{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "1f43149f",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import string\n",
    "from collections import defaultdict\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "cecb763a",
   "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>Freshness</th>\n",
       "      <th>Review</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>fresh</td>\n",
       "      <td>Manakamana doesn't answer any questions, yet ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>fresh</td>\n",
       "      <td>Wilfully offensive and powered by a chest-thu...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>rotten</td>\n",
       "      <td>It would be difficult to imagine material mor...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>rotten</td>\n",
       "      <td>Despite the gusto its star brings to the role...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>rotten</td>\n",
       "      <td>If there was a good idea at the core of this ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>479995</th>\n",
       "      <td>rotten</td>\n",
       "      <td>Zemeckis seems unable to admit that the motio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>479996</th>\n",
       "      <td>fresh</td>\n",
       "      <td>Movies like The Kids Are All Right -- beautif...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>479997</th>\n",
       "      <td>rotten</td>\n",
       "      <td>Film-savvy audiences soon will catch onto Win...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>479998</th>\n",
       "      <td>fresh</td>\n",
       "      <td>An odd yet enjoyable film.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>479999</th>\n",
       "      <td>fresh</td>\n",
       "      <td>No other animation studio, even our beloved P...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>480000 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Freshness                                             Review\n",
       "0          fresh   Manakamana doesn't answer any questions, yet ...\n",
       "1          fresh   Wilfully offensive and powered by a chest-thu...\n",
       "2         rotten   It would be difficult to imagine material mor...\n",
       "3         rotten   Despite the gusto its star brings to the role...\n",
       "4         rotten   If there was a good idea at the core of this ...\n",
       "...          ...                                                ...\n",
       "479995    rotten   Zemeckis seems unable to admit that the motio...\n",
       "479996     fresh   Movies like The Kids Are All Right -- beautif...\n",
       "479997    rotten   Film-savvy audiences soon will catch onto Win...\n",
       "479998     fresh                        An odd yet enjoyable film. \n",
       "479999     fresh   No other animation studio, even our beloved P...\n",
       "\n",
       "[480000 rows x 2 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the dataset\n",
    "data = pd.read_csv('rt_reviews.csv',encoding='ISO-8859–1')\n",
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5310f806",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "480000"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "068c4d8b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(480000, 2)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#shape of data\n",
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d4d26a3b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "\n",
    "# the number of occurrences of each class\n",
    "class_counts = data['Freshness'].value_counts()\n",
    "\n",
    "# Create the bar graph\n",
    "fig, ax = plt.subplots()\n",
    "ax.bar(class_counts.index, class_counts.values,color=['#EFD3F5', '#9C919F'])\n",
    "\n",
    "# Set the title and axis labels\n",
    "ax.set_title('Fresh and rotten ')\n",
    "ax.set_xlabel('Class')\n",
    "ax.set_ylabel('Count')\n",
    "\n",
    "# Display the graph\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0f93991d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#remove duplicate rows from data set\n",
    "data=data.drop_duplicates()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b38f655e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "339716"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#length of data set after deleting duplicated rows\n",
    "len(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "17be97e6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(339716, 2)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "8397b342",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#bar graph after deleting duplicate rows\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "\n",
    "# Count the number of occurrences of each class\n",
    "class_counts = data['Freshness'].value_counts()\n",
    "\n",
    "# Create the bar graph\n",
    "fig, ax = plt.subplots()\n",
    "ax.bar(class_counts.index, class_counts.values,color=['#EFD3F5', '#9C919F'])\n",
    "\n",
    "# Set the title and axis labels\n",
    "ax.set_title('Fresh and rotten')\n",
    "ax.set_xlabel('Class')\n",
    "ax.set_ylabel('Count')\n",
    "\n",
    "# Display the graph\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6f82c95e",
   "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>Freshness</th>\n",
       "      <th>Review</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>fresh</td>\n",
       "      <td>Manakamana doesn't answer any questions, yet ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>fresh</td>\n",
       "      <td>Wilfully offensive and powered by a chest-thu...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>rotten</td>\n",
       "      <td>It would be difficult to imagine material mor...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>rotten</td>\n",
       "      <td>Despite the gusto its star brings to the role...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>rotten</td>\n",
       "      <td>If there was a good idea at the core of this ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Freshness                                             Review\n",
       "0     fresh   Manakamana doesn't answer any questions, yet ...\n",
       "1     fresh   Wilfully offensive and powered by a chest-thu...\n",
       "2    rotten   It would be difficult to imagine material mor...\n",
       "3    rotten   Despite the gusto its star brings to the role...\n",
       "4    rotten   If there was a good idea at the core of this ..."
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "835e9c31",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Freshness', 'Review'], dtype='object')"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "61ff0dbe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Freshness    0\n",
       "Review       1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "62f4754f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\harsh\\AppData\\Local\\Temp\\ipykernel_24880\\2675036435.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  data['Review'] = data['Review'].apply(lambda Re: str(Re).replace(\"-\",\" \"))\n",
      "C:\\Users\\harsh\\AppData\\Local\\Temp\\ipykernel_24880\\2675036435.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  data['Review'] = data['Review'].apply(lambda Re: re.sub(r'[^a-zA-Z\\s]', '', Re))\n",
      "C:\\Users\\harsh\\AppData\\Local\\Temp\\ipykernel_24880\\2675036435.py:5: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  data['Review'] = data['Review'].str.lower()\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0          manakamana doesnt answer any questions yet ma...\n",
       "1          wilfully offensive and powered by a chest thu...\n",
       "2          it would be difficult to imagine material mor...\n",
       "3          despite the gusto its star brings to the role...\n",
       "4          if there was a good idea at the core of this ...\n",
       "                                ...                        \n",
       "479994     roland joffes deeply ridiculous movie is caug...\n",
       "479996     movies like the kids are all right    beautif...\n",
       "479997     film savvy audiences soon will catch onto win...\n",
       "479998                           an odd yet enjoyable film \n",
       "479999     no other animation studio even our beloved pi...\n",
       "Name: Review, Length: 339716, dtype: object"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#removing special charecters and chaning words into lower case\n",
    "import re\n",
    "data['Review'] = data['Review'].apply(lambda Re: str(Re).replace(\"-\",\" \"))\n",
    "data['Review'] = data['Review'].apply(lambda Re: re.sub(r'[^a-zA-Z\\s]', '', Re))\n",
    "data['Review'] = data['Review'].str.lower()\n",
    "data['Review']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "09cc3798",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set size: 237801\n",
      "Development set size: 50957\n",
      "Test set size: 50958\n"
     ]
    }
   ],
   "source": [
    "# Decide on the proportions of your data\n",
    "train_size = 0.7\n",
    "dev_test_size = 0.3\n",
    "\n",
    "# Split your data into the training set and the rest of the data\n",
    "train_data, remaining_data = train_test_split(data, train_size=train_size, random_state=42)\n",
    "\n",
    "# Split the remaining data into the development set and the test set\n",
    "dev_data, test_data = train_test_split(remaining_data, train_size=dev_test_size/(dev_test_size+dev_test_size), random_state=42)\n",
