Bolt wins third gold of world championship .\n",
      "Anchors Jamaica to 4x100m relay victory .\n",
      "Eighth gold at the championships for Bolt .\n",
      "Jamaica double up in women's 4x100m relay .\n"
     ]
    }
   ],
   "source": [
    "# check\n",
    "i = 1\n",
    "print(\"--- Full text ---\")\n",
    "print(dtf[\"text\"][i])\n",
    "print(\" \")\n",
    "print(\"--- Summary ---\")\n",
    "print(dtf[\"y\"][i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1 - Text Analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Text cleaning\n",
    "- Word frequency\n",
    "- Length analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Text cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "lst_stopwords = create_stopwords()\n",
    "lst_stopwords = lst_stopwords + [\"cnn\",\"say\",\"said\",\"new\",\"wa\",\"ha\"]"
   ]
  },
  {
   "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>Unnamed: 0</th>\n",
       "      <th>text</th>\n",
       "      <th>y</th>\n",
       "      <th>text_clean</th>\n",
       "      <th>y_clean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>It's official: U.S. President Barack Obama wan...</td>\n",
       "      <td>Syrian official: Obama climbed to the top of t...</td>\n",
       "      <td>official president barack obama want lawmaker ...</td>\n",
       "      <td>syrian official obama climbed top tree doe kno...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>(CNN) -- Usain Bolt rounded off the world cham...</td>\n",
       "      <td>Usain Bolt wins third gold of world championsh...</td>\n",
       "      <td>usain bolt rounded world championship sunday c...</td>\n",
       "      <td>usain bolt win third gold world championship a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>Kansas City, Missouri (CNN) -- The General Ser...</td>\n",
       "      <td>The employee in agency's Kansas City office is...</td>\n",
       "      <td>kansa city missouri general service administra...</td>\n",
       "      <td>employee agency kansa city office among hundre...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>Los Angeles (CNN) -- A medical doctor in Vanco...</td>\n",
       "      <td>NEW: A Canadian doctor says she was part of a ...</td>\n",
       "      <td>los angeles medical doctor vancouver british c...</td>\n",
       "      <td>canadian doctor part team examining harry burk...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>(CNN) -- Police arrested another teen Thursday...</td>\n",
       "      <td>Another arrest made in gang rape outside Calif...</td>\n",
       "      <td>police arrested another teen thursday sixth su...</td>\n",
       "      <td>another arrest made gang rape outside californ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0                                               text  \\\n",
       "0           0  It's official: U.S. President Barack Obama wan...   \n",
       "1           1  (CNN) -- Usain Bolt rounded off the world cham...   \n",
       "2           2  Kansas City, Missouri (CNN) -- The General Ser...   \n",
       "3           3  Los Angeles (CNN) -- A medical doctor in Vanco...   \n",
       "4           4  (CNN) -- Police arrested another teen Thursday...   \n",
       "\n",
       "                                                   y  \\\n",
       "0  Syrian official: Obama climbed to the top of t...   \n",
       "1  Usain Bolt wins third gold of world championsh...   \n",
       "2  The employee in agency's Kansas City office is...   \n",
       "3  NEW: A Canadian doctor says she was part of a ...   \n",
       "4  Another arrest made in gang rape outside Calif...   \n",
       "\n",
       "                                          text_clean  \\\n",
       "0  official president barack obama want lawmaker ...   \n",
       "1  usain bolt rounded world championship sunday c...   \n",
       "2  kansa city missouri general service administra...   \n",
       "3  los angeles medical doctor vancouver british c...   \n",
       "4  police arrested another teen thursday sixth su...   \n",
       "\n",
       "                                             y_clean  \n",
       "0  syrian official obama climbed top tree doe kno...  \n",
       "1  usain bolt win third gold world championship a...  \n",
       "2  employee agency kansa city office among hundre...  \n",
       "3  canadian doctor part team examining harry burk...  \n",
       "4  another arrest made gang rape outside californ...  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dtf = add_preprocessed_text(dtf, column=\"text\", \n",
    "                            punkt=True, lower=True, slang=True, lst_stopwords=lst_stopwords, lemm=True)\n",
    "dtf = add_preprocessed_text(dtf, column=\"y\", \n",
    "                            punkt=True, lower=True, slang=True, lst_stopwords=lst_stopwords, lemm=True)\n",
    "dtf.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Full text ---\n",
      "usain bolt rounded world championship sunday claiming third gold moscow anchored jamaica victory men 4x100m relay fastest man world charged clear united state rival justin gatlin jamaican quartet nesta carter kemar baileycole nickel ashmeade bolt 3736 second finished second 3756 second canada taking bronze britain disqualified faulty handover 26yearold bolt collected eight gold medal world championship equaling record held american trio carl lewis michael johnson allyson felix mention small matter six olympic title relay triumph followed individual success 100 200 meter russian capital I proud ill continue work dominate long possible bolt previously expressed intention carry 2016 rio olympics victory never seriously doubt got baton safely hand ashmeade gatlin united state third leg runner rakieem salaam problem gatlin strayed lane struggled get full control baton never able get term bolt earlier jamaica woman underlined dominance sprint event winning 4x100m relay gold anchored shellyann fraserpryce like bolt completing triple quartet recorded championship record 4129 second well clear france crossed line second place 4273 second defending champion united state initially back bronze medal position losing time second handover alexandria anderson english gardner promoted silver france subsequently disqualified illegal handover british quartet initially fourth promoted bronze eluded men team fraserpryce like bolt aged 26 became first woman achieve three gold 100200 relay final action last day championship france teddy tamgho became third man leap 18m triple jump exceeding mark four centimeter take gold germany christina obergfoll finally took gold global level woman javelin five previous silver kenya asbel kiprop easily tactical men 1500m final kiprops compatriot eunice jepkoech sum surprise winner woman 800m bolt final dash golden glory brought eightday championship rousing finale host topped medal table united state criticism poor attendance luzhniki stadium concern pole vault gold medalist yelena isinbayeva made controversial remark support russia law make propagandizing nontraditional sexual relation among minor criminal offense later attempted clarify comment renewed call gay right group boycott 2014 winter game sochi next major sport event russia\n",
