code/title_generation/title_generation_T5_of_the_shelf.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "228f3b49",
   "metadata": {},
   "source": [
    "#### Generate headline based on the article text using T5-base model on train and test set of clickbait and no-clickbait samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "387948b4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import T5ForConditionalGeneration, T5Tokenizer\n",
    "import pandas as pd\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a3c01b84",
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cpu'\n",
    "model = T5ForConditionalGeneration.from_pretrained(\"Michau/t5-base-en-generate-headline\")\n",
    "tokenizer = T5Tokenizer.from_pretrained(\"Michau/t5-base-en-generate-headline\")\n",
    "model = model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "702c9bbf",
   "metadata": {},
   "outputs": [],
   "source": [
    "max_len = 128\n",
    "\n",
    "def get_generated_headline(text, i):    \n",
    "    print(i)\n",
    "    encoding = tokenizer.encode_plus(text, return_tensors = \"pt\")\n",
    "    input_ids = encoding[\"input_ids\"].to(device)\n",
    "    attention_masks = encoding[\"attention_mask\"].to(device)\n",
    "    \n",
    "    beam_outputs = model.generate(\n",
    "    input_ids = input_ids,\n",
    "    attention_mask = attention_masks,\n",
    "    max_length = max_len,\n",
    "    num_beams = 3,\n",
    "    early_stopping = True,\n",
    "    )\n",
    "\n",
    "    result = tokenizer.decode(beam_outputs[0])\n",
    "#     print(result)\n",
    "    result = result.split('>')[1].split('</')[0].strip()\n",
    "\n",
    "    return result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "b11feb38",
   "metadata": {},
   "outputs": [],
   "source": [
    "path = \"./TitleGenData/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ee94a5d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "clickbait_test_data.csv\n",
      "no_clickbait_test_data.csv\n",
      "clickbait_train_data.csv\n",
      ".ipynb_checkpoints\n",
      "no_clickbait_train_data.csv\n"
     ]
    }
   ],
   "source": [
    "for f in os.listdir(path):\n",
    "    print(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ef5ba3b4",
   "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>postText</th>\n",
       "      <th>targetParagraphs</th>\n",
       "      <th>truthClass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>UK’s response to modern slavery leaving victim...</td>\n",
       "      <td>Thousands of modern slavery victims have not c...</td>\n",
       "      <td>no-clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>The \"forgotten\" Trump roast: Relive his brutal...</td>\n",
       "      <td>When the White House correspondents’ dinner is...</td>\n",
       "      <td>no-clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Tokyo's subway is shut down amid fears over an...</td>\n",
       "      <td>One of Tokyo's major subways systems says it s...</td>\n",
       "      <td>no-clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Ban lifted on Madrid doping laboratory</td>\n",
       "      <td>Share this with Madrid's Anti-Doping Laborator...</td>\n",
       "      <td>no-clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Despite the ‘Yuck Factor,’ Leeches Are Big in ...</td>\n",
       "      <td>MOSCOW — They are small as physician assistant...</td>\n",
       "      <td>no-clickbait</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            postText  \\\n",
       "0  UK’s response to modern slavery leaving victim...   \n",
       "1  The \"forgotten\" Trump roast: Relive his brutal...   \n",
       "2  Tokyo's subway is shut down amid fears over an...   \n",
       "3             Ban lifted on Madrid doping laboratory   \n",
       "4  Despite the ‘Yuck Factor,’ Leeches Are Big in ...   \n",
       "\n",
       "                                    targetParagraphs    truthClass  \n",
       "0  Thousands of modern slavery victims have not c...  no-clickbait  \n",
       "1  When the White House correspondents’ dinner is...  no-clickbait  \n",
       "2  One of Tokyo's major subways systems says it s...  no-clickbait  \n",
       "3  Share this with Madrid's Anti-Doping Laborator...  no-clickbait  \n",
       "4  MOSCOW — They are small as physician assistant...  no-clickbait  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# run the scriipt on each of these CSVs\n",
    "\n",
    "readp = path + 'no_clickbait_train_data.csv'\n",
    "no_clickbait_train_data = pd.read_csv(readp)\n",
    "no_clickbait_train_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9799b3e7",
   "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>postText</th>\n",
       "      <th>targetParagraphs</th>\n",