    "\n",
    "# Print the sizes of each set\n",
    "print(f\"Training set size: {len(train_data)}\")\n",
    "print(f\"Development set size: {len(dev_data)}\")\n",
    "print(f\"Test set size: {len(test_data)}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "399b6eb1",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df=train_data.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "68205dbe",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df_Review=train_df.Review.tolist()\n",
    "train_df_Freshness=train_df.Freshness.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "fabeb153",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'fresh': {}, 'rotten': {}}\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'fresh': 0.0, 'rotten': 0.0}"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a dictionary to store the frequency of each word for each class\n",
    "class_word_counts = dict()\n",
    "class_word_counts['fresh']={}\n",
    "class_word_counts['rotten']={}\n",
    "print(class_word_counts)\n",
    "# Get the total number of reviews in each class\n",
    "class_count = {}\n",
    "class_count['fresh']=0.0\n",
    "class_count['rotten']=0.0\n",
    "class_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "1b014dee",
   "metadata": {},
   "outputs": [],
   "source": [
    "#creat a set to add words\n",
    "vocabulary=set()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "e53bfed5",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(train_df_Review)):\n",
    "    sentence = train_df_Review[i]\n",
    "    sentence_words = sentence.split()\n",
    "    class_name = train_df_Freshness[i] \n",
    "    if class_name == \"fresh\":\n",
    "        for word in sentence_words:\n",
    "            if word in class_word_counts[\"fresh\"]:\n",
    "                class_word_counts[\"fresh\"][word] += 1\n",
    "            else:\n",
    "                class_word_counts[\"fresh\"][word] = 1\n",
    "            vocabulary.add(word)  #adding word to vocabulary\n",
    "    if class_name == \"rotten\":\n",
    "        for word in sentence_words:\n",
    "            if word in class_word_counts[\"rotten\"]:\n",
    "                class_word_counts[\"rotten\"][word] += 1\n",
    "            else:\n",
    "                class_word_counts[\"rotten\"][word] = 1\n",
    "            vocabulary.add(word)\n",
    "    class_count[class_name]+=1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "30cc1f67",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'fresh': 131979.0, 'rotten': 105822.0}"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "44d00aa4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'unaided',\n",
       " 'dashboards',\n",
       " 'hpbj',\n",
       " 'mackenzie',\n",
       " 'intentioned',\n",
       " 'greases',\n",
       " 'kinderkultur',\n",
       " 'nikolaj',\n",
       " 'deepthink',\n",
       " 'eskil',\n",
       " 'howitts',\n",
       " 'hardens',\n",
       " 'formal',\n",
       " 'rinzlers',\n",
       " 'refilmagem',\n",
       " 'dans',\n",
       " 'hudsonkatherine',\n",
       " 'admire',\n",
       " 'sportscenter',\n",
       " 'polaroids',\n",
       " 'reif',\n",
       " 'versions',\n",
       " 'smuel',\n",
       " 'anguish',\n",
       " 'machinations',\n",
       " 'twistings',\n",
       " 'untraveled',\n",
       " 'clutch',\n",
       " 'impishness',\n",
       " 'observer',\n",
       " 'physicists',\n",
       " 'comcasts',\n",
       " 'messagesbut',\n",
       " 'whallop',\n",
       " 'distressingly',\n",
       " 'fale',\n",
       " 'shakycams',\n",
       " 'tribespeople',\n",
       " 'alienate',\n",
       " 'impetuous',\n",
       " 'shambly',\n",
       " 'smoulder',\n",
       " 'jg',\n",
       " 'unrelievedly',\n",
       " 'doppler',\n",
       " 'munsons',\n",
       " 'expectingand',\n",
       " 'grashaw',\n",
       " 'soundscape',\n",
       " 'sidney',\n",
       " 'csokas',\n",
       " 'existenceof',\n",
       " 'antin',\n",
       " 'steals',\n",
       " 'underarticulated',\n",
       " 'craggy',\n",
       " 'baileys',\n",
       " 'megadeth',\n",
       " 'ash',\n",
       " 'incorrectness',\n",
       " 'fanswho',\n",
       " 'lowish',\n",
       " 'stewpot',\n",
       " 'gummy',\n",
       " 'courtisane',\n",
       " 'congratulatory',\n",
       " 'princbaby',\n",
       " 'hmmmmight',\n",
       " 'swarms',\n",
       " 'bergmanian',\n",
       " 'camps',\n",
       " 'gdp',\n",
       " 'napalms',\n",
       " 'clobberin',\n",
       " 'moscows',\n",
       " 'ansiedad',\n",
       " 'hickenlooper',\n",
       " 'airbender',\n",
       " 'midsectionbut',\n",
       " 'raincoats',\n",
       " 'fyodor',\n",
       " 'taos',\n",
       " 'countrymen',\n",
       " 'brute',\n",
       " 'expendables',\n",
       " 'finissent',\n",
       " 'vithobas',\n",
       " 'tory',\n",
       " 'carin',\n",
       " 'kallos',\n",
       " 'wahlberg',\n",
       " 'climaxyet',\n",
       " 'ballyhooed',\n",
       " 'popeye',\n",
       " 'lofty',\n",
       " 'posterior',\n",
       " 'rolex',\n",
       " 'peekaboo',\n",
       " 'swingin',\n",
       " 'citadel',\n",
       " 'jeu',\n",
       " 'impassionate',\n",
       " 'dimitris',\n",
       " 'rejected',\n",
       " 'marisas',\n",
       " 'overstate',\n",
       " 'resolving',\n",
       " 'satirical',\n",
       " 'offfers',\n",
       " 'gum',\n",
       " 'artsy',\n",
       " 'unidimensionality',\n",
       " 'akcijskih',\n",
       " 'starijim',\n",
       " 'hypocrisy',\n",
       " 'ces',\n",
       " 'indulge',\n",
       " 'irritants',\n",
       " 'sprinkle',\n",
       " 'colloquially',\n",
       " 'cfs',\n",
       " 'clever',\n",
       " 'seeming',\n",
       " 'backache',\n",
       " 'beane',\n",
       " 'confere',\n",
       " 'gallagher',\n",
       " 'surrey',\n",
       " 'emotioanlly',\n",
       " 'wonderstruck',\n",
       " 'developmentthis',\n",
       " 'mputu',\n",
       " 'crams',\n",
       " 'hipsterism',\n",
       " 'rational',\n",
       " 'bazin',\n",
       " 'childhoods',\n",
       " 'designers',\n",
       " 'clients',\n",
       " 'rustic',\n",
       " 'atroci',\n",
       " 'loath',\n",
       " 'mystifyingly',\n",
       " 'clinks',\n",
       " 'gymnopedie',\n",
       " 'cup',\n",
       " 'finagles',\n",
       " 'waits',\n",
       " 'burrells',\n",
       " 'levetts',\n",
       " 'evergreen',\n",
       " 'talbot',\n",
       " 'stuntcraft',\n",
       " 'greengrasss',\n",
       " 'fumbles',\n",
       " 'smoking',\n",
       " 'totem',\n",
       " 'aftermaths',\n",
       " 'sketchiness',\n",
       " 'uncompelling',\n",
       " 'mero',\n",
       " 'mccarthys',\n",
       " 'reasondie',\n",
       " 'odball',\n",
       " 'elocution',\n",
       " 'proofreading',\n",
       " 'disqualified',\n",
       " 'mandarin',\n",
       " 'matine',\n",
       " 'crockery',\n",
       " 'possess',\n",
       " 'novie',\n",
       " 'dever',\n",
       " 'lau',\n",
       " 'relationsh',\n",
       " 'shtload',\n",
       " 'isthere',\n",
       " 'faggot',\n",
       " 'prende',\n",
       " 'herding',\n",
       " 'spectatorship',\n",
       " 'brawls',\n",
       " 'generoso',\n",
       " 'configures',\n",
       " 'gritty',\n",
       " 'stepmother',\n",
       " 'judd',\n",
       " 'reassure',\n",
       " 'sairat',\n",
       " 'moussaka',\n",
       " 'deduce',\n",
       " 'venerable',\n",
       " 'previsible',\n",
       " 'umbilicus',\n",
       " 'zdt',\n",
       " 'bathrobe',\n",
       " 'jug',\n",
       " 'coyotes',\n",
       " 'humane',\n",
       " 'touts',\n",
       " 'imaxs',\n",
       " 'nasty',\n",
       " 'skipped',\n",
       " 'retrospectives',\n",
       " 'dogmas',\n",
       " 'maid',\n",
       " 'dou',\n",
       " 'recallsannie',\n",
       " 'gearshift',\n",
       " 'hereafter',\n",
       " 'hatar',\n",
       " 'quarrelsome',\n",
       " 'ferdanes',\n",
       " 'gather',\n",
       " 'hounsous',\n",
       " 'yet',\n",
       " 'zeva',\n",
       " 'faradays',\n",
       " 'allergy',\n",
       " 'intercommunication',\n",
       " 'whew',\n",
       " 'palid',\n",
       " 'streetwise',\n",
       " 'logistics',\n",
       " 'practising',\n",
       " 'debie',\n",
       " 'tomeis',\n",
       " 'glibdrowns',\n",
       " 'deejayed',\n",
       " 'sevigny',\n",
       " 'ochentero',\n",
       " 'bight',\n",
       " 'diggs',\n",
       " 'flatten',\n",
       " 'quilleveres',\n",
       " 'discounts',\n",
       " 'hardbound',\n",
       " 'unload',\n",
       " 'curioso',\n",
       " 'splotchy',\n",
       " 'mcgs',\n",
       " 'tsukamoto',\n",
       " 'kamaleswar',\n",
       " 'robbing',\n",
       " 'overgrown',\n",
       " 'amores',\n",
       " 'loggia',\n",
       " 'conceptual',\n",
       " 'prabhas',\n",
       " 'shawn',\n",
       " 'avails',\n",
       " 'chomkos',\n",
       " 'mcdonalds',\n",
       " 'romancing',\n",
       " 'imbuing',\n",
       " 'charted',\n",
       " 'beardens',\n",
       " 'crossword',\n",
       " 'roteirista',\n",
       " 'millie',\n",
       " 'disinclination',\n",
       " 'babysitter',\n",
       " 'crawler',\n",
       " 'pantsing',\n",
       " 'warlord',\n",
       " 'decolonised',\n",
       " 'dunstan',\n",
       " 'preternaturally',\n",
       " 'sofias',\n",
       " 'foe',\n",
       " 'trinca',\n",
       " 'conspire',\n",
       " 'transpecos',\n",
       " 'rancid',\n",
       " 'calculated',\n",
       " 'warners',\n",
       " 'hecho',\n",
       " 'mclean',\n",