      " \n",
      "--- Summary ---\n",
      "usain bolt win third gold world championship anchor jamaica 4x100m relay victory eighth gold championship bolt jamaica double woman 4x100m relay\n"
     ]
    }
   ],
   "source": [
    "# check\n",
    "print(\"--- Full text ---\")\n",
    "print(dtf[\"text_clean\"][i])\n",
    "print(\" \")\n",
    "print(\"--- Summary ---\")\n",
    "print(dtf[\"y_clean\"][i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Word frequency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Find most common words in text\n",
    "dtf_freq = word_freq(corpus=dtf[\"text_clean\"], ngrams=[1], top=30, figsize=(10,7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "43002"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "thres = 5\n",
    "X_top_words = len(dtf_freq[dtf_freq[\"freq\"]>thres])\n",
    "X_top_words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Find most common words in summaries\n",
    "dtf_freq = word_freq(corpus=dtf[\"y_clean\"], ngrams=[1], top=30, figsize=(10,7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10246"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "thres = 5\n",
    "y_top_words = len(dtf_freq[dtf_freq[\"freq\"]>thres])\n",
    "y_top_words"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Length analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                            min         mean          max\n",
      "char_count           103.000000  2260.440850  10038.00000\n",
      "word_count            19.000000   364.190450   1309.00000\n",
      "sentence_count         1.000000     1.000000      1.00000\n",
      "avg_word_length        3.992945     6.204663     35.34507\n",
      "avg_sentence_lenght   19.000000   364.190450   1309.00000\n"
     ]
    }
   ],
   "source": [
    "# Texts\n",
    "X = add_text_length(dtf, \"text_clean\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkoAAADYCAYAAAAUEsz4AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAArF0lEQVR4nO3deXxcdb3/8ddnJplksu9r0yZd6EJLF0tLAWUpQtnRi15FrnKv+vOqV/mpP+/loter17u4IC5XveAVRRQUQVBA1AKlUKAtdKF0S5vuTdssTdpszZ7v74+ZlDQkbVIyOcnM+/l4zGNmzpnlPd+eznzyPd/zPeacQ0RERETeyud1ABEREZGxSoWSiIiIyCBUKImIiIgMQoWSiIiIyCBUKImIiIgMQoWSiIiIyCBUKIlIRJnZo2a20uscZ8vMFpnZV73OISLeUKEkInJ6i4B/9TqEiHhDhZKIvG1mFvQ6g4hIJKhQEoliZnaZmTkzK+qzbLWZdZtZRp9lm83sP8K355nZc2Z2wsyOmdmDZpbf57Gl4df8kJk9YGbHgSfD60rM7GkzazWzfWb2sbPI7DezfzaznWbWbmaVZnZ/v8f8g5lVhNfvMrPP9Vt/v5mt67esN/d1fZY5M7vdzP7TzGrNrMbMfmRmCeH1twH/3eexbjzvRhSR4VOhJBLd1gKdwDsBzCwJeAfQAVwUXpYFnAusMrNcYCWQBNwCfAa4BHjGzAL9XvsuoAl4H/CfZmbAH4DZwEeBzwO3A0uGmfle4GvAb4HrgC+E8xDO+3FCxcsTwPXAI8B3zOyOYb5Pry8ARcCtwLeBT4RzA/wR+E749pLw5VNn+T4iMg7FeR1ARCLHOXfCzNYTKpQeBi4AGoDnwsv+CFwMOOAV4M7wU69yzjUCmFkFsAb4K+DXfV5+jXPu0713zOwaYD5wgXNubXjZemA3UDGUvGY2g1CRdbtz7gd9Vj0cXu8Dvgrc75z7QnjdcjNLB/7ZzL7nnGsbynv1sc85d1v49l/M7CLgvcC3nHO1ZrYPwDm3ZpivKyJRQD1KItHvRcI9SsC7gJeAF/ot2xQujBYBy3uLJIBw0bOPUEHV1x/73V8EVPcWSeHn7gfWDyPrZeHr+wdZP4FQ788j/ZY/DKQBc4bxXr2W97u/Lfw+IiIqlERiwCpgdnhM0jvD91cBC80ssc8ygEKgeoDXqAayBljWVwFQM8BzB1o2mGygpW+h1k/hIO/de79/xqE43u9+B5B4Fq8jIlFIhZJI9Hs5fH0poV1vLwJbgWZgKbCANwulI0DeAK+RD9T3W+b63a8a5LkDLRtMHZBsZmmDrD8yyGv2DjbvzdgG9B9TlTmMHCIigAolkajnnDsGbAE+B3QDG51zjtAuuH8kNFaxt1BaC1xlZqm9zzez84HS8ONP5zUg38wW93nuREKF2FCtCF9/eJD1lcBhQgPI+3o/0Ahs7vO40nCPWa8rh5Gjrw6Afq8lIjFChZJIbFhFaCzSK8657n7LKpxzvbuu7g5f/8XMbjSzDwGPESpAfneG93ga2AQ8YmYfNLP3EhrHNORdb865HcBPCB3F9m9mdoWZ3Wxmvwmv7yE0mPs2M/u2mV0Zntbgk8B/9RnI/XsgBfhp+DW+CPzdUHP0Ux6+vt3Mzjez6Wf5OiIyDqlQEokNvT1GLw6w7GRPkXOultCA6jZCR7j9KPy4dzvnOk73BuFeqhsIDYb+GfBd4IfA6mFm/RSh6QFuJVR8fQ840ed9/pfQ4fvvAZ4CPgh8wTn3jT6P2UKoMFpCaBqBS4C/HWaOXqsITRtwO6Eet3vP8nVEZByy0HebiIiIiPSnHiURERGRQWjCSREZNWZ2uu8c12f8lIjImKAeJREZTZ2nuTznYS4RkQGpR0lERtP5p1nXNGopRESGSIO5RURERAahXW8iIiIig1ChJCIiIjIIFUoiIiIig1ChJCIiIjKIiBz1lpOT40pLSyPx0hHV0tJCcnKy1zHGHLXLwNQug1PbDEztMjC1y+DUNgMb6XZZv379Uedc7kDrIlIolZaWsm7duki8dEStXLmSSy+91OsYY47aZWBql8GpbQamdhmY2mVwapuBjXS7mNn+wdZp15uIiIjIIFQoiYiIiAxChZKIiIjIIFQoiYiIiAxChZKIiIjIIHRS3Bj30NoDZ3xM0SjkEBERGYvUoyQiIiIyCBVKIiIiIoPQrjcZtoYTnew52sz+uhPUt3TQ3tXDxgPHSAz4SQnEkZEUT35aIvH+U+vwWxZP9CixiIjI2VGhJEOybl89f95axcryWg43tFKcESQ/LYHUxHjifEZ1Yxud3T20dfbQ0NbJ8ROdZCcHmF6QyqzCNEqykrz+CCIiIsOmQkkG5Zxj8+EGTnQ187M1G7lwcjYfXjKJKbkp+Hx2ymOf2159yv2uHkddczuVx1p5dEMlHZ09HG1q55bFE8lLSxzNjyEiInLWVCjJgOpa2nlkXSVtnd3MnOHj6nMLMYMD9Sc4UH/ijM+P8xn5aYnkpyXyjkmZ1LV0sPVII0vvfoFl5xbw2aXT1MskIiJjngZzy1u8tq+eHz+/m8L0RK4/r4hAnA+zMz/vdLKTA/zdRWXc/b55dPc4rv3BKr70+GaOtXSMTGgREZEIUI+SnOL58mpe3VfPtecVkhGMH9HX7t09l5Uc4Pq5Rbx+8DgXfXMFS2fksbgsC5/PpwHfIiIypqhQEgAcjuVbq9lc2cC1c4pICvgj+n7BeD9LJmczPT+VNXvqWLf/GDe/Y0JE31NERGS4VCgJAC/sqGXL4QaunlNAMD6yRVJfWckBrp5TyM7qJn66ai/OwWeXTiMQp73CIiLiPf0aCbtrm3lp11HePTN/VIukXgZMz0/lpvnFrNlTxw0/fInyqsZRzyEiItKfepRiXENrB7959QCXnJNLSoK3m0NywM/ckgwqqpt4z49e4Z3TcnjXObn4+o0k1zgmEREZLSqUYlh3j+PBtQeYWZhGcUbQ6zhAqHfpnPxUCtODvLTrKNurGnn/whKykxO8jiYiIjFIhVIUe2jtgdOuX7v3KF09jrklGaMTaBhSE+NYNruArYcb+PHzu1k6M48lk7OxtztPgYiIyDCoUIpRze2dPLOthqvOLWCslh4GzC5KZ0JmEi9V1LL5UAM3L9CRcSIiMno0mDtGPb25iql5KWQnB7yOckYZwXiunVNEQVoiP165mx88W0F7V7fXsUREJAaoUIpB++pa2FXTxLwxuMttMGYwpzidG+YWsWpXLcu+t4qVO2q8jiUiIlFOu95ijMPx5y1HeMfELAL+8VcnpybG8bkrzmHDgeN86fEtTM5N5ivXzWJafqrX0UREJAqNv19KeVt217TQ2NbFlLwUr6OcNTPjHZMy+cZ751CWk8z77lnN5x5+nYNDOFmviIjIcKhHKYY4HMu3VTGvJAPfWB3BPQxxfh9Xzy7kknNyeXrzEa79wSpumFvEZ5ZO47ntZ94tp/mYRETkTFQoxZBd1c20tHcxJWf89ibBmyfX7SszKXSi3U2VDbzrW8/zjkmZXDYjj+SANnERETl72vUWIxyOZ7ZVM7ckg2idiigY72dxWRbvXTCB4yc6uHv5Tp7bXq0j5ERE5KypUIoRB+pO0NTeOe57k4YiOeDnwik5XD+3kH11J7h7+U7W76/HOed1NBERGWe0XyJGvFRxlFmFaVHbmzSQtMR4LpueS01jGy9VHOWV3XXcOK+IiVnJXkcTEZFxQj1KMeB4awe7a5tj9hD6vLRErptbxDn5qfxy9X4eXX+QlvZOr2