       "      <th>truthClass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>this is good</td>\n",
       "      <td>President Donald Trump has appointed the pro-l...</td>\n",
       "      <td>clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Meet the happiest #dog in the world!</td>\n",
       "      <td>Adorable is probably an understatement. This a...</td>\n",
       "      <td>clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>The top 10 best selling comic books of all time</td>\n",
       "      <td>More By Brian Prowse-Gany In celebration of Na...</td>\n",
       "      <td>clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Players to know, QBs to watch and more 👀nnGuid...</td>\n",
       "      <td>The first round of the 2017 NFL draft was nota...</td>\n",
       "      <td>clickbait</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>White House reporter @alivitali on her first 1...</td>\n",
       "      <td>The most consistent thing about my life for th...</td>\n",
       "      <td>clickbait</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            postText  \\\n",
       "0                                       this is good   \n",
       "1               Meet the happiest #dog in the world!   \n",
       "2    The top 10 best selling comic books of all time   \n",
       "3  Players to know, QBs to watch and more 👀nnGuid...   \n",
       "4  White House reporter @alivitali on her first 1...   \n",
       "\n",
       "                                    targetParagraphs truthClass  \n",
       "0  President Donald Trump has appointed the pro-l...  clickbait  \n",
       "1  Adorable is probably an understatement. This a...  clickbait  \n",
       "2  More By Brian Prowse-Gany In celebration of Na...  clickbait  \n",
       "3  The first round of the 2017 NFL draft was nota...  clickbait  \n",
       "4  The most consistent thing about my life for th...  clickbait  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "readp = path + 'clickbait_train_data.csv'\n",
    "clickbait_train_data = pd.read_csv(readp)\n",
    "clickbait_train_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e5a40a16",
   "metadata": {},
   "outputs": [],
   "source": [
    "chop = 512 # take the first 512 characters of targetParagraphs\n",
    "\n",
    "all_targetParagraphs = [] \n",
    "result = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "79861c67",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i, row in clickbait_train_data.iterrows():\n",
    "    all_targetParagraphs.append(row['targetParagraphs'][:chop])        \n",
    "\n",
    "# all_targetParagraphs = all_targetParagraphs[:10]\n",
    "print('all_targetParagraphs:', len(all_targetParagraphs))\n",
    "\n",
    "# all_targetParagraphs=all_targetParagraphs[698:700]\n",
    "for i, text in enumerate(all_targetParagraphs):\n",
    "    result.append(get_generated_headline(text, i))\n",
    "\n",
    "print('done', len(result))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "39ec6f58",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5437\n"
     ]
    }
   ],
   "source": [
    "print(len(result))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2c3ff48b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "clickbait_train_data['generatedPostText'] = result\n",
    "clickbait_train_data.to_csv('./gen_clickbait_train_data.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "215b0774",
   "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>postText</th>\n",
       "      <th>targetParagraphs</th>\n",
       "      <th>truthClass</th>\n",
       "      <th>generatedPostText</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>this is good</td>\n",
       "      <td>President Donald Trump has appointed the pro-l...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>Trump Appoints Pro-Life Advocate Dr. Charmaine...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Meet the happiest #dog in the world!</td>\n",
       "      <td>Adorable is probably an understatement. This a...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>Maru the Husky - Part Husky, Part Panda, Part ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>The top 10 best selling comic books of all time</td>\n",
       "      <td>More By Brian Prowse-Gany In celebration of Na...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>Top 10 Comic Books of All Time</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Players to know, QBs to watch and more 👀nnGuid...</td>\n",
       "      <td>The first round of the 2017 NFL draft was nota...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>The 2017 NFL Draft was notable for a number of...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>White House reporter @alivitali on her first 1...</td>\n",