       " 'supernova',\n",
       " 'chum',\n",
       " 'oozer',\n",
       " 'lovesometimes',\n",
       " 'imitative',\n",
       " 'gerhard',\n",
       " 'lorre',\n",
       " 'hologram',\n",
       " 'panders',\n",
       " 'danas',\n",
       " 'restaurant',\n",
       " 'betsey',\n",
       " 'impossi',\n",
       " 'palito',\n",
       " 'convicts',\n",
       " 'obviates',\n",
       " 'commentative',\n",
       " 'tenta',\n",
       " 'finland',\n",
       " 'bookson',\n",
       " 'tinges',\n",
       " 'fawkes',\n",
       " 'marred',\n",
       " 'wondering',\n",
       " 'sojourns',\n",
       " 'bug',\n",
       " 'vibey',\n",
       " 'directorcomposer',\n",
       " 'daring',\n",
       " 'enlightenment',\n",
       " 'outshining',\n",
       " 'kasbah',\n",
       " 'pixars',\n",
       " 'overcompensating',\n",
       " 'hurwitz',\n",
       " 'engrossingly',\n",
       " 'motes',\n",
       " 'andor',\n",
       " 'lessens',\n",
       " 'klux',\n",
       " 'featureh',\n",
       " 'harlin',\n",
       " 'stamm',\n",
       " 'formulamatic',\n",
       " 'oslo',\n",
       " 'drom',\n",
       " 'fugeddaboutit',\n",
       " 'intoning',\n",
       " 'ulm',\n",
       " 'whitewashing',\n",
       " 'repaved',\n",
       " 'qing',\n",
       " 'hostels',\n",
       " 'chases',\n",
       " 'sunken',\n",
       " 'penumbral',\n",
       " 'copilot',\n",
       " 'tech',\n",
       " 'goosed',\n",
       " 'snobbish',\n",
       " 'groin',\n",
       " 'craziest',\n",
       " 'salma',\n",
       " 'complaisance',\n",
       " 'gloomiest',\n",
       " 'glastonbury',\n",
       " 'snorecraft',\n",
       " 'contrasted',\n",
       " 'privy',\n",
       " 'neckties',\n",
       " 'enought',\n",
       " 'lorn',\n",
       " 'moneyed',\n",
       " 'heinemans',\n",
       " 'menwomen',\n",
       " 'desensitising',\n",
       " 'lineman',\n",
       " 'ackerman',\n",
       " 'inventively',\n",
       " 'kalytero',\n",
       " 'rasgulla',\n",
       " 'newport',\n",
       " 'tenner',\n",
       " 'supergroup',\n",
       " 'toxicity',\n",
       " 'quickly',\n",
       " 'backdrops',\n",
       " 'offy',\n",
       " 'sieve',\n",
       " 'combative',\n",
       " 'jungermanns',\n",
       " 'mormonsploitation',\n",
       " 'oratory',\n",
       " 'segal',\n",
       " 'squawky',\n",
       " 'wraith',\n",
       " 'wolfbitch',\n",
       " 'irittu',\n",
       " 'ohmagosh',\n",
       " 'moviemaker',\n",
       " 'faded',\n",
       " 'deviate',\n",
       " 'storewindow',\n",
       " 'fairytale',\n",
       " 'smugbut',\n",
       " 'minnellis',\n",
       " 'billboards',\n",
       " 'infants',\n",
       " 'shlockily',\n",
       " 'fowlers',\n",
       " 'trent',\n",
       " 'low',\n",
       " 'legoland',\n",
       " 'foundations',\n",
       " 'stoli',\n",
       " 'stanford',\n",
       " 'rehearsed',\n",
       " 'reeled',\n",
       " 'ladrn',\n",
       " 'unwrapping',\n",
       " 'hayseed',\n",
       " 'seasickness',\n",
       " 'advised',\n",
       " 'portend',\n",
       " 'franzoni',\n",
       " 'arrgh',\n",
       " 'holliday',\n",
       " 'profiles',\n",
       " 'cer',\n",
       " 'bloodsipping',\n",
       " 'lapsed',\n",
       " 'esperan',\n",
       " 'perspicacity',\n",
       " 'bodied',\n",
       " 'adorbs',\n",
       " 'pawlikowski',\n",
       " 'summit',\n",
       " 'inkiness',\n",
       " 'resurfaces',\n",
       " 'metaphysical',\n",
       " 'blinging',\n",
       " 'calm',\n",
       " 'impetuosity',\n",
       " 'orson',\n",
       " 'cleos',\n",
       " 'antarcticas',\n",
       " 'affiliation',\n",
       " 'telecast',\n",
       " 'kempner',\n",
       " 'decathlon',\n",
       " 'houseguest',\n",
       " 'shackling',\n",
       " 'plushily',\n",
       " 'electable',\n",
       " 'silberman',\n",
       " 'hernandezs',\n",
       " 'franois',\n",
       " 'babbling',\n",
       " 'inexpressible',\n",
       " 'burgeoning',\n",
       " 'atwaters',\n",
       " 'unrated',\n",
       " 'ific',\n",
       " 'virtuosity',\n",
       " 'valentinos',\n",
       " 'rebuilds',\n",
       " 'dempseys',\n",
       " 'eggnogged',\n",
       " 'spielbergian',\n",
       " 'komakur',\n",
       " 'demoliton',\n",
       " 'acquainting',\n",
       " 'hemispheres',\n",
       " 'upn',\n",
       " 'ophuls',\n",
       " 'roles',\n",
       " 'noimosynis',\n",
       " 'testy',\n",
       " 'mufti',\n",
       " 'frameworks',\n",
       " 'bank',\n",
       " 'sours',\n",
       " 'kreskin',\n",
       " 'headings',\n",
       " 'excusable',\n",
       " 'conceit',\n",
       " 'throaty',\n",
       " 'dacosta',\n",
       " 'toscas',\n",
       " 'infertile',\n",
       " 'passable',\n",
       " 'cyril',\n",
       " 'disclaimer',\n",
       " 'quarry',\n",
       " 'prohibited',\n",
       " 'shacking',\n",
       " 'tripeline',\n",
       " 'offside',\n",
       " 'moaz',\n",
       " 'strangely',\n",
       " 'gavron',\n",
       " 'mcguigans',\n",
       " 'carefree',\n",
       " 'likova',\n",
       " 'putrification',\n",
       " 'mabel',\n",
       " 'creaminess',\n",
       " 'frissons',\n",
       " 'broderick',\n",
       " 'chateaubriand',\n",
       " 'sadden',\n",
       " 'pointseems',\n",
       " 'bounties',\n",
       " 'waitll',\n",
       " 'fasta',\n",
       " 'metamorphosis',\n",
       " 'praises',\n",
       " 'sloppiness',\n",
       " 'below',\n",
       " 'po',\n",
       " 'pasadena',\n",
       " 'kaui',\n",
       " 'rejtman',\n",
       " 'cavort',\n",
       " 'blotchy',\n",
       " 'srk',\n",
       " 'stevenss',\n",
       " 'incredulity',\n",
       " 'fleishers',\n",
       " 'courier',\n",
       " 'foster',\n",
       " 'parrs',\n",
       " 'boisterously',\n",
       " 'applicant',\n",
       " 'haim',\n",
       " 'pleasantly',\n",
       " 'bettanydunst',\n",
       " 'brosnahan',\n",
       " 'engorges',\n",
       " 'oneals',\n",
       " 'shakespearea',\n",
       " 'stigmatizing',\n",
       " 'azimi',\n",
       " 'ilsa',\n",
       " 'nostalghia',\n",
       " 'defas',\n",
       " 'blogger',\n",
       " 'yearenormously',\n",
       " 'cahoonas',\n",
       " 'fermata',\n",
       " 'ayanaba',\n",
       " 'levis',\n",
       " 'tensi',\n",
       " 'antically',\n",
       " 'leconte',\n",
       " 'sangfroid',\n",
       " 'agallas',\n",
       " 'qualley',\n",
       " 'jonbent',\n",
       " 'inexcusably',\n",
       " 'goons',\n",
       " 'dragons',\n",
       " 'louganis',\n",
       " 'morocco',\n",
       " 'experi',\n",
       " 'conservative',\n",
       " 'bings',\n",
       " 'irans',\n",
       " 'soullessness',\n",
       " 'sauerkraut',\n",
       " 'mcthis',\n",
       " 'varga',\n",
       " 'erection',\n",
       " 'mugging',\n",
       " 'coppery',\n",
       " 'dissenters',\n",
       " 'treitz',\n",
       " 'working',\n",
       " 'obarry',\n",
       " 'guzmans',\n",
       " 'aswell',\n",
       " 'impressingly',\n",
       " 'entertain',\n",
       " 'correct',\n",
       " 'enlarge',\n",
       " 'enshrouded',\n",
       " 'dating',\n",
       " 'share',\n",
       " 'daftness',\n",
       " 'appeasing',\n",
       " 'vibrates',\n",
       " 'overawed',\n",
       " 'seaboarder',\n",
       " 'microcosmos',\n",
       " 'utonium',\n",
       " 'yeoh',\n",
       " 'emma',\n",
       " 'akashs',\n",
       " 'assumindo',\n",
       " 'unsynchronised',\n",
       " 'hangar',\n",
       " 'mismatched',\n",
       " 'battaglia',\n",
       " 'concur',\n",
       " 'cluedo',\n",
       " 'bluebell',\n",
       " 'scofield',\n",
       " 'bharat',\n",
       " 'maui',\n",
       " 'sharkey',\n",
       " 'mage',\n",
       " 'grimault',\n",
       " 'compasin',\n",
       " 'golfs',\n",
       " 'cryogenic',\n",
       " 'assess',\n",
       " 'dybbuk',\n",
       " 'workmanship',\n",
       " 'angled',\n",
       " 'sight',\n",
       " 'detest',\n",
       " 'tratado',\n",
       " 'bosworths',\n",
       " 'myjust',\n",
       " 'dramafrom',\n",
       " 'sidewinding',\n",
       " 'trane',\n",
       " 'hales',\n",
       " 'gravitate',\n",
       " 'tweedledee',\n",
       " 'unflappable',\n",
       " 'fragonard',\n",
       " 'landmines',\n",
       " 'eisley',\n",
       " 'rp',\n",
       " 'fictionalize',\n",
       " 'deees',\n",
       " 'rough',\n",
       " 'ei',\n",
       " 'martians',\n",
       " 'melissas',\n",
       " 'mandatory',\n",
       " 'burning',\n",
       " 'equalize',\n",
       " 'stellas',\n",
       " 'pare',\n",
       " 'inkheart',\n",
       " 'saleems',\n",
       " 'tacks',\n",
       " 'kimya',\n",
       " 'spears',\n",
       " 'rooftops',\n",
       " 'nudging',\n",
       " 'trophies',\n",
       " 'underdogfamily',\n",
       " 'privately',\n",
       " 'fittingly',\n",
       " 'goormaghtigh',\n",
       " 'microfilm',\n",
       " 'dweezil',\n",
       " 'acostas',\n",
       " 'inhabituelle',\n",
       " 'mythologize',\n",
       " 'kuso',\n",
       " 'dick',\n",
       " 'mule',\n",
       " 'nards',\n",
       " 'milewski',\n",
       " 'joisey',\n",
       " 'jesting',\n",
       " 'belter',\n",
       " 'grafts',\n",
       " 'psps',\n",
       " 'diterlizzi',\n",
       " 'instrumen',\n",
       " 'moreish',\n",
       " 'llamativos',\n",
       " 'everybody',\n",
       " 'credence',\n",
       " 'raffertys',\n",
       " 'maximise',\n",
       " 'clifton',\n",
       " 'klieg',\n",
       " 'junked',\n",
       " 'decorous',\n",
       " 'kondabolu',\n",
       " 'euphorically',\n",
       " 'dampening',\n",
       " 'horowitz',\n",