OJiMg4oB6lGLB6dx3T8lPH5bxJI8WAaXkpTMpOYuOB43z32QqCAT8fXDQJfzQcAigiIhERu7+cMaKju5t1+44xszA2e5P6C/h9LC7LYtm5BTy09iA3/PAl1u8/5nUsEREZo1QoRbnXDzSQn5ZAWmK811HGlKzkAF++diaXnJPLJ365jn94aAOHjrd6HUtERMYYFUpRbu3eOmYUpHkdY0wyM945LZdv3zyXQJyPq7/3Il9+fDPVjW1eRxMRkTFChVIUO9LQSlNbF8UZQa+jjGmJ8X7e944SvnXzXI63drL0Oy/w+YdfZ+vhBq+jiYiIxzSYO4q9tq+eafkpMTUlwHAMNMN3QVoiN84rYkdVE39972rKclL4qwXFXDe3iJyUBA9SioiIl1QoRam2zm42HWzghrmFXkcZd4LxfuaVZDBnQjp5qQmsKK/hruU7mFGQxlXn5vPuWQWU5mguJhGRWKBCKUot31ZNdkqAVA3iPmt+M+aVZDKvJJOOrh62HGpg7d567nlhDykJcfz99HZW767j/NJM4mJ46gURkWimQilK/XrtAablRf/pSkZLIM7HgkmZLJiUiXOOfXUn6D68lX/5wxZqm9q5YmYef7VgAhdMzsaneZlERKKGCqUodOh4K1sPN/DeBRO8jhKVzIyynGTqjwf4+o2zOdrczq/W7Oezv9lIe1cPi0qzWFSW9ZbevFsWT/QosYiInC0VSlHoidcPsagsizj1bIyKnJQE5hSnM6c4nbqWDsqPNHL3MzuZWZDKZTPzyE1J9DqiiIicJQ2siEKPbzzEkik5XseISdnJAS6amsP7F5YQ5/dxz8o9/PrVA9Q1t3sdTUREzoJ6lKLMzuom6ls6mFGQyhHNNP22DTSFQK8p3Z1sHGR9QpyPeSUZzCpKY+vhBn68chcNrZ189oppmiVdRGQcUY9SlPn9xkMsmZyNT5MnjQkBv4/5JZm8Z/4E9h5t4bJvr+SJTYdxznkdTUREhkA9SlHEOcfvNx7iM0uneR1F+kkK+CnLSSY54Oerf9jKD1dU8N75E0gLasC3iMhYph6lKLLhwDHi/MakrCSvo8gg8sIzf6ckxPGDFRU6TYqIyBinHqUo8vuNh7lgcjam3W5jmt9nLJiYSXFGkCc3HWZndRPXzy0izqe/W0RExhoVSlGip8fxpy1HuPPqmV5HkSHKT0vkpvnFrKqo5Scv7OFvlkziobUHzvg87Z4TERk9+hM2Sqw/cIzUxHgKM4JeR5FhCPh9XD4jn/z0BH64YheVx054HUlERPpQoRQl/vjGYc4vzfQ6hpwFA+aXZLJ4chY/f2UfO6ubvI4kIiJhKpSiQE+P4+nNVSwqzfY6irwNpdnJXDEjj9+uO8jrB495HUdERNAYpXGr71iW/fUtdHU7yqsaKa9q9DCVvF35aYksO7eApzcfoavbsbA0y+tIIiIxTT1KUWDLoQYm5WhKgGiRlRzg6tmFLN9WxYb99V7HERGJaSqUxjmHY3NlA2XZyV5HkRGUHoxn2exC/rSlio0HtBtORMQrKpTGuUPHWvGZkZkc8DqKjLCMYDzLZhfwx81H2FmtXaoiIl5QoTTObTvSyMTsJDTFZHTKTAqwdEYeD79WqakDREQ8oEJpnNt2uJGJOmVJVMtPS+Tiadn84pV91Ld0eB1HRCSmqFAax+pbOmhq7yI/NdHrKBJhk7KSmVuSwf2v7KWxrdPrOCIiMUOF0ji27UgDk7KS0KndYsOswjTyUhP4hwc30N3jvI4jIhITVCiNY1sPNVKi3W4xZfHkbI63dvIff9zmdRQRkZigQmmcOtHRxZGGVop1breY4jfjM5dP409bqvjD64e8jiMiEvWGVCiZ2WNmdq2ZqbAaI8qrmijOCBLn0363WJOSEMftS6fxlT9s1UzsIiIRNtTC58fALUCFmX3DzKZHMJMMwbbDDdrtFsMmZSdz6wUT+fgD62ho1eBuEZFIGdK53pxzzwLPmlk68MHw7YPA/wK/cs7pm3oUdXT1sLu2hfkTM72OIh54bnv1ydupCXF84N7V3LpkEtZnNq1bFk/0IpqISNQZ8q40M8sGbgM+BmwEvg8sAJ6JSDIZ1Gv76skIxhOM93sdRTy2uCybupYOXt511OsoIiJRaahjlB4HVgFJwPXOuRuccw875z4DpEQyoLzVM9uqmZClQdwCfp9x2fRcVu6o5UB9i9dxRESizlB7lP7XOTfLOfdfzrkjAGaWAOCcWxixdDKgFeU1lGTqJLgSkpoYz0VTs3lo7QFOdHR5HUdEJKoMtVD69wGWrR7JIDI0e4+20NLeRXaKToIrb5qUlcyk7GQeWV+JQ5NRioiMlNMWSmZWYGbvAIJmNt/MFoQvlxLaDSejbEV5DfMnZugkuPIWCydlcqylg1c0XklEZMSc6ai3qwgN4J4A3N1neRNwZ4QyyWks31rFRVNzaNQh4dKP32dcOj2XJzcdYdPB48wtyfA6kojIuHfaHiXn3C+cc5cBtznnLutzucE599goZZSw5vYu3qhsYE5xutdRZIxKS4znwinZfPqhDTp5rojICDhtj5KZ3eqc+xVQamaf77/eOXf3AE+TCHmp4ijTC1JJ1LQAchplOcn4fcYdv3uDH92yANNZk0VEztqZBnP3HlqVAqQOcJFRtKK8Rr1JMiQfWjyJ7Uca+c1rB72OIiIyrp22R8k5d2/4+mujE0cG45xj5Y4a/nHZDK+jyDgQiPPx6cum8fWntrFgYibTC/R3jYjI2RjqhJPfMrM0M4s3s+fMrNbMbo10OHlTRU0zZlCUnuh1FBknijOC3LJ4Ip98cL3mVxIROUtDnUfpSudcI3AdsA+YCnwxUqHkrZ4vr2HuhAyNN5Fhede0XCZmJvGV32/1OoqIyLg01EKpdxfdtcAjzrmGCOWRQWh8kpytj1xYypq9dfxufaXXUURExp0zzaPU6ykzKwdagU+aWS7QFrlYse2htQdOud/e1c2GA8eYXpB6ypnjRYYiMd7PP1w2la8/tY25JelMzdN4JRGRoRpSj5Jz7g7gQmChc64TaAFujGQwedOe2hbyUhMJ+IfaAShyqknZybz//BI+8UuNVxIRGY7h/PLOAP7azD4M3AxcGZlI0t+OqiaKM4Nex5Bx7tJzcinODPLlx7d4HUVEZNwY6lFvvwTuAi4Gzg9fFkYwl4Q5HDuqG5mQoUJJ3h4z428vLOO1ffX85tUDZ36CiIgMeYzSQmCWc06nJR9ldc3tdPc4MpMDXkeRceR0Y9kWl2XztSe3Mbs4ndk6QEBE5LSGuuttC1AQySAysJ3VzRRnBtGkADJSMpLiWTI5m0/8cj0NJ3Q+OBGR0xlqoZQDbDOzv5jZE72XSAaTkB1VTRSnJ3kdQ6LM5Nxk5pVk8Jlfb6C7Rx3FIiKDGequt69GMoQMrKunh/11LVxQlu11FIlCH1hUwjf/VM5dy3fwTzo1jojIgIY6PcALhGbkjg/ffg3YEMFcAuyvO0FmUoCEeE0LICMvzufjM5dP47ENlTy9+YjXcURExqShHvX2ceBR4N7womLg9xHKJGE7qpoo0rQAEkFpwXhuX3oOdz62mW2HG72OIyIy5gx119ungUXAWgDnXIWZ5UUslQCws7qJxWVZXseQKNX3yLgFEzP4wE9W8+nLppKaGH9y+S2LJ3oRTURkzBjqPp1251xH7x0ziwM0AjSCmto6aWjtJDc10esoEgMm56YwLT+FX6zeR2d3j9dxRETGjKEWSi+Y2Z1A0MzeDTwCPBm5WFJRE5oWwKd5AWSUzCvJJCkQx8OvHaBHU6aJiABDL5TuAGqBzcAngKeBL0cqlEB5VSPF6RqfJKPHgIun5nC8tZMnNx3GqdNYRGTIR731EBq8/Snn3M3Ouf/VLN2R0+Mcu2tadH43GXVxPuOKmfnsqmlm5Y4ar+OIiHjutIWShXzVzI4CO4AdZlZrZl8ZnXix6fDxVoIBPykJQx1rLzJyAn4fV55bwJo99Ty4dr/XcUREPHWmHqXPARcB5zvnspxzWcBi4CIz+1zE08WoHVVNFOskuOKh5ICfZecW8N1ndvLo+oNexxER8cyZCqW/AT7onNvbu8A5twe4FfhwJIPFsh3VKpTEe+nBeO5YNpP/erqcJzcd9jqOiIgnzlQoxTvnjvZf6JyrBeIHeLy8TQ2tndQ0tlGQrmkBxHvFmUG+eNV0/vWJrTy+odLrOCIio+5MhVLHWa6Ts/TKrqMUpCcSp3kBZIyYlJ3MHctm8O9/3M5vXj3gdRwRkVF1ptHCc81soPMaGKAujwhYUV5DkXa7yRhTkpXEl66dyTf+VE5zexcfe+dkryOJiIyK0xZKzjn/aAURcM7xYkUt75yW63UUEeDU05wAXD4jjx89v4sXdtaybHYBhuk0JyIS1XRa+jGkoqYZIzSIVmQsSkmI49o5heysauKR1yrp6tHpTkQkuqlQGkNWlNcwd0IGGp0kY1livJ9lsws41trB/S/vo7Gt0+tIIiIRo0JpDFmxvZo5EzK8jiFyRvF+H1fMyCcY8PPeH7/CoeOtXkcSEYkIFUpjRFNbJ5sPNXJuUZrXUUSGxAwumJzNksnZ3PTDl9lc2eB1JBGREadCaYx4eVcd0wtSSYzX+HkZPwy4Zk4hf3PBJG69by3Lt1Z5HUlEZETpZGJjxIryGmYXpXsdQ2TYeo+Mu3R6Lp97+HUumZ7LFRkdPLT2zTmXdGSciIxX6lEaA5xzrNxRw7yJGV5HETlruSkJXHdeEWt213O8tZMeHREnIlFAhdIYsLO6GZ9BkU5bIuNcamIc180tpLOrhwfW7Keju9vrSCIib4sKpTFgRXl1aFoA08QAMv4F/D7SgvEY8JMX99Ck6QNEZBxToTQGLN9Wrd1uEnUunpZLQVoi/7NyN3tqm72OIyJyVlQoeay+pYOdVU3MKtRAbokuBiyYmMmcCencfM9qNh445nUkEZFhU6HksefLa5hdnE4gTv8UEp2m56fy0YvKuO3nr73l3HEiImOdfp09tnxbFfNKMryOIRJRCyZl8v+uPIcvPvoGv371wJmfICIyRqhQ8lB7Vzcv76pj/sRMr6OIRNzUvFS+fM1MfvBcBd97ZifOOa8jiYickSac9NDaPfVMyAySHoz3OopIRPXd5XbZ9DweXLufVbtquWleMX5f6O81TUopImORepQ89My2au12k5iTFPBzzZxCapraeWD1ftq7NNeSiIxd6lEaZb2ndXA4nth0iKUz8jXAVWJOvN/HFTPzWb27jntf2M1tF5Z5HUlEZEDqUfLIkeNtAGQmBzxOIuINvxkXTc1hYlYSP3p+F5srG7yOJCLyFiqUPLL5UAOl2cloLm6JZQacNyGDRWVZ3HrfWv74xhGvI4mInEKFkgccjs2VDZTmJHsdRWRMKMtJ5p+WzeDfntrKN/60ne4eHREnImODCiUPVDe20dnTQ05KgtdRRMaMspxk/u2G2by06ygf+dmr1Ld0eB1JRESFkhc2V2q3m8hA0oLx/NOyGWQlB7j6+y/y2r56ryOJSIxToeSB0PikJK9jiIxJcT4fH1w0kQ8vKeX/PLCO7z+7k67uHq9jiUiMUqE0ymqa2mjr7CYvLdHrKCJj2oKJmXz9xtmsKK/hr+5Zzb6jLV5HEpEYpEJplG2ubGCSdruJDEl2SgL/uGwG80rSufFHL/Ozl/ZooLeIjCpNODmKnHNsPHCci6Zmex1FZMw53cSr8T4fV87K53cbDvGH1w/zrZvnMr0gdRTTiUisUo/SKNpU2UC369FuN5GzkB6M585rZrKwNIv337uarz+1jeb2Lq9jiUiUU4/SKHp03UGm5qVot5vIWXq+vAYDrjuvkFUVtfz2tYNcNbuA+SUZmIX+Z+nkuiIyktSjNEo6unp46o0jTM1N8TqKyLgXjPfzrmm5XDo9l1U7a/nvFRXsqmnGofFLIjKy1KM0SlbuqKE4M0hqYrzXUUSiRn5aItfNLWJvbQuPbagkPSmeaXmpLCrL8jqaiEQJFUqj5NH1lSyZko3+4BUZWQZMzk2mNCeZipomPvvrjUzIDPKpy6Zw2fS8k7vkRETOhna9jYLjJzp4eddRLijT0W4ikeIzmJ6fyl3vm8sFk7P5+lPbufw7L3D/y3tpauv0Op6IjFPqURoFD792kIWlWSQnqLlFIs3vMy6amsOFU7LZfqSR5duq+c4zO7lyVj7vW1jCotIsfD71MonI0OiXO8J6ehwPrN7P318y2esoIjGh/3xMswrTKM1OZldtM59+cANdPT184PyJ3DivmJmFqdo1JyKnpUIpwl7YWUsw4GeKjnYT8UxSwM95xenMKU6nvqWDNXvqeOjVAwTifMydkMHcknRyU06d30zTDIgIqFCKuJ+/vJelMzSgVGQsMCA7OUB2chYLS7OoaWxjb10LP3mhjrRgPAsmZjCvJIPkBB2dKiIhKpQiaH9dC5sqG/joxdrtJjLWGKHpBfLTEllcms3hhlYqqpt5dnsNU/NSKM1OYsmUbP2RIxLjVChF0AOr9/OuaTkE4nRwochYZgbFGUGKM4K0d/Wwu7aZOx7bTEKcj//zrsncMK+IhDi/1zFFxAMqlCKkvqWDR9Yd5D/eM8frKCIyDAlxPmYVpjGzMI1Dx1r5/nMV/NuT27hkei7nl2YR73/zDx+NYxKJfiqUIuSnq/awuCyLnJQEr6OIyFkwYEJmkAmZQWqb23n94HFW7qjl0um5LCrLIs6nnmKRWKD/6RHQcKKTX63Zz/Vzi7yOIiIjIDclgXfPzOeKmfm8UdnAXX/Zwbp99XR193gdTUQiTIVSBNz30h4WlmaSm5p45geLyLiRkxLgyln5XHJOLmv31rH0Oy/w1BuH6enRuYlEopUKpRHW0NrJL1bv5/rzir2OIiIRkp+WyLLZhdyyeCLff7aCZd9/kT9vqcI5FUwi0UZjlEbY957dycJJmRSkqzdJJJoZcN6EDOYUp7PxwHG+/ZdyvvvMTj59+VSumV1AnF9/h4pEA/1PHkE7q5t4bMMh3rewxOsoIjJKzIwFkzL5+o2zuWFeEfes3M2ld63kZy/pZLwi0UA9SiPEOccnHljHrMI01u2r9zqOiIyC/ueVA7hwSjZVjW38au1+vvnncuaWZPAv185idnGaBwlF5O1SoTRC/rSlisa2Li6boS9DkVhXkJZIQVoize1d7Kxu4uMPrCM5wc/fn9POrppmpubp3I8i44UKpRFQ19zOV5/YygWTs/HpbAciEpaSEMeCiZl84co8dlQ10Vqznb++dzWpifFcOj2XS6bnsqg0i+QEfRWLjFX63/k2Oef4/G83ccHkbPJSNbmkiLzV8+U1AEzx+7hpfjFHm9vZcqiB5VurqG1uJzslwLtnFjC3JIPZxWlMzU3RYHCRMUKF0tt0/yv7OHy8lY9dXMYLO2u9jiMiY5wRmsAyNyWBeSUZdPc4jja3s6O6idV76jja3E5TWxfZyQHy0xIpSE8gLzWR3NQEPnXpVJ07UmSUqVB6G96oPM73nq3gazecq7/+ROSs+H1Gfloi+WlvTinS1dPDsROdHGvpoLqxnYrqZo6f6OS/n9tFWjCOnJQE8tISyEkJXbJTAqQnxmNmOv+cyAhToXSW9h1t4e/uf42PXlx2yheciMjbFefznex16qu7x9HU1sXx1g4aTnRSXtVEY2s9Da2dtHf1kJEUz5+3VjEpK4mJWUlMyAxSnBmkOCNIVnIAMw2iFBkuFUpnoaapjVvvW8tN84s5vzTL6zgiEiP8PiMjKZ6MpHjIPnVdZ3cPTW1dlOUkU9vczsaDx3h2ezVHm9upaWqnq9tRlJFISVYSHV095KQEyA3v0ktNjMM4tYhSz5RIiAqlYapubONv7lvLhVOyWToj3+s4IiIAxPt9ZCUHaGjtJOD3UZQepCg9eHJ9R7iQamrt5ESPo6KmmXX7jnHsRCc9zpGflkhheuLJHqiu7h4NKRBBhdKwbDvcyN/e/yqXz8jj+vOKvI4jIjJkAb+P7OQA2cmBt6xr7eymvqWDupYONlc2sKK8hp+u2suMwlQWTMxkXkkG8ydmUJwR1O47iTkqlIbozsc28/jrh1hSlkVyII4V4cN9RUTGu2C8n+KMUE9Sr47uHmqb2tl08DgrtldT3dQOwITMYHj8UxKfXTqNrAEKL5FookLpDI42t/Pl329hze463j0zj7xUDdwWkegX8PtOKZ4c0NLeRW1TOzWN7Ww/0sQj6w6SFoxnZmEas4vSmFGYxjn5KUzKTiZeu+0kSqhQGkRTWye/eGUf9720l3dOy+Wm+cXEadptEYlRRmim8ZSEOMpykoFQ8dTY2kldSwcvVhzlyU2HOXaik6a2LtKD8WSlBLigLIuSrCSKM4MUhKdByE1NIDHe7+nnERkqFUr9lFc18viGQ/x23UHmFKfzpWtmUZwZHPDklyIiscyA9GA86cF4JoeLJ3hzGoOGtg521TazqbKBlo4uTrR30dLeTUtHF/F+H8mBOFIS/Xx0Sht3Prb5lLmhclND0yPkpamoEm/FdKHknKO2qZ0NB46zZk8dRa3NfPHlV1kyOZt/uXYWhX3214uIyNCcMo3BABzQ0dlDa2c3rZ3dQBV7j7aw7UgjbZ3dtIWXn2jv5kRHN34/FKQFQ8VTbwGVmkBWSmhwemZSgPSkeDKCAdKCcQTj/Rp0LiNm3BZKrR3dPPnGYdo7u3GAc9DjHN09oUtXj6Ojq4f2rh7aOrtDh8W2dZ6cU6SqoY2uHgdAWmIcMwrTmF4Qx+1Lp2GEjgLZU9t88v1qm9u9+aBjwKQER21r7H7+wahdBqe2GZja5a3i/IbPZ+SlDX6uzI6uUFF1oqObQ8da2VXdTGtnN+1d3bR1hr7je7/PT/c+Ab+POJ8R5/MR5zf8PiPOb8SZ4ff58PsNv0FZTkpouS/8eL8R5ws/xseb1xbK3nttBkbvNSfv9+U4fc7+spo7+OmqPQAniz/fydcPvRfw5rv0KRD7l4qnvLNzJ5eFb+Kco8f1LnOnPHQouXs/e2/WN9ugzzLrlyv8BHvropMWlWYxLT/1jO8fKda3MUbsRc1qgf0j/sJ93yMhOTU+s/Cct/Marru7E9fd1Xu/p7U5zhdM6Trdc2KR2mVgapfBqW0GpnYZ2Mi3ixnm82HmM7Nx3b3UfaIBf1K61zE81dPWUt91/MjefotzgKMj+DaTnHO5A62ISKE0XpnZOufcQq9zjDVql4GpXQanthmY2mVgapfBqW0GNprtouM3RURERAahQklERERkECqUTvUTrwOMUWqXgaldBqe2GZjaZWBql8GpbQY2au2iMUoiIiIig1CPkoiIiMggVCgBZrbMzHaY2S4zu8PrPKPJzErM7Hkz22ZmW83s9vDyLDN7xswqwteZ4eVmZj8It9UbZrbA208QWWbmN7ONZvZU+H6Zma0Nf/6HzSwQXp4Qvr8rvL7U0+ARZmYZZvaomZWb2XYzW6JtBszsc+H/R1vM7Ndmlhir24yZ/czMasxsS59lw95GzOwj4cdXmNlHvPgsI2mQdvl2+P/SG2b2uJll9Fn3z+F22WFmV/VZHlW/WwO1S591XzAzZ2Y54fuju70452L6AviB3cBkIABsAmZ5nWsUP38hsCB8OxXYCcwCvgXcEV5+B/DN8O1rgD8Rmh/sAmCt158hwu3zeeAh4Knw/d8CHwjfvgf4ZPj2p4B7wrc/ADzsdfYIt8svgI+FbweAjFjfZoBiYC8Q7LOt3Bar2wzwLmABsKXPsmFtI0AWsCd8nRm+nen1Z4tAu1wJxIVvf7NPu8wK/yYlAGXh3yp/NP5uDdQu4eUlwF8Izc2Y48X2oh4lWATscs7tcc51AL8BbvQ406hxzh1xzm0I324CthP6wr+R0I8h4eubwrdvBB5wIWuADDMrHN3Uo8PMJgDXAj8N3zfgcuDR8EP6t0tvez0KLA0/PuqYWTqhL7X7AJxzHc6542ibgdDZDoJmFgckAUeI0W3GOfciUN9v8XC3kauAZ5xz9c65Y8AzwLKIh4+ggdrFObfcOdc74eYaYEL49o3Ab5xz7c65vcAuQr9ZUfe7Ncj2AvBd4B85dWLxUd1eVCiFioKDfe5XhpfFnHDX/3xgLZDvnDsSXlUF5Idvx1J7fY/Qf9Ce8P1s4HifL7S+n/1ku4TXN4QfH43KgFrg5+Hdkj81s2RifJtxzh0C7gIOECqQGoD1aJvpa7jbSExsO/38HaHeEojxdjGzG4FDzrlN/VaNaruoUBIAzCwF+B3wf51zjX3XuVCfZkwdHmlm1wE1zrn1XmcZg+IIdZH/j3NuPtBCaDfKSTG6zWQS+ku3DCgCkhnnvR+RFIvbyJmY2ZeALuBBr7N4zcySgDuBr3idRYUSHCK0D7TXhPCymGFm8YSKpAedc4+FF1f37h4JX9eEl8dKe10E3GBm+wh1a18OfJ9QF2/vyaT7fvaT7RJenw7UjWbgUVQJVDrn1obvP0qocIr1beYKYK9zrtY51wk8Rmg70jbzpuFuI7Gy7WBmtwHXAR8KF5EQ2+0yhdAfHZvC38MTgA1mVsAot4sKJXgNmBY+MiVAaFDlEx5nGjXhMRH3Adudc3f3WfUE0HvEwEeAP/RZ/uHwUQcXAA19utKjhnPun51zE5xzpYS2iRXOuQ8BzwM3hx/Wv1162+vm8OOj8q9l51wVcNDMpocXLQW2EePbDKFdbheYWVL4/1Vvu8T8NtPHcLeRvwBXmllmuMfuyvCyqGJmywjt5r/BOXeiz6ongA+Ej5AsA6YBrxIDv1vOuc3OuTznXGn4e7iS0IFHVYz29hKpEezj6UJoBP1OQkcRfMnrPKP82S8m1P39BvB6+HINobESzwEVwLNAVvjxBvwo3FabgYVef4ZRaKNLefOot8mEvqh2AY8ACeHlieH7u8LrJ3udO8JtMg9YF95ufk/oCJOY32aArwHlwBbgl4SOVorJbQb4NaGxWp2EfuQ+ejbbCKExO7vCl7/1+nNFqF12ERpb0/sdfE+fx38p3C47gKv7LI+q362B2qXf+n28edTbqG4vmplbREREZBDa9SYiIiIyCBVKIiIiIoNQoSQiIiIyCBVKIiIiIoNQoSQiIiIyCBVKIiIiIoNQoSQiIiIyCBVKIiIiIoP4/8omxW82CBaoAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_distributions(X, x=\"word_count\", figsize=(10,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_len = 400"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                       min        mean    max\n",
      "char_count           35.00  164.701550  319.0\n",
      "word_count            4.00   27.021300   50.0\n",
      "sentence_count        1.00    1.000000    1.0\n",
      "avg_word_length       4.15    6.115849   10.0\n",
      "avg_sentence_lenght   4.00   27.021300   50.0\n"
     ]
    }
   ],
   "source": [
    "# Summaries\n",
    "y = add_text_length(dtf, \"y_clean\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_distributions(y, x=\"word_count\", max_cat=1, figsize=(10,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_len = 40"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>text</th>\n",
       "      <th>text_clean</th>\n",
       "      <th>y</th>\n",
       "      <th>y_clean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>It's official: U.S. President Barack Obama wan...</td>\n",
       "      <td>official president barack obama want lawmaker ...</td>\n",
       "      <td>Syrian official: Obama climbed to the top of t...</td>\n",
       "      <td>syrian official obama climbed top tree doe kno...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(CNN) -- Usain Bolt rounded off the world cham...</td>\n",
       "      <td>usain bolt rounded world championship sunday c...</td>\n",
       "      <td>Usain Bolt wins third gold of world championsh...</td>\n",
       "      <td>usain bolt win third gold world championship a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Kansas City, Missouri (CNN) -- The General Ser...</td>\n",