       "      <td>The most consistent thing about my life for th...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>The Most Consistent Thing About My Life</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5432</th>\n",
       "      <td>RT @capitalweather: The Derecho, from D to O: ...</td>\n",
       "      <td>This post was originally published on June 11,...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>Derecho Season in the Northwest</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5433</th>\n",
       "      <td>Restaurant won't apologize for tasteless Caitl...</td>\n",
       "      <td>Whoever comes up with the menus for Nacho Mama...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>How the Kardashians and Jenners Supported Cait...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5434</th>\n",
       "      <td>As a BASE jumper leaped, his girlfriend snappe...</td>\n",
       "      <td>Dean Potter jumped. Graham Hunt followed. Pott...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>YOSEMITE NATIONAL PARK, Calif.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5435</th>\n",
       "      <td>Women scientists are tweeting \"sexy” photos of...</td>\n",
       "      <td>Marine biologists, archeologists, and many oth...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>#DistractinglySexy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5436</th>\n",
       "      <td>Man who received world's first penis transplan...</td>\n",
       "      <td>The man who underwent the world's first succes...</td>\n",
       "      <td>clickbait</td>\n",
       "      <td>The first successful penis transplant is set t...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5437 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               postText  \\\n",
       "0                                          this is good   \n",
       "1                  Meet the happiest #dog in the world!   \n",
       "2       The top 10 best selling comic books of all time   \n",
       "3     Players to know, QBs to watch and more 👀nnGuid...   \n",
       "4     White House reporter @alivitali on her first 1...   \n",
       "...                                                 ...   \n",
       "5432  RT @capitalweather: The Derecho, from D to O: ...   \n",
       "5433  Restaurant won't apologize for tasteless Caitl...   \n",
       "5434  As a BASE jumper leaped, his girlfriend snappe...   \n",
       "5435  Women scientists are tweeting \"sexy” photos of...   \n",
       "5436  Man who received world's first penis transplan...   \n",
       "\n",
       "                                       targetParagraphs truthClass  \\\n",
       "0     President Donald Trump has appointed the pro-l...  clickbait   \n",
       "1     Adorable is probably an understatement. This a...  clickbait   \n",
       "2     More By Brian Prowse-Gany In celebration of Na...  clickbait   \n",
       "3     The first round of the 2017 NFL draft was nota...  clickbait   \n",
       "4     The most consistent thing about my life for th...  clickbait   \n",
       "...                                                 ...        ...   \n",
       "5432  This post was originally published on June 11,...  clickbait   \n",
       "5433  Whoever comes up with the menus for Nacho Mama...  clickbait   \n",
       "5434  Dean Potter jumped. Graham Hunt followed. Pott...  clickbait   \n",
       "5435  Marine biologists, archeologists, and many oth...  clickbait   \n",
       "5436  The man who underwent the world's first succes...  clickbait   \n",
       "\n",
       "                                      generatedPostText  \n",
       "0     Trump Appoints Pro-Life Advocate Dr. Charmaine...  \n",
       "1     Maru the Husky - Part Husky, Part Panda, Part ...  \n",
       "2                        Top 10 Comic Books of All Time  \n",
       "3     The 2017 NFL Draft was notable for a number of...  \n",
       "4               The Most Consistent Thing About My Life  \n",
       "...                                                 ...  \n",
       "5432                    Derecho Season in the Northwest  \n",
       "5433  How the Kardashians and Jenners Supported Cait...  \n",
       "5434                     YOSEMITE NATIONAL PARK, Calif.  \n",
       "5435                                 #DistractinglySexy  \n",
       "5436  The first successful penis transplant is set t...  \n",
       "\n",
       "[5437 rows x 4 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clickbait_train_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e9b4dddd",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "language": "python",
   "name": "venv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}