       " 'statesman',\n",
       " 'shitman',\n",
       " 'chappies',\n",
       " 'youits',\n",
       " 'appropriateness',\n",
       " 'reevaluate',\n",
       " 'prominentlyas',\n",
       " 'celebrat',\n",
       " 'mastani',\n",
       " 'moviethis',\n",
       " 'esquec',\n",
       " 'ubiquity',\n",
       " 'marinis',\n",
       " 'supplies',\n",
       " 'hitter',\n",
       " 'grandmothers',\n",
       " 'sponge',\n",
       " 'guionista',\n",
       " 'prolix',\n",
       " 'miroslav',\n",
       " 'gems',\n",
       " 'krybei',\n",
       " 'efibiko',\n",
       " 'yousafzais',\n",
       " 'lapoirie',\n",
       " 'technosceptic',\n",
       " 'homeliness',\n",
       " 'thumper',\n",
       " 'mohawk',\n",
       " 'perfection',\n",
       " 'numa',\n",
       " 'tussauds',\n",
       " 'formulatic',\n",
       " 'marlboro',\n",
       " 'abernathy',\n",
       " 'droplet',\n",
       " 'splutter',\n",
       " 'bacri',\n",
       " 'gbh',\n",
       " 'elaborate',\n",
       " 'studying',\n",
       " 'daldry',\n",
       " 'pabsts',\n",
       " 'triples',\n",
       " 'bankability',\n",
       " 'fitful',\n",
       " 'domhnall',\n",
       " 'ragnarok',\n",
       " 'bypass',\n",
       " 'rael',\n",
       " 'bruisingly',\n",
       " 'roguishly',\n",
       " 'resorts',\n",
       " 'bulgariagreece',\n",
       " 'liciosa',\n",
       " 'satisfactorily',\n",
       " 'hopkung',\n",
       " 'mordant',\n",
       " 'crackpots',\n",
       " 'friedman',\n",
       " 'admires',\n",
       " 'archness',\n",
       " 'spinet',\n",
       " 'knottiness',\n",
       " 'gann',\n",
       " 'derren',\n",
       " 'jordin',\n",
       " 'senos',\n",
       " 'sth',\n",
       " 'debut',\n",
       " 'roachs',\n",
       " 'godawfully',\n",
       " 'transport',\n",
       " 'jew',\n",
       " 'kriges',\n",
       " 'parmets',\n",
       " 'deveria',\n",
       " 'varma',\n",
       " 'exigencies',\n",
       " 'iconic',\n",
       " 'jeeva',\n",
       " 'ladkani',\n",
       " 'pantomime',\n",
       " 'repetitivea',\n",
       " 'horsed',\n",
       " 'interrogator',\n",
       " 'bawdiest',\n",
       " 'intensamente',\n",
       " 'jepsen',\n",
       " 'scratchingly',\n",
       " 'dashcam',\n",
       " 'brannagh',\n",
       " 'deeply',\n",
       " 'buellers',\n",
       " 'nsync',\n",
       " 'amaretto',\n",
       " 'outdone',\n",
       " 'staying',\n",
       " 'wells',\n",
       " 'tragedians',\n",
       " 'undersized',\n",
       " 'timing',\n",
       " 'beastlyi',\n",
       " 'conking',\n",
       " 'fdrs',\n",
       " 'gents',\n",
       " 'repositioned',\n",
       " 'chintz',\n",
       " 'importantes',\n",
       " 'williamss',\n",
       " 'rachelellis',\n",
       " 'dearly',\n",
       " 'commies',\n",
       " 'coma',\n",
       " 'burnhams',\n",
       " 'rihnei',\n",
       " 'flaccidness',\n",
       " 'climatic',\n",
       " 'cantonese',\n",
       " 'archaism',\n",
       " 'fanmade',\n",
       " 'robitussin',\n",
       " 'encanta',\n",
       " 'spoofiness',\n",
       " 'seasonally',\n",
       " 'reveling',\n",
       " 'totaled',\n",
       " 'sluts',\n",
       " 'popcorniest',\n",
       " 'tangle',\n",
       " 'nombreuses',\n",
       " 'willard',\n",
       " 'frodos',\n",
       " 'stacking',\n",
       " 'souths',\n",
       " 'tearjerk',\n",
       " 'reach',\n",
       " 'masochistically',\n",
       " 'annoyances',\n",
       " 'forjaded',\n",
       " 'revire',\n",
       " 'legrand',\n",
       " 'determining',\n",
       " 'tethered',\n",
       " 'general',\n",
       " 'papier',\n",
       " 'bothe',\n",
       " 'graciosos',\n",
       " 'endow',\n",
       " 'aggressing',\n",
       " 'intelligent',\n",
       " 'aherns',\n",
       " 'energization',\n",
       " 'scope',\n",
       " 'tomcat',\n",
       " 'mulling',\n",
       " 'baena',\n",
       " 'cutouts',\n",
       " 'progress',\n",
       " 'troupes',\n",
       " 'levelling',\n",
       " 'taunted',\n",
       " 'polio',\n",
       " 'compassionate',\n",
       " 'loomed',\n",
       " 'guirados',\n",
       " 'vibrancies',\n",
       " 'broughton',\n",
       " 'prepackaged',\n",
       " 'designing',\n",
       " 'dre',\n",
       " 'inciters',\n",
       " 'textures',\n",
       " 'fortune',\n",
       " 'bile',\n",
       " 'anointed',\n",
       " 'ended',\n",
       " 'barre',\n",
       " 'blairs',\n",
       " 'dangerousveterans',\n",
       " 'sith',\n",
       " 'subcharacters',\n",
       " 'felled',\n",
       " 'ju',\n",
       " 'pyschosis',\n",
       " 'gelo',\n",
       " 'roadrunner',\n",
       " 'ocurrir',\n",
       " 'papeles',\n",
       " 'polgergeist',\n",
       " 'horrific',\n",
       " 'hc',\n",
       " 'plattitudes',\n",
       " 'toggling',\n",
       " 'phile',\n",
       " 'jills',\n",
       " 'zinnemann',\n",
       " 'bloodied',\n",
       " 'righted',\n",
       " 'outmatched',\n",
       " 'gift',\n",
       " 'emcee',\n",
       " 'skeevy',\n",
       " 'twig',\n",
       " 'trinians',\n",
       " 'explosiveness',\n",
       " 'ornament',\n",
       " 'kerpow',\n",
       " 'marmaduke',\n",
       " 'woro',\n",
       " 'hawaa',\n",
       " 'islamophobic',\n",
       " 'athletic',\n",
       " 'bernays',\n",
       " 'meron',\n",
       " 'laughsand',\n",
       " 'operativethan',\n",
       " 'porgnip',\n",
       " 'runrevenge',\n",
       " 'techniques',\n",
       " 'mesmeric',\n",
       " 'became',\n",
       " 'twi',\n",
       " 'goofus',\n",
       " 'toughest',\n",
       " 'recurrence',\n",
       " 'outguess',\n",
       " 'guess',\n",
       " 'investor',\n",
       " 'unobtrusively',\n",
       " 'coagulated',\n",
       " 'terrios',\n",
       " 'ponytail',\n",
       " 'christophersons',\n",
       " 'vacancy',\n",
       " 'row',\n",
       " 'pillage',\n",
       " 'nadzams',\n",
       " 'motherless',\n",
       " 'oprahfied',\n",
       " 'edgar',\n",
       " 'envelop',\n",
       " 'adventuretoy',\n",
       " 'consigo',\n",
       " 'hairdos',\n",
       " 'noriko',\n",
       " 'fractious',\n",
       " 'zeffirelli',\n",
       " 'scena',\n",
       " 'mongers',\n",
       " 'tesla',\n",
       " 'proclivity',\n",
       " 'abusing',\n",
       " 'wanes',\n",
       " 'pasado',\n",
       " 'fentons',\n",
       " 'gratifications',\n",
       " 'mutterer',\n",
       " 'decasia',\n",
       " 'punches',\n",
       " 'joans',\n",
       " 'vernica',\n",
       " 'absolved',\n",
       " 'lannister',\n",
       " 'eggshells',\n",
       " 'emblazoned',\n",
       " 'robertson',\n",
       " 'sprightly',\n",
       " 'magnified',\n",
       " 'contempornea',\n",
       " 'leduc',\n",
       " 'partridges',\n",
       " 'moronically',\n",
       " 'deceitful',\n",
       " 'dunnes',\n",
       " 'kraemers',\n",
       " 'ladylove',\n",
       " 'sybil',\n",
       " 'augmented',\n",
       " 'scorecard',\n",
       " 'auf',\n",
       " 'thurston',\n",
       " 'images',\n",
       " 'biennial',\n",
       " 'angling',\n",
       " 'dern',\n",
       " 'baruchel',\n",
       " 'pirates',\n",
       " 'prosperous',\n",
       " 'kurdish',\n",
       " 'toughness',\n",
       " 'levity',\n",
       " 'compounded',\n",
       " 'dissonance',\n",
       " 'mystified',\n",
       " 'hendersons',\n",
       " 'backtrace',\n",
       " 'miramax',\n",
       " 'pratts',\n",
       " 'dismantles',\n",
       " 'artiest',\n",
       " 'grounded',\n",
       " 'nutcases',\n",
       " 'norquists',\n",
       " 'civilians',\n",
       " 'desperate',\n",
       " 'reckless',\n",
       " 'presidencythat',\n",
       " 'nosey',\n",
       " 'costs',\n",
       " 'mentor',\n",
       " 'marianis',\n",
       " 'leukocytes',\n",
       " 'zeus',\n",
       " 'martinessis',\n",
       " 'maynard',\n",
       " 'yukinoj',\n",
       " 'brezhnev',\n",
       " 'overcame',\n",
       " 'u',\n",
       " 'bulgakov',\n",
       " 'candysteve',\n",
       " 'anvils',\n",
       " 'schwarzeneggerian',\n",
       " 'important',\n",
       " 'basement',\n",
       " 'vulnerabilidade',\n",
       " 'gimmicky',\n",
       " 'wander',\n",
       " 'mariner',\n",
       " 'nonsensically',\n",
       " 'hreiazetai',\n",
       " 'ritualistic',\n",
       " ...}"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vocabulary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "54ddfa7e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'fresh': {'marked': 81,\n",
       "  'by': 11033,\n",
       "  'a': 109093,\n",
       "  'tossed': 30,\n",
       "  'off': 2281,\n",
       "  'feeling': 760,\n",
       "  'distinct': 85,\n",
       "  'impression': 155,\n",
       "  'that': 34330,\n",
       "  'americas': 175,\n",
       "  'most': 6947,\n",
       "  'viable': 17,\n",
       "  'auteur': 91,\n",
       "  'brats': 6,\n",
       "  'had': 1359,\n",
       "  'taken': 440,\n",
       "  'an': 19259,\n",
       "  'unfamiliar': 69,\n",
       "  'path': 195,\n",
       "  'with': 26689,\n",
       "  'minimal': 52,\n",
       "  'investment': 59,\n",
       "  'for': 22086,\n",
       "  'uncertain': 47,\n",
       "  'gain': 52,\n",
       "  'all': 9273,\n",
       "  'the': 151903,\n",
       "  'classic': 1360,\n",
       "  'moments': 1956,\n",
       "  'are': 9764,\n",
       "  'there': 3848,\n",
       "  'crammed': 50,\n",
       "  'in': 44853,\n",
       "  'as': 25987,\n",
       "  'if': 7346,\n",
       "  'we': 3889,\n",
       "  'might': 2258,\n",
       "  'not': 11430,\n",
       "  'get': 2539,\n",