       "      <td>kansa city missouri general service administra...</td>\n",
       "      <td>The employee in agency's Kansas City office is...</td>\n",
       "      <td>employee agency kansa city office among hundre...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Los Angeles (CNN) -- A medical doctor in Vanco...</td>\n",
       "      <td>los angeles medical doctor vancouver british c...</td>\n",
       "      <td>NEW: A Canadian doctor says she was part of a ...</td>\n",
       "      <td>canadian doctor part team examining harry burk...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(CNN) -- Police arrested another teen Thursday...</td>\n",
       "      <td>police arrested another teen thursday sixth su...</td>\n",
       "      <td>Another arrest made in gang rape outside Calif...</td>\n",
       "      <td>another arrest made gang rape outside californ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                text  \\\n",
       "0  It's official: U.S. President Barack Obama wan...   \n",
       "1  (CNN) -- Usain Bolt rounded off the world cham...   \n",
       "2  Kansas City, Missouri (CNN) -- The General Ser...   \n",
       "3  Los Angeles (CNN) -- A medical doctor in Vanco...   \n",
       "4  (CNN) -- Police arrested another teen Thursday...   \n",
       "\n",
       "                                          text_clean  \\\n",
       "0  official president barack obama want lawmaker ...   \n",
       "1  usain bolt rounded world championship sunday c...   \n",
       "2  kansa city missouri general service administra...   \n",
       "3  los angeles medical doctor vancouver british c...   \n",
       "4  police arrested another teen thursday sixth su...   \n",
       "\n",
       "                                                   y  \\\n",
       "0  Syrian official: Obama climbed to the top of t...   \n",
       "1  Usain Bolt wins third gold of world championsh...   \n",
       "2  The employee in agency's Kansas City office is...   \n",
       "3  NEW: A Canadian doctor says she was part of a ...   \n",
       "4  Another arrest made in gang rape outside Calif...   \n",
       "\n",
       "                                             y_clean  \n",
       "0  syrian official obama climbed top tree doe kno...  \n",
       "1  usain bolt win third gold world championship a...  \n",
       "2  employee agency kansa city office among hundre...  \n",
       "3  canadian doctor part team examining harry burk...  \n",
       "4  another arrest made gang rape outside californ...  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dtf = dtf[[\"text\",\"text_clean\",\"y\",\"y_clean\"]]\n",
    "dtf.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2 - Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>text</th>\n",
       "      <th>text_clean</th>\n",
       "      <th>y</th>\n",
       "      <th>y_clean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>It's official: U.S. President Barack Obama wan...</td>\n",
       "      <td>official president barack obama want lawmaker ...</td>\n",
       "      <td>Syrian official: Obama climbed to the top of t...</td>\n",
       "      <td>syrian official obama climbed top tree doe kno...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(CNN) -- Usain Bolt rounded off the world cham...</td>\n",
       "      <td>usain bolt rounded world championship sunday c...</td>\n",
       "      <td>Usain Bolt wins third gold of world championsh...</td>\n",
       "      <td>usain bolt win third gold world championship a...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                text  \\\n",
       "0  It's official: U.S. President Barack Obama wan...   \n",
       "1  (CNN) -- Usain Bolt rounded off the world cham...   \n",
       "\n",
       "                                          text_clean  \\\n",
       "0  official president barack obama want lawmaker ...   \n",
       "1  usain bolt rounded world championship sunday c...   \n",
       "\n",
       "                                                   y  \\\n",
       "0  Syrian official: Obama climbed to the top of t...   \n",
       "1  Usain Bolt wins third gold of world championsh...   \n",
       "\n",
       "                                             y_clean  \n",
       "0  syrian official obama climbed top tree doe kno...  \n",
       "1  usain bolt win third gold world championship a...  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Partitioning\n",
    "dtf_train = dtf.iloc[i+1:]\n",
    "dtf_test = dtf.iloc[:i+1]\n",
    "dtf_test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3 - Baseline (Extractive: TextRank)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test (no Train)\n",
    "predicted = textrank(corpus=dtf_test[\"text\"], ratio=y_len/X_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rouge1: 0.18 | rouge2: 0.04 | rougeL: 0.04 --> avg rouge: 0.14\n"
     ]
    }
   ],
   "source": [
    "# Evaluate\n",
    "evaluate_summary(dtf_test[\"y\"][i], predicted[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Real Summary</strong><br>Usain <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> wins third <span style=\"background-color:rgba(255,215,0,0.3);\">gold</span> of world championship . Anchors Jamaica to <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m</span> <span style=\"background-color:rgba(255,215,0,0.3);\">relay</span> victory . Eighth <span style=\"background-color:rgba(255,215,0,0.3);\">gold</span> at <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> championships for <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> . Jamaica double up <span style=\"background-color:rgba(255,215,0,0.3);\">in</span> women's <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m</span> <span style=\"background-color:rgba(255,215,0,0.3);\">relay</span> .<br><br><strong>Predicted Summary</strong><br>Earlier, Jamaica's women underlined their dominance <span style=\"background-color:rgba(255,215,0,0.3);\">in</span> <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> sprint events by winning <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m</span> <span style=\"background-color:rgba(255,215,0,0.3);\">relay</span> <span style=\"background-color:rgba(255,215,0,0.3);\">gold,</span> anchored by Shelly-Ann Fraser-Pryce, who like <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> was completing a triple."