       "  'another': 1762,\n",
       "  'shot': 957,\n",
       "  'at': 10780,\n",
       "  'this': 21794,\n",
       "  'its': 36544,\n",
       "  'very': 3901,\n",
       "  'sweet': 1102,\n",
       "  'and': 98971,\n",
       "  'charming': 869,\n",
       "  'should': 2102,\n",
       "  'be': 13143,\n",
       "  'thankful': 23,\n",
       "  'isnt': 2152,\n",
       "  'childhood': 216,\n",
       "  'ruining': 8,\n",
       "  'disaster': 222,\n",
       "  'mulholland': 12,\n",
       "  'drive': 222,\n",
       "  'is': 55220,\n",
       "  'maddeningly': 25,\n",
       "  'abstract': 59,\n",
       "  'inscrutable': 22,\n",
       "  'on': 14766,\n",
       "  'first': 3255,\n",
       "  'viewing': 466,\n",
       "  'but': 24895,\n",
       "  'hypnotic': 159,\n",
       "  'haunting': 431,\n",
       "  'every': 2227,\n",
       "  'after': 1646,\n",
       "  'jason': 201,\n",
       "  'reitman': 60,\n",
       "  'does': 3176,\n",
       "  'it': 33809,\n",
       "  'again': 1285,\n",
       "  'although': 957,\n",
       "  'full': 4881,\n",
       "  'of': 91787,\n",
       "  'product': 160,\n",
       "  'placements': 2,\n",
       "  'dark': 1473,\n",
       "  'comedy': 4254,\n",
       "  'surprisingly': 911,\n",
       "  'enjoyable': 1350,\n",
       "  'highlighted': 29,\n",
       "  'anna': 104,\n",
       "  'kendrick': 37,\n",
       "  'who': 6115,\n",
       "  'made': 3017,\n",
       "  'me': 1678,\n",
       "  'laugh': 728,\n",
       "  'out': 5826,\n",
       "  'loud': 385,\n",
       "  'ambitious': 468,\n",
       "  'passionate': 179,\n",
       "  'grief': 254,\n",
       "  'stricken': 17,\n",
       "  'work': 3530,\n",
       "  'film': 21677,\n",
       "  'art': 1338,\n",
       "  'energetic': 215,\n",
       "  'slapstick': 219,\n",
       "  'aimed': 108,\n",
       "  'indiscriminating': 1,\n",
       "  'youngsters': 50,\n",
       "  'will': 5848,\n",
       "  'want': 1220,\n",
       "  'to': 58174,\n",
       "  'buy': 105,\n",
       "  'lots': 296,\n",
       "  'crood': 1,\n",
       "  'toys': 83,\n",
       "  'one': 11718,\n",
       "  'van': 188,\n",
       "  'dammes': 8,\n",
       "  'best': 5630,\n",
       "  'flicksgood': 1,\n",
       "  'story': 7620,\n",
       "  'good': 5780,\n",
       "  'skirmishes': 7,\n",
       "  'masterfully': 93,\n",
       "  'directed': 768,\n",
       "  'xavier': 32,\n",
       "  'beauvois': 13,\n",
       "  'hands': 366,\n",
       "  'down': 1624,\n",
       "  'euthanasia': 4,\n",
       "  'ever': 2248,\n",
       "  'films': 5164,\n",
       "  'humor': 1533,\n",
       "  'sauce': 7,\n",
       "  'elevate': 140,\n",
       "  'from': 10224,\n",
       "  'whining': 5,\n",
       "  'self': 1314,\n",
       "  'absorbed': 34,\n",
       "  'unattached': 2,\n",
       "  'women': 790,\n",
       "  'populate': 23,\n",
       "  'too': 3431,\n",
       "  'many': 2524,\n",
       "  'books': 181,\n",
       "  'today': 369,\n",
       "  'easily': 710,\n",
       "  'ryan': 188,\n",
       "  'reynolds': 100,\n",
       "  'performance': 3341,\n",
       "  'since': 1103,\n",
       "  'buried': 62,\n",
       "  'ben': 218,\n",
       "  'mendelsohn': 18,\n",
       "  'portrays': 111,\n",
       "  'flawed': 357,\n",
       "  'helpless': 21,\n",
       "  'individuals': 115,\n",
       "  'so': 6785,\n",
       "  'incredibly': 293,\n",
       "  'well': 5395,\n",
       "  'when': 3944,\n",
       "  'you': 12675,\n",
       "  'take': 2044,\n",
       "  'into': 6028,\n",
       "  'consideration': 62,\n",
       "  'mississippi': 11,\n",
       "  'grind': 30,\n",
       "  'addictive': 32,\n",
       "  'gambling': 12,\n",
       "  'storyline': 186,\n",
       "  'involving': 308,\n",
       "  'visually': 667,\n",
       "  'stunning': 656,\n",
       "  'ive': 569,\n",
       "  'seen': 1695,\n",
       "  'insidious': 72,\n",
       "  'chapter': 203,\n",
       "  'lack': 428,\n",
       "  'unique': 660,\n",
       "  'charm': 767,\n",
       "  'both': 3016,\n",
       "  'predecessors': 141,\n",
       "  'still': 3493,\n",
       "  'scary': 518,\n",
       "  'fun': 3725,\n",
       "  'thrill': 316,\n",
       "  'ride': 865,\n",
       "  'delivers': 1246,\n",
       "  'some': 5198,\n",
       "  'serious': 610,\n",
       "  'blood': 519,\n",
       "  'pressure': 69,\n",
       "  'raising': 84,\n",
       "  'tension': 683,\n",
       "  'intriguing': 631,\n",
       "  'tale': 2291,\n",
       "  'about': 9628,\n",
       "  'individual': 194,\n",
       "  'whose': 926,\n",
       "  'mounting': 35,\n",
       "  'debt': 38,\n",
       "  'leads': 523,\n",
       "  'unfortunate': 41,\n",
       "  'circumstances': 121,\n",
       "  'thai': 18,\n",
       "  'arthouse': 86,\n",
       "  'director': 4169,\n",
       "  'apichatpong': 7,\n",
       "  'weerasethakul': 9,\n",
       "  'uncle': 52,\n",
       "  'boonmee': 6,\n",
       "  'won': 160,\n",
       "  'palme': 14,\n",
       "  'dor': 17,\n",
       "  'cannes': 68,\n",
       "  'few': 1678,\n",
       "  'years': 2198,\n",
       "  'ago': 215,\n",
       "  'comes': 1388,\n",
       "  'strange': 546,\n",
       "  'swooning': 18,\n",
       "  'dream': 479,\n",
       "  'cemetery': 9,\n",
       "  'splendour': 10,\n",
       "  'engaging': 1140,\n",
       "  'exciting': 799,\n",
       "  'sci': 722,\n",
       "  'fi': 746,\n",
       "  'flick': 768,\n",
       "  'bring': 453,\n",
       "  'year': 2279,\n",
       "  'old': 2187,\n",
       "  'us': 3897,\n",
       "  'while': 3970,\n",
       "  'find': 1415,\n",
       "  'jarmuschs': 34,\n",
       "  'can': 4902,\n",
       "  'call': 527,\n",
       "  'dialogue': 894,\n",
       "  'poor': 122,\n",
       "  'pretentious': 83,\n",
       "  'others': 526,\n",
       "  'adore': 37,\n",
       "  'artistic': 277,\n",
       "  'sensibilities': 86,\n",
       "  'war': 1482,\n",
       "  'stalemate': 2,\n",
       "  'aleksandr': 4,\n",
       "  'burovs': 1,\n",
       "  'bleached': 12,\n",
       "  'images': 483,\n",
       "  'creating': 282,\n",
       "  'alien': 246,\n",
       "  'landscape': 227,\n",
       "  'which': 4190,\n",
       "  'colour': 90,\n",
       "  'rare': 942,\n",
       "  'compassion': 220,\n",
       "  'rarely': 477,\n",
       "  'has': 9315,\n",
       "  'combat': 96,\n",
       "  'seemed': 132,\n",
       "  'savage': 73,\n",
       "  'or': 5726,\n",
       "  'futile': 28,\n",
       "  'nick': 99,\n",
       "  'morans': 3,\n",
       "  'based': 662,\n",
       "  'stage': 344,\n",
       "  'play': 894,\n",
       "  'he': 3899,\n",
       "  'wrote': 143,\n",
       "  'james': 581,\n",
       "  'hicks': 7,\n",
       "  'eccentric': 174,\n",
       "  'sometimes': 1320,\n",
       "  'underpowered': 1,\n",
       "  'always': 1426,\n",
       "  'watchable': 415,\n",
       "  'early': 514,\n",
       "  's': 1685,\n",
       "  'prehistory': 3,\n",
       "  'pop': 535,\n",
       "  'culture': 695,\n",
       "  'part': 1527,\n",
       "  'screenwriters': 59,\n",
       "  'never': 3275,\n",
       "  'lose': 138,\n",
       "  'sight': 250,\n",
       "  'moral': 506,\n",
       "  'whatever': 279,\n",
       "  'else': 610,\n",
       "  'gets': 1371,\n",
       "  'mix': 434,\n",
       "  'shrek': 84,\n",
       "  'must': 842,\n",
       "  'heart': 2248,\n",
       "  'soul': 580,\n",
       "  'myers': 41,\n",
       "  'master': 406,\n",
       "  'makes': 3785,\n",
       "  'seem': 764,\n",
       "  'easy': 1018,\n",
       "  'being': 2203,\n",
       "  'green': 223,\n",
       "  'doesnt': 2899,\n",
       "  'match': 248,\n",
       "  'high': 1660,\n",
       "  'bar': 137,\n",
       "  'set': 1369,\n",
       "  'royale': 40,\n",
       "  'neither': 255,\n",
       "  'let': 689,\n",
       "  'franchise': 751,\n",
       "  'fall': 400,\n",
       "  'unduly': 5,\n",
       "  'far': 1861,\n",
       "  'mark': 381,\n",
       "  'appealingly': 29,\n",
       "  'cheesy': 153,\n",
       "  'tribute': 400,\n",
       "  'hope': 784,\n",
       "  'springs': 38,\n",
       "  'eternal': 66,\n",
       "  'hopelessly': 29,\n",
       "  'inept': 16,\n",
       "  'such': 1959,\n",
       "  'photogenic': 16,\n",
       "  'charismatic': 222,\n",
       "  'cast': 2398,\n",
       "  'how': 3613,\n",
       "  'could': 2544,\n",
       "  'earth': 288,\n",
       "  'miss': 397,\n",
       "  'quick': 161,\n",
       "  'show': 1029,\n",
       "  'characters': 3521,\n",
       "  'skin': 252,\n",
       "  'less': 1540,\n",
       "  'inclined': 40,\n",
       "  'explore': 133,\n",
       "  'what': 6534,\n",
       "  'lies': 310,\n",
       "  'beneath': 276,\n",
       "  'those': 2694,\n",
       "  'saw': 206,\n",
       "  'animated': 875,\n",
       "  'version': 680,\n",
       "  'musical': 704,\n",
       "  'numbers': 180,\n",
       "  'yet': 2484,\n",
       "  'favreau': 53,\n",
       "  'manages': 1102,\n",
       "  'create': 443,\n",
       "  'beautiful': 1381,\n",
       "  'balanced': 119,\n",
       "  'review': 3537,\n",
       "  'spanish': 3206,\n",
       "  'new': 3119,\n",
       "  'hur': 6,\n",
       "  'impressive': 842,\n",
       "  'epic': 745,\n",
       "  'memory': 403,\n",
       "  'charlton': 4,\n",
       "  'heston': 5,\n",
       "  'chariot': 4,\n",
       "  'race': 299,\n",
       "  'remains': 702,\n",
       "  'lanthimoss': 7,\n",
       "  'terrifically': 95,\n",
       "  'twisted': 222,\n",
       "  'satire': 571,\n",
       "  'takes': 1671,\n",
       "  'places': 329,\n",
       "  'fear': 355,\n",
       "  'tread': 18,\n",
       "  'symbolism': 45,\n",
       "  'humanized': 6,\n",
       "  'top': 945,\n",
       "  'flight': 122,\n",
       "  'ending': 650,\n",
       "  'brutal': 342,\n",
       "  'rapturously': 13,\n",
       "  'romantic': 1240,\n",
       "  'same': 1342,\n",
       "  'time': 4835,\n",
       "  'despite': 1381,\n",
       "  'shortcomings': 127,\n",
       "  'happy': 540,\n",
       "  'feet': 165,\n",
       "  'brings': 686,\n",
       "  'lot': 1723,\n",
       "  'energy': 659,\n",
       "  'creativity': 133,\n",
       "  'genre': 1971,\n",
       "  'increasingly': 205,\n",
       "  'relies': 115,\n",
       "  'over': 2634,\n",
       "  'stimulation': 7,\n",
       "  'blythe': 7,\n",
       "  'danner': 10,\n",
       "  'lands': 63,\n",
       "  'her': 3952,\n",
       "  'screen': 1811,\n",
       "  'role': 1021,\n",
       "  'retired': 14,\n",
       "  'widow': 9,\n",
       "  'decides': 27,\n",
       "  'dip': 22,\n",
       "  'tentative': 24,\n",
       "  'toe': 44,\n",
       "  'dating': 29,\n",
       "  'pool': 45,\n",
       "  'indie': 354,\n",
       "  'ill': 212,\n",
       "  'see': 3147,\n",
       "  'my': 1269,\n",
       "  'dreams': 288,\n",
       "  'truth': 541,\n",
       "  'beauty': 742,\n",
       "  'theyre': 491,\n",
       "  'necessarily': 161,\n",
       "  'nine': 51,\n",
       "  'committed': 184,\n",
       "  'relationship': 537,\n",
       "  'two': 2810,\n",
       "  'kids': 1017,\n",
       "  'backseat': 7,\n",
       "  'moulin': 16,\n",
       "  'rouge': 22,\n",
       "  'le': 118,\n",
       "  'queda': 3,\n",
       "  'chica': 2,\n",
       "  'la': 501,\n",
       "  'pantalla': 4,\n",
       "  'no': 3985,\n",
       "  'slo': 13,\n",
       "  'satura': 1,\n",
       "  'desborda': 2,\n",
       "  'es': 86,\n",
       "  'una': 109,\n",
       "  'fiesta': 5,\n",
       "  'de': 1076,\n",
       "  'color': 250,\n",
       "  'msica': 2,\n",
       "  'heartwarming': 155,\n",
       "  'message': 634,\n",
       "  'combined': 99,\n",
       "  'creative': 476,\n",
       "  'fantastical': 79,\n",
       "  'elements': 599,\n",
       "  'make': 3920,\n",
       "  'memorable': 444,\n",
       "  'stardust': 21,\n",
       "  'just': 5432,\n",
       "  'note': 302,\n",
       "  'movie': 13582,\n",
       "  'pizzazz': 20,\n",
       "  'zero': 89,\n",
       "  'substance': 186,\n",
       "  'more': 9369,\n",
       "  'thoughtful': 554,\n",
       "  'motion': 362,\n",
       "  'picture': 1313,\n",
       "  'ponders': 13,\n",
       "  'mysteries': 92,\n",
       "  'life': 4260,\n",
       "  'way': 4127,\n",
       "  'accessible': 186,\n",
       "  'emotionally': 701,\n",
       "  'satisfying': 869,\n",
       "  'give': 1012,\n",
       "  'tati': 11,\n",
       "  'his': 9920,\n",
       "  'posthumous': 10,\n",
       "  'chance': 331,\n",
       "  'express': 134,\n",
       "  'sad': 444,\n",
       "  'may': 3617,\n",
       "  'have': 7115,\n",
       "  'been': 2856,\n",
       "  'chomets': 5,\n",
       "  'only': 3133,\n",
       "  'true': 1531,\n",
       "  'course': 375,\n",
       "  'honest': 500,\n",
       "  'beautifully': 1011,\n",
       "  'thoughtfully': 40,\n",
       "  'realized': 199,\n",
       "  'uma': 71,\n",
       "  'daquelas': 3,\n",
       "  'obras': 5,\n",
       "  'que': 312,\n",
       "  'por': 75,\n",
       "  'mais': 46,\n",
       "  'eu': 3,\n",
       "  'tenha': 1,\n",
       "  'apreciado': 1,\n",
       "  'sei': 1,\n",
       "  'jamais': 18,\n",
       "  'voltarei': 2,\n",
       "  'assistir': 3,\n",
       "  'e': 240,\n",
       "  'isto': 6,\n",
       "  'acreditem': 1,\n",
       "  'um': 124,\n",
       "  'elo': 3,\n",
       "  'al': 117,\n",
       "  'daradji': 1,\n",
       "  'directs': 119,\n",
       "  'sure': 922,\n",
       "  'hand': 548,\n",
       "  'avoiding': 50,\n",
       "  'icy': 39,\n",
       "  'realism': 300,\n",
       "  'chest': 38,\n",
       "  'beating': 88,\n",
       "  'melodrama': 488,\n",
       "  'result': 643,\n",
       "  'warm': 588,\n",
       "  'human': 1764,\n",
       "  'quietly': 367,\n",
       "  'devastating': 260,\n",
       "  'day': 1137,\n",
       "  'turns': 1044,\n",
       "  'enjoying': 89,\n",
       "  'traditional': 282,\n",
       "  'than': 8080,\n",
       "  'admonishment': 4,\n",
       "  'stop': 413,\n",
       "  'wasting': 11,\n",
       "  'business': 285,\n",
       "  'living': 382,\n",
       "  'enjoy': 895,\n",
       "  'single': 439,\n",
       "  'moment': 791,\n",
       "  'ones': 615,\n",
       "  'love': 3499,\n",
       "  'anyone': 745,\n",
       "  'interest': 407,\n",
       "  'dance': 293,\n",
       "  'documentary': 2453,\n",
       "  'among': 556,\n",
       "  'holocaust': 109,\n",
       "  'agnieszka': 5,\n",
       "  'hollands': 4,\n",
       "  'academy': 106,\n",
       "  'award': 167,\n",
       "  'nominated': 95,\n",
       "  'darkness': 215,\n",
       "  'during': 528,\n",
       "  'world': 3021,\n",
       "  'ii': 230,\n",
       "  'small': 952,\n",
       "  'town': 225,\n",
       "  'poland': 11,\n",
       "  'named': 108,\n",
       "  'lvov': 2,\n",
       "  'now': 1202,\n",
       "  'ukraine': 5,\n",
       "  'propulsive': 53,\n",
       "  'punk': 109,\n",
       "  'rock': 488,\n",
       "  'soundtrack': 269,\n",
       "  'rough': 214,\n",
       "  'cinematography': 415,\n",
       "  'dragonslayer': 6,\n",
       "  'care': 462,\n",
       "  'scrawny': 4,\n",
       "  'young': 1452,\n",
       "  'man': 2539,\n",
       "  'skating': 16,\n",
       "  'nowhere': 94,\n",
       "  'viewers': 868,\n",
       "  'wont': 828,\n",
       "  'feel': 2189,\n",
       "  'like': 8219,\n",
       "  'theyve': 160,\n",
       "  'dealt': 20,\n",
       "  'cruelly': 19,\n",
       "  'pirates': 65,\n",
       "  'defining': 73,\n",
       "  'wondermentbut': 1,\n",
       "  'whole': 967,\n",
       "  'tap': 77,\n",
       "  'wish': 387,\n",
       "  'were': 1524,\n",
       "  'pike': 41,\n",
       "  'gives': 1459,\n",
       "  'thoroughly': 451,\n",
       "  'convincing': 296,\n",
       "  'colvin': 17,\n",
       "  'drama': 3495,\n",
       "  'though': 2225,\n",
       "  'dimensional': 97,\n",
       "  'filmmakers': 793,\n",
       "  'focused': 196,\n",
       "  'they': 3089,\n",
       "  'dont': 1986,\n",
       "  'notice': 99,\n",
       "  'anything': 1027,\n",
       "  'contrived': 151,\n",
       "  'wonder': 592,\n",
       "  'wheel': 56,\n",
       "  'kate': 89,\n",
       "  'winslet': 35,\n",
       "  'pluses': 5,\n",
       "  'plenty': 858,\n",
       "  'laughs': 894,\n",
       "  'outrageous': 178,\n",
       "  'situations': 201,\n",
       "  'guys': 279,\n",
       "  'themselves': 474,\n",
       "  'lawrence': 116,\n",
       "  'luxuriates': 6,\n",
       "  'tremendous': 158,\n",
       "  'hits': 430,\n",
       "  'spot': 196,\n",
       "  'entertaining': 2930,\n",
       "  'informative': 118,\n",
       "  'deeply': 832,\n",
       "  'moving': 1478,\n",
       "  'adult': 392,\n",
       "  'beginners': 15,\n",
       "  'feels': 1866,\n",
       "  'baby': 231,\n",
       "  'step': 366,\n",
       "  'figuring': 22,\n",
       "  'kroll': 8,\n",
       "  'transition': 72,\n",
       "  'star': 1532,\n",
       "  'potential': 309,\n",
       "  'certainly': 1003,\n",
       "  'ways': 858,\n",
       "  'unremarkable': 35,\n",
       "  'rainmaker': 1,\n",
       "  'other': 2111,\n",
       "  'astonishing': 262,\n",
       "  'complex': 839,\n",
       "  'dismissed': 39,\n",
       "  'riotously': 25,\n",
       "  'funny': 3645,\n",
       "  'extremely': 411,\n",
       "  'clever': 854,\n",
       "  'abundantly': 19,\n",
       "  'given': 642,\n",
       "  'monumental': 49,\n",
       "  'task': 94,\n",
       "  'these': 1846,\n",
       "  'managed': 142,\n",
       "  'tell': 506,\n",
       "  'grow': 150,\n",
       "  'band': 198,\n",
       "  'fascinating': 1273,\n",
       "  'tactfully': 4,\n",
       "  'charts': 44,\n",
       "  'sort': 672,\n",
       "  'philosophical': 140,\n",
       "  'differences': 61,\n",
       "  'eventually': 225,\n",
       "  'undermine': 14,\n",
       "  'any': 2538,\n",
       "  'group': 245,\n",
       "  'endeavor': 53,\n",
       "  'even': 5289,\n",
       "  'seemingly': 248,\n",
       "  'minded': 162,\n",
       "  'collaborators': 31,\n",
       "  'salesman': 28,\n",
       "  'meticulously': 102,\n",
       "  'plotted': 83,\n",
       "  'puzzle': 141,\n",
       "  'once': 1360,\n",
       "  'shows': 940,\n",
       "  'great': 4082,\n",
       "  'transcend': 71,\n",
       "  'restrained': 138,\n",
       "  'macro': 9,\n",
       "  'marvel': 532,\n",
       "  'wisely': 87,\n",
       "  'marketed': 12,\n",
       "  'fry': 8,\n",
       "  'closer': 173,\n",
       "  'antz': 13,\n",
       "  'mutant': 39,\n",
       "  'radioactive': 3,\n",
       "  'variety': 107,\n",
       "  'them': 2288,\n",
       "  'would': 2171,\n",
       "  'perfect': 1339,\n",
       "  'damned': 69,\n",
       "  'annoying': 70,\n",
       "  'mango': 1,\n",
       "  'tree': 71,\n",
       "  'song': 294,\n",
       "  'submarine': 35,\n",
       "  'dads': 20,\n",
       "  'simple': 833,\n",
       "  'headed': 109,\n",
       "  'enough': 3261,\n",
       "  'hit': 577,\n",
       "  'quite': 1892,\n",