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compare y_test and predicted\n",
    "match = display_string_matching(dtf_test[\"y\"][i], predicted[i], both=True, sentences=False, \n",
    "                                titles=[\"Real Summary\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Full Text</strong><br>(CNN) -- Usain Bolt rounded off the world championships Sunday by claiming his third gold in Moscow as he anchored Jamaica to victory in the men's 4x100m relay. The fastest man in the world charged clear of United States rival Justin Gatlin as the Jamaican quartet of Nesta Carter, Kemar Bailey-Cole, Nickel Ashmeade and Bolt won in 37.36 seconds. The U.S finished second in 37.56 seconds with Canada taking the bronze after Britain were disqualified for a faulty handover. The 26-year-old Bolt has now collected eight gold medals at world championships, equaling the record held by American trio Carl Lewis, Michael Johnson and Allyson Felix, not to mention the small matter of six Olympic titles. The relay triumph followed individual successes in the 100 and 200 meters in the Russian capital. \"I'm proud of myself and I'll continue to work to dominate for as long as possible,\" Bolt said, having previously expressed his intention to carry on until the 2016 Rio Olympics. Victory was never seriously in doubt once he got the baton safely in hand from Ashmeade, while Gatlin and the United States third leg runner Rakieem Salaam had problems. Gatlin strayed out of his lane as he struggled to get full control of their baton and was never able to get on terms with Bolt. <span style=\"background-color:rgba(255,215,0,0.3);\">Earlier, Jamaica's women underlined their dominance in the sprint events by winning the 4x100m relay gold, anchored by Shelly-Ann Fraser-Pryce, who like Bolt was completing a triple.</span>  Their quartet recorded a championship record of 41.29 seconds, well clear of France, who crossed the line in second place in 42.73 seconds. Defending champions, the United States, were initially back in the bronze medal position after losing time on the second handover between Alexandria Anderson and English Gardner, but promoted to silver when France were subsequently disqualified for an illegal handover. The British quartet, who were initially fourth, were promoted to the bronze which eluded their men's team. Fraser-Pryce, like Bolt aged 26, became the first woman to achieve three golds in the 100-200 and the relay. In other final action on the last day of the championships, France's Teddy Tamgho became the third man to leap over 18m in the triple jump, exceeding the mark by four centimeters to take gold. Germany's Christina Obergfoll finally took gold at global level in the women's javelin after five previous silvers, while Kenya's Asbel Kiprop easily won a tactical men's 1500m final. Kiprop's compatriot Eunice Jepkoech Sum was a surprise winner of the women's 800m. Bolt's final dash for golden glory brought the eight-day championship to a rousing finale, but while the hosts topped the medal table from the United States there was criticism of the poor attendances in the Luzhniki Stadium. There was further concern when their pole vault gold medalist Yelena Isinbayeva made controversial remarks in support of Russia's new laws, which make \"the propagandizing of non-traditional sexual relations among minors\" a criminal offense. She later attempted to clarify her comments, but there were renewed calls by gay rights groups for a boycott of the 2014 Winter Games in Sochi, the next major sports event in Russia.<br><br><strong>Predicted Summary</strong><br><span style=\"background-color:rgba(255,215,0,0.3);\">Earlier, Jamaica's women underlined their dominance in the sprint events by winning the 4x100m relay gold, anchored by Shelly-Ann Fraser-Pryce, who like Bolt was completing a triple.</span>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Explainability\n",
    "match = display_string_matching(dtf_test[\"text\"][i], predicted[i], both=True, sentences=True, \n",
    "                                titles=[\"Full Text\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4 - Model (Abstractive: Seq2Seq)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Features Engineering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- tokenization ---\n",
      "43004 words\n",
      "--- padding to sequence ---\n",
      "19998 sequences of length 400\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create X_train for seq2seq (sequences of tokens)\n",
    "dic_seq = text2seq(corpus=dtf_train[\"text_clean\"], top=X_top_words, maxlen=X_len)\n",
    "\n",
    "X_train, X_tokenizer, X_dic_vocabulary = dic_seq[\"X\"], dic_seq[\"tokenizer\"], dic_seq[\"dic_vocabulary\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'<PAD>': 0, 'one': 1, 'people': 2, 'year': 3, 'would': 4, 'time': 5}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict(list(X_dic_vocabulary.items())[0:6])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- tokenization ---\n",
      "--- padding to sequence ---\n",
      "2 sequences of length 400\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Preprocess X_test with the same tokenizer\n",
    "X_test = text2seq(corpus=dtf_test[\"text_clean\"], fitted_tokenizer=X_tokenizer, maxlen=X_train.shape[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'<START> usain bolt win third gold world championship anchor jamaica 4x100m relay victory eighth gold championship bolt jamaica double woman 4x100m relay <END>'"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Add START and END tokens to the summaries (y)\n",
    "special_tokens = (\"<START>\", \"<END>\")\n",
    "dtf_train[\"y_clean\"] = dtf_train['y_clean'].apply(lambda x: special_tokens[0]+' '+x+' '+special_tokens[1])\n",
    "dtf_test[\"y_clean\"] = dtf_test['y_clean'].apply(lambda x: special_tokens[0]+' '+x+' '+special_tokens[1])\n",
    "dtf_test[\"y_clean\"][i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- tokenization ---\n",
      "10248 words\n",
      "--- padding to sequence ---\n",
      "19998 sequences of length 40\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Tokenize the summaries (y_train)\n",
    "dic_seq = text2seq(corpus=dtf_train[\"y_clean\"], top=y_top_words, maxlen=y_len)\n",
    "\n",
    "y_train, y_tokenizer, y_dic_vocabulary = dic_seq[\"X\"], dic_seq[\"tokenizer\"], dic_seq[\"dic_vocabulary\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'<PAD>': 0, '<START>': 1, '<END>': 2, 'year': 3, 'people': 4, 'one': 5}"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict(list(y_dic_vocabulary.items())[0:6])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- tokenization ---\n",
      "--- padding to sequence ---\n",
      "2 sequences of length 40\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Preprocess y_test with the same tokenizer\n",
    "y_test = text2seq(corpus=dtf_test[\"y_clean\"], fitted_tokenizer=y_tokenizer, maxlen=y_train.shape[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Create Embedding Weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load pre-trained Word2Vec\n",
    "nlp = gensim_api.load(\"glove-wiki-gigaword-300\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Or train Word2Vec from scratch\n",
    "#corpus =  dtf_train[\"text_clean\"].append(dtf_train[\"y_clean\"])\n",
    "#lst_corpus, nlp = fit_w2v(corpus=corpus, min_count=1, size=300, window=y_avg_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(300,)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check nlp model\n",
    "word = \"home\"\n",
    "nlp[word].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "vocabulary mapped to 43005 vectors of size 300\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(43005, 300)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_embeddings = vocabulary_embeddings(X_dic_vocabulary, nlp)\n",
    "X_embeddings.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "vocabulary mapped to 10249 vectors of size 300\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(10249, 300)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_embeddings = vocabulary_embeddings(y_dic_vocabulary, nlp)\n",
    "y_embeddings.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Model Design"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Seq2Seq\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "x_in (InputLayer)               [(None, 400)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "y_in (InputLayer)               [(None, None)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "x_emb (Embedding)               (None, 400, 300)     12901200    x_in[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "y_emb (Embedding)               (None, None, 300)    3074400     y_in[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "x_lstm (LSTM)                   [(None, 400, 250), ( 551000      x_emb[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "y_lstm (LSTM)                   [(None, None, 250),  551000      y_emb[0][0]                      \n",
      "                                                                 x_lstm[0][1]                     \n",
      "                                                                 x_lstm[0][2]                     \n",
      "__________________________________________________________________________________________________\n",
      "dense (TimeDistributed)         (None, None, 10248)  2572248     y_lstm[0][0]                     \n",
      "==================================================================================================\n",
      "Total params: 19,649,848\n",
      "Trainable params: 19,649,848\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Basic Seq2Seq\n",
    "lstm_units = 250\n",
    "embeddings_size = 300\n",
    "\n",
    "##------------ ENCODER (embeddings + lstm) -----------------------------##\n",
    "x_in = layers.Input(name=\"x_in\", shape=(X_train.shape[1],))\n",
    "### embedding\n",
    "layer_x_emb = layers.Embedding(name=\"x_emb\", input_dim=len(X_dic_vocabulary), output_dim=embeddings_size, \n",
    "                               trainable=True)\n",
    "x_emb = layer_x_emb(x_in)\n",
    "### lstm \n",
    "layer_x_lstm = layers.LSTM(name=\"x_lstm\", units=lstm_units, dropout=0.4,  \n",
    "                           return_sequences=True, return_state=True)\n",
    "x_out, state_h, state_c = layer_x_lstm(x_emb)\n",
    "\n",
    "##------------ DECODER (embeddings + lstm + dense) ---------------------##\n",
    "y_in = layers.Input(name=\"y_in\", shape=(None,))\n",
    "### embedding\n",
    "layer_y_emb = layers.Embedding(name=\"y_emb\", input_dim=len(y_dic_vocabulary), output_dim=embeddings_size, \n",
    "                               trainable=True)\n",
    "y_emb = layer_y_emb(y_in)\n",
    "### lstm \n",
    "layer_y_lstm = layers.LSTM(name=\"y_lstm\", units=lstm_units, dropout=0.4,\n",
    "                           return_sequences=True, return_state=True)\n",
    "y_out, _, _ = layer_y_lstm(y_emb, initial_state=[state_h, state_c])\n",
    "### final dense layers\n",
    "layer_dense = layers.TimeDistributed(name=\"dense\", \n",
    "                                     layer=layers.Dense(units=len(y_dic_vocabulary), activation='softmax'))\n",
    "y_out = layer_dense(y_out)\n",
    "\n",
    "##---------------------------- COMPILE --------------------------------##\n",
    "model = models.Model(inputs=[x_in, y_in], outputs=y_out, name=\"Seq2Seq\")\n",
    "model.compile(optimizer='rmsprop', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Seq2Seq\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "x_in (InputLayer)               [(None, 400)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "x_emb (Embedding)               (None, 400, 300)     12901500    x_in[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "x_lstm_1 (Bidirectional)        [(None, 400, 500), ( 1102000     x_emb[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "x_lstm_2 (Bidirectional)        [(None, 400, 500), ( 1502000     x_lstm_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "y_in (InputLayer)               [(None, None)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "x_lstm_3 (Bidirectional)        [(None, 400, 500), ( 1502000     x_lstm_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "y_emb (Embedding)               (None, None, 300)    3074700     y_in[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "concatenate (Concatenate)       (None, 500)          0           x_lstm_3[0][1]                   \n",
      "                                                                 x_lstm_3[0][3]                   \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 500)          0           x_lstm_3[0][2]                   \n",
      "                                                                 x_lstm_3[0][4]                   \n",
      "__________________________________________________________________________________________________\n",
      "y_lstm (LSTM)                   [(None, None, 500),  1602000     y_emb[0][0]                      \n",
      "                                                                 concatenate[0][0]                \n",
      "                                                                 concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dense (TimeDistributed)         (None, None, 10248)  5134248     y_lstm[0][0]                     \n",
      "==================================================================================================\n",
      "Total params: 26,818,448\n",
      "Trainable params: 10,842,248\n",
      "Non-trainable params: 15,976,200\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Advanced Seq2Seq\n",
    "lstm_units = 250\n",
    "\n",
    "##------------ ENCODER (pre-trained embeddings + 3 bi-lstm) ---------------##\n",
    "x_in = layers.Input(name=\"x_in\", shape=(X_train.shape[1],))\n",
    "### embedding\n",
    "layer_x_emb = layers.Embedding(name=\"x_emb\", input_dim=X_embeddings.shape[0], output_dim=X_embeddings.shape[1], \n",
    "                               weights=[X_embeddings], trainable=False)\n",
    "x_emb = layer_x_emb(x_in)\n",
    "### bi-lstm 1\n",
    "layer_x_bilstm = layers.Bidirectional(layers.LSTM(units=lstm_units, dropout=0.4, recurrent_dropout=0.4,\n",
    "                                                  return_sequences=True, return_state=True), \n",
    "                                      name=\"x_lstm_1\")\n",
    "x_out, _, _, _, _ = layer_x_bilstm(x_emb)\n",
    "### bi-lstm 2\n",
    "layer_x_bilstm = layers.Bidirectional(layers.LSTM(units=lstm_units, dropout=0.4, recurrent_dropout=0.4,\n",
    "                                                  return_sequences=True, return_state=True),\n",
    "                                      name=\"x_lstm_2\")\n",
    "x_out, _, _, _, _ = layer_x_bilstm(x_out)\n",
    "### bi-lstm 3 (here final states are collected)\n",
    "layer_x_bilstm = layers.Bidirectional(layers.LSTM(units=lstm_units, dropout=0.4, recurrent_dropout=0.4, \n",
    "                                                  return_sequences=True, return_state=True),\n",
    "                                      name=\"x_lstm_3\")\n",
    "x_out, forward_h, forward_c, backward_h, backward_c = layer_x_bilstm(x_out)\n",
    "state_h = layers.Concatenate()([forward_h, backward_h])\n",
    "state_c = layers.Concatenate()([forward_c, backward_c])\n",
    "\n",
    "##------------ DECODER (pre-trained embeddings + lstm + dense) ------------##\n",
    "y_in = layers.Input(name=\"y_in\", shape=(None,))\n",
    "### embedding\n",
    "layer_y_emb = layers.Embedding(name=\"y_emb\", input_dim=y_embeddings.shape[0], output_dim=y_embeddings.shape[1], \n",
    "                               weights=[y_embeddings], trainable=False)\n",
    "y_emb = layer_y_emb(y_in)\n",
    "### lstm\n",
    "layer_y_lstm = layers.LSTM(name=\"y_lstm\", units=lstm_units*2, dropout=0.2, recurrent_dropout=0.2,\n",
    "                           return_sequences=True, return_state=True)\n",
    "y_out, _, _ = layer_y_lstm(y_emb, initial_state=[state_h, state_c])\n",
    "### final dense layers\n",
    "layer_dense = layers.TimeDistributed(name=\"dense\", \n",
    "                                     layer=layers.Dense(units=len(y_dic_vocabulary), activation='softmax'))\n",
    "y_out = layer_dense(y_out)\n",
    "\n",
    "##---------------------------- COMPILE ------------------------------------##\n",
    "model = models.Model(inputs=[x_in, y_in], outputs=y_out, name=\"Seq2Seq\")\n",
    "model.compile(optimizer='rmsprop', loss='sparse_categorical_crossentropy')\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "219/219 [==============================] - 355s 2s/step - loss: 5.4287 - accuracy: 0.3696 - val_loss: 5.2587 - val_accuracy: 0.3759\n",
      "Epoch 2/100\n",
      "219/219 [==============================] - 355s 2s/step - loss: 5.1787 - accuracy: 0.3801 - val_loss: 5.1966 - val_accuracy: 0.3771\n",
      "Epoch 3/100\n",
      "219/219 [==============================] - 357s 2s/step - loss: 5.0799 - accuracy: 0.3826 - val_loss: 5.1201 - val_accuracy: 0.3802\n",
      "Epoch 4/100\n",
      "219/219 [==============================] - 809s 4s/step - loss: 4.9903 - accuracy: 0.3858 - val_loss: 5.0544 - val_accuracy: 0.3821\n",
      "Epoch 5/100\n",
      "219/219 [==============================] - 358s 2s/step - loss: 4.8997 - accuracy: 0.3883 - val_loss: 4.9996 - val_accuracy: 0.3839\n",
      "Epoch 6/100\n",
      "219/219 [==============================] - 356s 2s/step - loss: 4.8151 - accuracy: 0.3907 - val_loss: 4.9544 - val_accuracy: 0.3858\n",
      "Epoch 7/100\n",
      "219/219 [==============================] - 355s 2s/step - loss: 4.7357 - accuracy: 0.3931 - val_loss: 4.9188 - val_accuracy: 0.3876\n",
      "Epoch 8/100\n",
      "219/219 [==============================] - 359s 2s/step - loss: 4.6615 - accuracy: 0.3962 - val_loss: 4.8885 - val_accuracy: 0.3899\n",
      "Epoch 9/100\n",
      "219/219 [==============================] - 358s 2s/step - loss: 4.5905 - accuracy: 0.3987 - val_loss: 4.8642 - val_accuracy: 0.3915\n",
      "Epoch 10/100\n",
      "219/219 [==============================] - 356s 2s/step - loss: 4.5218 - accuracy: 0.4016 - val_loss: 4.8409 - val_accuracy: 0.3929\n",
      "Epoch 11/100\n",
      "219/219 [==============================] - 355s 2s/step - loss: 4.4556 - accuracy: 0.4044 - val_loss: 4.8268 - val_accuracy: 0.3939\n",
      "Epoch 12/100\n",
      "219/219 [==============================] - 356s 2s/step - loss: 4.3915 - accuracy: 0.4069 - val_loss: 4.8139 - val_accuracy: 0.3949\n",
      "Epoch 13/100\n",
      "219/219 [==============================] - 363s 2s/step - loss: 4.3298 - accuracy: 0.4095 - val_loss: 4.8008 - val_accuracy: 0.3961\n",
      "Epoch 14/100\n",
      "219/219 [==============================] - 409s 2s/step - loss: 4.2719 - accuracy: 0.4120 - val_loss: 4.7971 - val_accuracy: 0.3971\n",
      "Epoch 15/100\n",
      "219/219 [==============================] - 416s 2s/step - loss: 4.2165 - accuracy: 0.4143 - val_loss: 4.7947 - val_accuracy: 0.3971\n",
      "Epoch 16/100\n",
      "219/219 [==============================] - 422s 2s/step - loss: 4.1637 - accuracy: 0.4170 - val_loss: 4.7921 - val_accuracy: 0.3977\n",
      "Epoch 17/100\n",
      "219/219 [==============================] - 428s 2s/step - loss: 4.1127 - accuracy: 0.4190 - val_loss: 4.7951 - val_accuracy: 0.3981\n",
      "Epoch 18/100\n",
      "219/219 [==============================] - 430s 2s/step - loss: 4.0624 - accuracy: 0.4215 - val_loss: 4.7964 - val_accuracy: 0.3978\n",
      "Epoch 00018: early stopping\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This takes a while\n",
    "model = fit_seq2seq(X_train, y_train, model, build_encoder_decoder=False, \n",
    "                    epochs=100, batch_size=64, verbose=1)\n",
    "#model, encoder_model, decoder_model = fit_seq2seq(X_train, y_train, X_embeddings, y_embeddings, model, \n",
    "#                                                  build_encoder_decoder=True, epochs=100, batch_size=64)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Prediction_Encoder\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "x_in (InputLayer)            [(None, 400)]             0         \n",
      "_________________________________________________________________\n",
      "x_emb (Embedding)            (None, 400, 300)          12901200  \n",
      "_________________________________________________________________\n",
      "x_lstm (LSTM)                [(None, 400, 250), (None, 551000    \n",
      "=================================================================\n",
      "Total params: 13,452,200\n",
      "Trainable params: 13,452,200\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Prediction Encoder\n",
    "encoder_model = models.Model(inputs=x_in, outputs=[x_out, state_h, state_c], name=\"Prediction_Encoder\")\n",
    "encoder_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Prediction_Decoder\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "y_in (InputLayer)               [(None, None)]       0                                            \n",
      "__________________________________________________________________________________________________\n",
      "y_emb (Embedding)               (None, None, 300)    3074400     y_in[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "input_2 (InputLayer)            [(None, 250)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_3 (InputLayer)            [(None, 250)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "y_lstm (LSTM)                   [(None, None, 250),  551000      y_emb[1][0]                      \n",
      "                                                                 input_2[0][0]                    \n",
      "                                                                 input_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "input_1 (InputLayer)            [(None, 400, 250)]   0                                            \n",
      "__________________________________________________________________________________________________\n",
      "dense (TimeDistributed)         (None, None, 10248)  2572248     y_lstm[1][0]                     \n",
      "==================================================================================================\n",
      "Total params: 6,197,648\n",
      "Trainable params: 6,197,648\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Prediction Decoder\n",
    "## double the lstm units if you used bidirectional lstm\n",
    "lstm_units = lstm_units*2 if any(\"Bidirectional\" in str(layer) for layer in model.layers) else lstm_units\n",
    "\n",
    "## states of the previous time step\n",
    "x_out2 = layers.Input(shape=(X_train.shape[1], lstm_units))\n",
    "state_h, state_c = layers.Input(shape=(lstm_units,)), layers.Input(shape=(lstm_units,))\n",
    "\n",
    "## decoder embeddings\n",
    "y_emb2 = layer_y_emb(y_in)\n",
    "\n",
    "## lstm to predict the next word\n",
    "y_out2, new_state_h, new_state_c = layer_y_lstm(y_emb2, initial_state=[state_h, state_c])\n",
    "\n",
    "## softmax to generate probability distribution over the target vocabulary\n",
    "probs = layer_dense(y_out2)\n",
    "\n",
    "## compile\n",
    "decoder_model = models.Model(inputs=[y_in, x_out2, state_h, state_c], \n",
    "                             outputs=[probs, new_state_h, new_state_c],\n",
    "                             name=\"Prediction_Decoder\")\n",
    "decoder_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted = predict_seq2seq(X_test, encoder_model, decoder_model, y_tokenizer, special_tokens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rouge1: 0.04 | rouge2: 0.0 | rougeL: 0.0 --> avg rouge: 0.03\n"
     ]
    }
   ],
   "source": [
    "# Evaluate\n",