       "  'ludicrous': 68,\n",
       "  'pulls': 236,\n",
       "  'mills': 27,\n",
       "  'last': 1371,\n",
       "  'gave': 99,\n",
       "  'semi': 105,\n",
       "  'autobiographical': 59,\n",
       "  'treatment': 169,\n",
       "  'dad': 78,\n",
       "  'draws': 160,\n",
       "  'strong': 1389,\n",
       "  'memories': 175,\n",
       "  'rich': 730,\n",
       "  'imagination': 406,\n",
       "  'recreate': 22,\n",
       "  'santa': 70,\n",
       "  'barbara': 30,\n",
       "  'circa': 23,\n",
       "  'denizens': 13,\n",
       "  'wonderful': 733,\n",
       "  'girls': 376,\n",
       "  'much': 4739,\n",
       "  'their': 4662,\n",
       "  'own': 2519,\n",
       "  'people': 2168,\n",
       "  'tough': 388,\n",
       "  'categorize': 11,\n",
       "  'unlike': 314,\n",
       "  'youve': 535,\n",
       "  'before': 1341,\n",
       "  'completely': 573,\n",
       "  'recognizable': 65,\n",
       "  'forceful': 51,\n",
       "  'amount': 307,\n",
       "  'wordless': 17,\n",
       "  'transmissions': 2,\n",
       "  'pain': 314,\n",
       "  'longing': 108,\n",
       "  'across': 439,\n",
       "  'dinner': 85,\n",
       "  'tables': 11,\n",
       "  'supermarket': 5,\n",
       "  'counters': 12,\n",
       "  'tuned': 48,\n",
       "  'kumas': 1,\n",
       "  'wavelength': 24,\n",
       "  'privileged': 29,\n",
       "  'pick': 123,\n",
       "  'up': 6325,\n",
       "  'bit': 1635,\n",
       "  'overcooked': 10,\n",
       "  'comfort': 159,\n",
       "  'results': 261,\n",
       "  'pretty': 1094,\n",
       "  'least': 1129,\n",
       "  'closing': 100,\n",
       "  'emotional': 1732,\n",
       "  'doors': 42,\n",
       "  'scarred': 21,\n",
       "  'bereaved': 2,\n",
       "  'night': 671,\n",
       "  'death': 713,\n",
       "  'called': 262,\n",
       "  'country': 373,\n",
       "  'pub': 11,\n",
       "  'taut': 241,\n",
       "  'direction': 804,\n",
       "  'performances': 3205,\n",
       "  'impart': 17,\n",
       "  'terrifying': 289,\n",
       "  'sense': 2045,\n",
       "  'claustrophobia': 33,\n",
       "  'isolation': 66,\n",
       "  'mcgarrys': 2,\n",
       "  'extraordinary': 503,\n",
       "  'access': 71,\n",
       "  'illuminating': 107,\n",
       "  'perspective': 324,\n",
       "  'medical': 43,\n",
       "  'system': 244,\n",
       "  'powerful': 1280,\n",
       "  'experience': 1588,\n",
       "  'our': 2052,\n",
       "  'week': 131,\n",
       "  'saying': 172,\n",
       "  'arrives': 79,\n",
       "  'shape': 114,\n",
       "  'atmospheric': 199,\n",
       "  'slice': 235,\n",
       "  'americana': 27,\n",
       "  'arranged': 16,\n",
       "  'marriage': 234,\n",
       "  'deep': 691,\n",
       "  'rural': 87,\n",
       "  'minnesota': 6,\n",
       "  'why': 778,\n",
       "  'back': 1684,\n",
       "  'neck': 41,\n",
       "  'little': 3258,\n",
       "  'seems': 1273,\n",
       "  'paper': 108,\n",
       "  'i': 6192,\n",
       "  'suppose': 36,\n",
       "  'credit': 345,\n",
       "  'go': 1774,\n",
       "  'lieberstein': 2,\n",
       "  'gained': 18,\n",
       "  'intuitive': 20,\n",
       "  'successfully': 181,\n",
       "  'blend': 275,\n",
       "  'absurdity': 133,\n",
       "  'romance': 827,\n",
       "  'x': 219,\n",
       "  'shouldve': 19,\n",
       "  'ticket': 100,\n",
       "  'cold': 403,\n",
       "  'weather': 46,\n",
       "  'syrupy': 19,\n",
       "  'sentimental': 285,\n",
       "  'soap': 107,\n",
       "  'opera': 175,\n",
       "  'tailor': 43,\n",
       "  'fans': 1548,\n",
       "  'nicholas': 54,\n",
       "  'sparks': 75,\n",
       "  'hartigan': 2,\n",
       "  'touches': 245,\n",
       "  'loneliness': 160,\n",
       "  'fatherhood': 20,\n",
       "  'casual': 117,\n",
       "  'racism': 88,\n",
       "  'difficulty': 37,\n",
       "  'trying': 597,\n",
       "  'figure': 276,\n",
       "  'where': 1749,\n",
       "  'fit': 183,\n",
       "  'matter': 818,\n",
       "  'your': 3524,\n",
       "  'age': 1223,\n",
       "  'certain': 471,\n",
       "  'distance': 127,\n",
       "  'gilliams': 22,\n",
       "  'uncontrollable': 4,\n",
       "  'nonsense': 129,\n",
       "  'emerges': 162,\n",
       "  'subtle': 491,\n",
       "  'poem': 75,\n",
       "  'look': 2304,\n",
       "  'humans': 147,\n",
       "  'under': 738,\n",
       "  'inhuman': 11,\n",
       "  'stress': 45,\n",
       "  'nuanced': 257,\n",
       "  'almost': 1714,\n",
       "  'grand': 380,\n",
       "  'scale': 319,\n",
       "  'aurora': 4,\n",
       "  'proves': 768,\n",
       "  'puiu': 3,\n",
       "  'impressiona': 2,\n",
       "  'pelo': 12,\n",
       "  'ritmo': 5,\n",
       "  'pelas': 2,\n",
       "  'atuaes': 6,\n",
       "  'pela': 7,\n",
       "  'intensidade': 1,\n",
       "  'excepcional': 4,\n",
       "  'edio': 1,\n",
       "  'som': 2,\n",
       "  'revelando': 4,\n",
       "  'se': 128,\n",
       "  'sempre': 4,\n",
       "  'tenso': 6,\n",
       "  'nos': 19,\n",
       "  'melhores': 6,\n",
       "  'momentos': 13,\n",
       "  'absolutam': 1,\n",
       "  'leonera': 1,\n",
       "  'gritty': 296,\n",
       "  'earnest': 241,\n",
       "  'mixes': 112,\n",
       "  'breast': 6,\n",
       "  'feeding': 24,\n",
       "  'screaming': 57,\n",
       "  'infants': 3,\n",
       "  'familiar': 915,\n",
       "  'scenarios': 41,\n",
       "  'lesbian': 29,\n",
       "  'shower': 22,\n",
       "  'assaults': 8,\n",
       "  'strip': 65,\n",
       "  'searches': 10,\n",
       "  'bonfire': 5,\n",
       "  'riots': 15,\n",
       "  'friday': 72,\n",
       "  'lights': 141,\n",
       "  'success': 395,\n",
       "  'failure': 111,\n",
       "  'beloved': 147,\n",
       "  'team': 361,\n",
       "  'resonate': 103,\n",
       "  'mood': 449,\n",
       "  'townsfolk': 3,\n",
       "  'entire': 423,\n",
       "  'video': 306,\n",
       "  'lisbon': 4,\n",
       "  'revived': 15,\n",
       "  'transports': 25,\n",
       "  'audience': 1550,\n",
       "  'because': 1780,\n",
       "  'end': 1954,\n",
       "  'tour': 238,\n",
       "  'language': 357,\n",
       "  'men': 969,\n",
       "  'greatness': 166,\n",
       "  'achieved': 81,\n",
       "  'without': 2089,\n",
       "  'burtons': 106,\n",
       "  'creepy': 371,\n",
       "  'frankenstein': 29,\n",
       "  'tweens': 24,\n",
       "  'martin': 183,\n",
       "  'scorseses': 110,\n",
       "  'thrilling': 609,\n",
       "  'throbbing': 22,\n",
       "  'reinterpretation': 11,\n",
       "  'infernal': 10,\n",
       "  'affairs': 41,\n",
       "  'rages': 14,\n",
       "  'within': 583,\n",
       "  'between': 2395,\n",
       "  'better': 2314,\n",
       "  'badder': 6,\n",
       "  'selves': 17,\n",
       "  'transforms': 80,\n",
       "  'treachery': 15,\n",
       "  'movies': 3592,\n",
       "  'watch': 1675,\n",
       "  'likes': 138,\n",
       "  'talking': 290,\n",
       "  'heads': 177,\n",
       "  'blondie': 2,\n",
       "  'succeeds': 543,\n",
       "  'unglamorized': 1,\n",
       "  'ode': 132,\n",
       "  'uncompromising': 136,\n",
       "  'scene': 858,\n",
       "  'york': 283,\n",
       "  'talented': 345,\n",
       "  'consumed': 25,\n",
       "  'masterpiece': 670,\n",
       "  'storytelling': 762,\n",
       "  'predecessor': 319,\n",
       "  'was': 3201,\n",
       "  'instead': 637,\n",
       "  'perfectly': 675,\n",
       "  'acceptable': 47,\n",
       "  'pleasantly': 125,\n",
       "  'diverting': 145,\n",
       "  'fairy': 293,\n",
       "  'alike': 148,\n",
       "  'dynamic': 226,\n",
       "  'supple': 10,\n",
       "  'open': 353,\n",
       "  'differing': 9,\n",
       "  'opinions': 26,\n",
       "  'ideas': 568,\n",
       "  'concluding': 16,\n",
       "  'dredd': 20,\n",
       "  'purest': 52,\n",
       "  'gratifying': 48,\n",
       "  'law': 141,\n",
       "  'enforcing': 2,\n",
       "  'robocop': 32,\n",
       "  'cleans': 2,\n",
       "  'corrupt': 46,\n",
       "  'motown': 9,\n",
       "  'victory': 58,\n",
       "  'hes': 1050,\n",
       "  'got': 632,\n",
       "  'machinery': 23,\n",
       "  'morals': 19,\n",
       "  'side': 656,\n",
       "  'tech': 88,\n",
       "  'meets': 242,\n",
       "  'noon': 11,\n",
       "  'agonies': 8,\n",
       "  'ecstasies': 2,\n",
       "  'equally': 267,\n",
       "  'darker': 202,\n",
       "  'tone': 777,\n",
       "  'suitable': 76,\n",
       "  'older': 231,\n",
       "  'disneys': 148,\n",
       "  'retains': 105,\n",
       "  'undeniably': 232,\n",
       "  'spirit': 683,\n",
       "  'earlier': 172,\n",
       "  'finest': 380,\n",
       "  'date': 455,\n",
       "  'japanese': 179,\n",
       "  ...},\n",
       " 'rotten': {'bendingers': 3,\n",
       "  'film': 14227,\n",
       "  'is': 40108,\n",
       "  'a': 79156,\n",
       "  'mess': 969,\n",
       "  'thats': 2487,\n",
       "  'really': 2602,\n",
       "  'all': 8355,\n",
       "  'there': 4158,\n",
       "  'to': 54307,\n",
       "  'it': 28393,\n",
       "  'how': 2588,\n",
       "  'would': 3352,\n",
       "  'story': 5396,\n",
       "  'written': 534,\n",
       "  'before': 1583,\n",
       "  'we': 2786,\n",
       "  'made': 2397,\n",
       "  'the': 120580,\n",
       "  'moon': 93,\n",
       "  'that': 27805,\n",
       "  'so': 7895,\n",
       "  'complicated': 163,\n",
       "  'didnt': 766,\n",
       "  'work': 1871,\n",
       "  'in': 32261,\n",