    "evaluate_summary(dtf_test[\"y_clean\"][i], predicted[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Real Summary</strong><br><START> usain bolt <span style=\"background-color:rgba(255,215,0,0.3);\">win</span> third gold world championship anchor jamaica 4x100m relay victory eighth gold championship bolt jamaica double woman 4x100m relay <END><br><br><strong>Predicted Summary</strong><br>manchester united state liverpool beat france 21 draw cup 21 draw score goal goal goal <span style=\"background-color:rgba(255,215,0,0.3);\">win</span> first goal <span style=\"background-color:rgba(255,215,0,0.3);\">win</span> first time since 22 minute arsenal 21 <span style=\"background-color:rgba(255,215,0,0.3);\">win</span>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compare y_test and predicted\n",
    "match = display_string_matching(dtf_test[\"y_clean\"][i], predicted[i], both=True, sentences=False, \n",
    "                                titles=[\"Real Summary\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Full Text</strong><br>usain bolt rounded world championship sunday claiming third gold moscow anchored jamaica victory men 4x100m relay fastest man world charged clear <span style=\"background-color:rgba(255,215,0,0.3);\">united</span> <span style=\"background-color:rgba(255,215,0,0.3);\">state</span> rival justin gatlin jamaican quartet nesta carter kemar baileycole nickel ashmeade bolt 3736 second finished second 3756 second canada taking bronze britain disqualified faulty handover 26yearold bolt collected eight gold medal world championship equaling record held american trio carl lewis michael johnson allyson felix mention small matter six olympic title relay triumph followed individual success 100 200 meter russian capital I proud ill continue work dominate long possible bolt previously expressed intention carry 2016 rio olympics victory never seriously doubt got baton safely hand ashmeade gatlin <span style=\"background-color:rgba(255,215,0,0.3);\">united</span> <span style=\"background-color:rgba(255,215,0,0.3);\">state</span> third leg runner rakieem salaam problem gatlin strayed lane struggled get full control baton never able get term bolt earlier jamaica woman underlined dominance sprint event winning 4x100m relay gold anchored shellyann fraserpryce like bolt completing triple quartet recorded championship record 4129 second well clear <span style=\"background-color:rgba(255,215,0,0.3);\">france</span> crossed line second place 4273 second defending champion <span style=\"background-color:rgba(255,215,0,0.3);\">united</span> <span style=\"background-color:rgba(255,215,0,0.3);\">state</span> initially back bronze medal position losing <span style=\"background-color:rgba(255,215,0,0.3);\">time</span> second handover alexandria anderson english gardner promoted silver <span style=\"background-color:rgba(255,215,0,0.3);\">france</span> subsequently disqualified illegal handover british quartet initially fourth promoted bronze eluded men team fraserpryce like bolt aged 26 became <span style=\"background-color:rgba(255,215,0,0.3);\">first</span> woman achieve three gold 100200 relay final action last day championship <span style=\"background-color:rgba(255,215,0,0.3);\">france</span> teddy tamgho became third man leap 18m triple jump exceeding mark four centimeter take gold germany christina obergfoll finally took gold global level woman javelin five previous silver kenya asbel kiprop easily tactical men 1500m final kiprops compatriot eunice jepkoech sum surprise winner woman 800m bolt final dash golden glory brought eightday championship rousing finale host topped medal table <span style=\"background-color:rgba(255,215,0,0.3);\">united</span> <span style=\"background-color:rgba(255,215,0,0.3);\">state</span> criticism poor attendance luzhniki stadium concern pole vault gold medalist yelena isinbayeva made controversial remark support russia law make propagandizing nontraditional sexual relation among minor criminal offense later attempted clarify comment renewed call gay right group boycott 2014 winter game sochi next major sport event russia<br><br><strong>Predicted Summary</strong><br>manchester <span style=\"background-color:rgba(255,215,0,0.3);\">united</span> <span style=\"background-color:rgba(255,215,0,0.3);\">state</span> liverpool beat <span style=\"background-color:rgba(255,215,0,0.3);\">france</span> 21 draw cup 21 draw score goal goal goal win <span style=\"background-color:rgba(255,215,0,0.3);\">first</span> goal win <span style=\"background-color:rgba(255,215,0,0.3);\">first</span> <span style=\"background-color:rgba(255,215,0,0.3);\">time</span> since 22 minute arsenal 21 win"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Explainability\n",
    "match = display_string_matching(dtf_test[\"text_clean\"][i], predicted[i], both=True, sentences=False, \n",
    "                                titles=[\"Full Text\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5 - Model (Abstractive: Language model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test (no Train)\n",
    "predicted = bart(corpus=dtf_test[\"text\"], max_len=y_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rouge1: 0.57 | rouge2: 0.26 | rougeL: 0.26 --> avg rouge: 0.49\n"
     ]
    }
   ],
   "source": [
    "# Evaluate\n",
    "evaluate_summary(dtf_test[\"y\"][i], predicted[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Real Summary</strong><br><span style=\"background-color:rgba(255,215,0,0.3);\">Usain</span> <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> <span style=\"background-color:rgba(255,215,0,0.3);\">wins</span> <span style=\"background-color:rgba(255,215,0,0.3);\">third</span> <span style=\"background-color:rgba(255,215,0,0.3);\">gold</span> of <span style=\"background-color:rgba(255,215,0,0.3);\">world</span> championship . Anchors <span style=\"background-color:rgba(255,215,0,0.3);\">Jamaica</span> <span style=\"background-color:rgba(255,215,0,0.3);\">to</span> <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m</span> relay <span style=\"background-color:rgba(255,215,0,0.3);\">victory</span> . Eighth <span style=\"background-color:rgba(255,215,0,0.3);\">gold</span> <span style=\"background-color:rgba(255,215,0,0.3);\">at</span> <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> <span style=\"background-color:rgba(255,215,0,0.3);\">championships</span> for <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> . <span style=\"background-color:rgba(255,215,0,0.3);\">Jamaica</span> double up <span style=\"background-color:rgba(255,215,0,0.3);\">in</span> women's <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m</span> relay .<br><br><strong>Predicted Summary</strong><br><span style=\"background-color:rgba(255,215,0,0.3);\">Usain</span> <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> <span style=\"background-color:rgba(255,215,0,0.3);\">wins</span> his <span style=\"background-color:rgba(255,215,0,0.3);\">third</span> <span style=\"background-color:rgba(255,215,0,0.3);\">gold</span> <span style=\"background-color:rgba(255,215,0,0.3);\">at</span> <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> <span style=\"background-color:rgba(255,215,0,0.3);\">world</span> <span style=\"background-color:rgba(255,215,0,0.3);\">championships</span> <span style=\"background-color:rgba(255,215,0,0.3);\">in</span> Moscow. <span style=\"background-color:rgba(255,215,0,0.3);\">Bolt</span> anchored <span style=\"background-color:rgba(255,215,0,0.3);\">Jamaica</span> <span style=\"background-color:rgba(255,215,0,0.3);\">to</span> <span style=\"background-color:rgba(255,215,0,0.3);\">victory</span> <span style=\"background-color:rgba(255,215,0,0.3);\">in</span> <span style=\"background-color:rgba(255,215,0,0.3);\">the</span> men's <span style=\"background-color:rgba(255,215,0,0.3);\">4x100m,</span>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compare y_test and predicted\n",
    "match = display_string_matching(dtf_test[\"y\"][i], predicted[i], both=True, sentences=False, \n",
    "                                titles=[\"Real Summary\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<strong>Full Text</strong><br>(CNN) -- Usain Bolt rounded off <span style=\"background-color:rgba(255,215,0,0.3);\"> the world championships </span> Sunday by claiming his third gold in Moscow as he <span style=\"background-color:rgba(255,215,0,0.3);\"> anchored Jamaica to victory in the men's 4x100m</span>  relay. The fastest man in the world charged clear of United States rival Justin Gatlin as the Jamaican quartet of Nesta Carter, Kemar Bailey-Cole, Nickel Ashmeade and Bolt won in 37.36 seconds. The U.S finished second in 37.56 seconds with Canada taking the bronze after Britain were disqualified for a faulty handover. The 26-year-old Bolt has now collected eight gold medals at world championships, equaling the record held by American trio Carl Lewis, Michael Johnson and Allyson Felix, not to mention the small matter of six Olympic titles. The relay triumph followed individual successes in the 100 and 200 meters in the Russian capital. \"I'm proud of myself and I'll continue to work to dominate for as long as possible,\" Bolt said, having previously expressed his intention to carry on until the 2016 Rio Olympics. Victory was never seriously in doubt once he got the baton safely in hand from Ashmeade, while Gatlin and the United States third leg runner Rakieem Salaam had problems. Gatlin strayed out of his lane as he struggled to get full control of their baton and was never able to get on terms with Bolt. Earlier, Jamaica's women underlined their dominance in the sprint events by winning the 4x100m relay gold, anchored by Shelly-Ann Fraser-Pryce, who like Bolt was completing a triple. Their quartet recorded a championship record of 41.29 seconds, well clear of France, who crossed the line in second place in 42.73 seconds. Defending champions, the United States, were initially back in the bronze medal position after losing time on the second handover between Alexandria Anderson and English Gardner, but promoted to silver when France were subsequently disqualified for an illegal handover. The British quartet, who were initially fourth, were promoted to the bronze which eluded their men's team. Fraser-Pryce, like Bolt aged 26, became the first woman to achieve three golds in the 100-200 and the relay. In other final action on the last day of the championships, France's Teddy Tamgho became the third man to leap over 18m in the triple jump, exceeding the mark by four centimeters to take gold. Germany's Christina Obergfoll finally took gold at global level in the women's javelin after five previous silvers, while Kenya's Asbel Kiprop easily won a tactical men's 1500m final. Kiprop's compatriot Eunice Jepkoech Sum was a surprise winner of the women's 800m. Bolt's final dash for golden glory brought the eight-day championship to a rousing finale, but while the hosts topped the medal table from the United States there was criticism of the poor attendances in the Luzhniki Stadium. There was further concern when their pole vault gold medalist Yelena Isinbayeva made controversial remarks in support of Russia's new laws, which make \"the propagandizing of non-traditional sexual relations among minors\" a criminal offense. She later attempted to clarify her comments, but there were renewed calls by gay rights groups for a boycott of the 2014 Winter Games in Sochi, the next major sports event in Russia.<br><br><strong>Predicted Summary</strong><br> Usain Bolt wins his third gold at <span style=\"background-color:rgba(255,215,0,0.3);\"> the world championships </span> in Moscow. Bolt <span style=\"background-color:rgba(255,215,0,0.3);\"> anchored Jamaica to victory in the men's 4x100m</span> ,"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Explainability\n",
    "match = display_string_matching(dtf_test[\"text\"][i], predicted[i], both=True, sentences=True, \n",
    "                                titles=[\"Full Text\", \"Predicted Summary\"])\n",
    "\n",
    "from IPython.core.display import display, HTML\n",
    "display(HTML(match))"
   ]
  }
 ],
 "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": "3.6.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}