       "  'and': 61029,\n",
       "  'doesnt': 3996,\n",
       "  'now': 897,\n",
       "  'if': 6393,\n",
       "  'you': 9569,\n",
       "  'fixate': 4,\n",
       "  'on': 12126,\n",
       "  'funny': 2055,\n",
       "  'things': 1219,\n",
       "  'like': 9419,\n",
       "  'plot': 2553,\n",
       "  'character': 2032,\n",
       "  'movies': 2929,\n",
       "  'anyway': 97,\n",
       "  'too': 6533,\n",
       "  'bad': 3201,\n",
       "  'willis': 74,\n",
       "  'appears': 262,\n",
       "  'have': 8065,\n",
       "  'finally': 439,\n",
       "  'grown': 170,\n",
       "  'bored': 229,\n",
       "  'of': 64360,\n",
       "  'his': 6057,\n",
       "  'own': 1843,\n",
       "  'shtick': 96,\n",
       "  'malkovich': 28,\n",
       "  'spends': 149,\n",
       "  'most': 3486,\n",
       "  'movie': 14494,\n",
       "  'looking': 817,\n",
       "  'humiliated': 8,\n",
       "  'with': 17445,\n",
       "  'love': 1704,\n",
       "  'my': 1222,\n",
       "  'eye': 431,\n",
       "  'i': 6547,\n",
       "  'liked': 125,\n",
       "  'see': 1905,\n",
       "  'little': 3743,\n",
       "  'more': 9363,\n",
       "  'cris': 2,\n",
       "  'abilities': 37,\n",
       "  'future': 253,\n",
       "  'than': 8280,\n",
       "  'just': 6496,\n",
       "  'another': 1936,\n",
       "  'sappy': 107,\n",
       "  'marc': 53,\n",
       "  'forsteris': 1,\n",
       "  'constrained': 13,\n",
       "  'by': 8638,\n",
       "  'reality': 313,\n",
       "  'left': 957,\n",
       "  'lacks': 1049,\n",
       "  'much': 5995,\n",
       "  'satisfying': 296,\n",
       "  'climax': 273,\n",
       "  'or': 7221,\n",
       "  'anything': 1576,\n",
       "  'proper': 84,\n",
       "  'ending': 655,\n",
       "  'target': 189,\n",
       "  'audience': 1595,\n",
       "  'for': 18265,\n",
       "  'this': 18608,\n",
       "  'are': 8720,\n",
       "  'overly': 369,\n",
       "  'horny': 20,\n",
       "  'teenage': 168,\n",
       "  'boys': 298,\n",
       "  'who': 3885,\n",
       "  'want': 1137,\n",
       "  'female': 332,\n",
       "  'nudity': 84,\n",
       "  'graphic': 114,\n",
       "  'violence': 697,\n",
       "  'no': 5048,\n",
       "  'real': 1758,\n",
       "  'ray': 111,\n",
       "  'winstones': 2,\n",
       "  'irascible': 4,\n",
       "  'english': 165,\n",
       "  'copper': 2,\n",
       "  'but': 21413,\n",
       "  'blows': 69,\n",
       "  'steam': 111,\n",
       "  'out': 5904,\n",
       "  'ears': 50,\n",
       "  'ashley': 13,\n",
       "  'jensen': 4,\n",
       "  'chews': 9,\n",
       "  'whatever': 302,\n",
       "  'scenery': 136,\n",
       "  'hasnt': 223,\n",
       "  'already': 730,\n",
       "  'been': 3723,\n",
       "  'trampled': 5,\n",
       "  'pacing': 252,\n",
       "  'slow': 619,\n",
       "  'intensity': 99,\n",
       "  'needs': 594,\n",
       "  'be': 12334,\n",
       "  'cranked': 10,\n",
       "  'up': 6257,\n",
       "  'notch': 54,\n",
       "  'two': 2689,\n",
       "  'has': 6928,\n",
       "  'emotion': 254,\n",
       "  'your': 2334,\n",
       "  'typical': 155,\n",
       "  'horror': 1593,\n",
       "  'stories': 415,\n",
       "  'problems': 326,\n",
       "  'arthur': 88,\n",
       "  'mostly': 879,\n",
       "  'ones': 463,\n",
       "  'scale': 148,\n",
       "  'set': 1186,\n",
       "  'pieces': 364,\n",
       "  'big': 1834,\n",
       "  'becomes': 1030,\n",
       "  'labored': 65,\n",
       "  'cast': 1684,\n",
       "  'does': 2201,\n",
       "  'they': 3326,\n",
       "  'can': 3648,\n",
       "  'end': 2006,\n",
       "  'dead': 638,\n",
       "  'man': 1453,\n",
       "  'down': 1840,\n",
       "  'plays': 903,\n",
       "  'as': 20449,\n",
       "  'confused': 238,\n",
       "  'noble': 121,\n",
       "  'misfire': 276,\n",
       "  'dialogue': 1178,\n",
       "  'flat': 817,\n",
       "  'animation': 374,\n",
       "  'workmanlike': 50,\n",
       "  'quirky': 160,\n",
       "  'pacy': 3,\n",
       "  'enough': 3324,\n",
       "  'keep': 731,\n",
       "  'younger': 135,\n",
       "  'children': 373,\n",
       "  'quiet': 101,\n",
       "  'dearth': 25,\n",
       "  'new': 1986,\n",
       "  'at': 8623,\n",
       "  'multiplex': 78,\n",
       "  'dwayne': 55,\n",
       "  'johnsons': 36,\n",
       "  'draw': 115,\n",
       "  'may': 2114,\n",
       "  'make': 3705,\n",
       "  'lucrative': 12,\n",
       "  'opening': 292,\n",
       "  'weekend': 95,\n",
       "  'not': 10505,\n",
       "  'beyond': 696,\n",
       "  'stepford': 21,\n",
       "  'threat': 40,\n",
       "  'isnt': 2836,\n",
       "  'tradition': 77,\n",
       "  'minded': 164,\n",
       "  'men': 691,\n",
       "  'corporate': 123,\n",
       "  'mother': 154,\n",
       "  'hens': 1,\n",
       "  'such': 1640,\n",
       "  'martha': 13,\n",
       "  'stewart': 70,\n",
       "  'bake': 7,\n",
       "  'their': 3255,\n",
       "  'cakes': 3,\n",
       "  'eat': 107,\n",
       "  'them': 2281,\n",
       "  'garden': 44,\n",
       "  'state': 204,\n",
       "  'braff': 16,\n",
       "  'takes': 1124,\n",
       "  'trials': 29,\n",
       "  'tribulations': 7,\n",
       "  'bloom': 33,\n",
       "  'family': 922,\n",
       "  'life': 2057,\n",
       "  'ties': 27,\n",
       "  'off': 2578,\n",
       "  'happily': 43,\n",
       "  'ever': 1361,\n",
       "  'after': 1659,\n",
       "  'finale': 210,\n",
       "  'wish': 523,\n",
       "  'was': 3841,\n",
       "  'here': 2386,\n",
       "  'light': 446,\n",
       "  'entertainment': 599,\n",
       "  'forces': 69,\n",
       "  'itself': 1473,\n",
       "  'upon': 271,\n",
       "  'disappointing': 627,\n",
       "  'directed': 507,\n",
       "  'good': 4211,\n",
       "  'actor': 440,\n",
       "  'peter': 205,\n",
       "  'berg': 42,\n",
       "  'thinks': 206,\n",
       "  'tale': 1179,\n",
       "  'clever': 449,\n",
       "  'witty': 124,\n",
       "  'whole': 1114,\n",
       "  'its': 28219,\n",
       "  'headache': 55,\n",
       "  'inducing': 150,\n",
       "  'ordeal': 44,\n",
       "  'which': 3602,\n",
       "  'might': 2337,\n",
       "  'hopefully': 74,\n",
       "  'put': 668,\n",
       "  'an': 12568,\n",
       "  'found': 605,\n",
       "  'footage': 247,\n",
       "  'sub': 133,\n",
       "  'genre': 1017,\n",
       "  'kind': 1503,\n",
       "  'people': 1749,\n",
       "  'will': 3811,\n",
       "  'discover': 59,\n",
       "  'hilarious': 210,\n",
       "  'ways': 421,\n",
       "  'mock': 39,\n",
       "  'years': 1026,\n",
       "  'come': 1243,\n",
       "  'shallows': 17,\n",
       "  'often': 1465,\n",
       "  'ankle': 4,\n",
       "  'deep': 417,\n",
       "  'unable': 164,\n",
       "  'past': 513,\n",
       "  'snout': 4,\n",
       "  'theres': 3226,\n",
       "  'impressive': 333,\n",
       "  'production': 667,\n",
       "  'values': 163,\n",
       "  'effective': 263,\n",
       "  'star': 1009,\n",
       "  'turn': 632,\n",
       "  'burtons': 45,\n",
       "  'favorite': 105,\n",
       "  'johnny': 99,\n",
       "  'depp': 100,\n",
       "  'comes': 1379,\n",
       "  'together': 1086,\n",
       "  'remarkably': 84,\n",
       "  'dull': 1260,\n",
       "  'non': 319,\n",
       "  'starter': 18,\n",
       "  'could': 2633,\n",
       "  'least': 1341,\n",
       "  'b': 329,\n",
       "  'grade': 139,\n",
       "  'flick': 713,\n",
       "  'had': 1832,\n",
       "  'conviction': 96,\n",
       "  'what': 5133,\n",
       "  'were': 1989,\n",
       "  'high': 1199,\n",
       "  'school': 579,\n",
       "  'geek': 26,\n",
       "  'porn': 197,\n",
       "  'moved': 68,\n",
       "  'next': 657,\n",
       "  'door': 119,\n",
       "  'answer': 168,\n",
       "  'surely': 184,\n",
       "  'outwit': 4,\n",
       "  'cut': 359,\n",
       "  'rate': 186,\n",
       "  'risky': 27,\n",
       "  'business': 196,\n",
       "  'child': 273,\n",
       "  'michael': 501,\n",
       "  'haneke': 30,\n",
       "  'eli': 46,\n",
       "  'roth': 45,\n",
       "  'meaning': 294,\n",
       "  'basically': 262,\n",
       "  'arty': 60,\n",
       "  'torture': 169,\n",
       "  'tranquilizers': 2,\n",
       "  'blatantly': 59,\n",
       "  'hearted': 118,\n",
       "  'product': 350,\n",
       "  'half': 1544,\n",
       "  'vibrant': 41,\n",
       "  'original': 1404,\n",
       "  'wired': 9,\n",
       "  'article': 17,\n",
       "  'based': 570,\n",
       "  'myopically': 2,\n",
       "  'neglects': 19,\n",
       "  'address': 45,\n",
       "  'arizonas': 1,\n",
       "  'troubling': 48,\n",
       "  'anti': 247,\n",
       "  'immigration': 18,\n",
       "  'legislation': 1,\n",
       "  'through': 2184,\n",
       "  'even': 4948,\n",
       "  'splash': 17,\n",
       "  'hindsight': 8,\n",
       "  'contraband': 15,\n",
       "  'thriller': 1711,\n",
       "  'offers': 557,\n",
       "  'way': 3022,\n",
       "  'elements': 585,\n",
       "  'assemble': 5,\n",
       "  'minute': 622,\n",
       "  'trailer': 155,\n",
       "  'death': 584,\n",
       "  'smoochy': 8,\n",
       "  'odd': 263,\n",
       "  'im': 750,\n",
       "  'almost': 1575,\n",
       "  'tempted': 27,\n",
       "  'recommend': 237,\n",
       "  'promising': 254,\n",
       "  'quite': 1654,\n",
       "  'ultimately': 1524,\n",
       "  'underwhelming': 238,\n",
       "  'first': 2406,\n",
       "  'makes': 2187,\n",
       "  'some': 4004,\n",
       "  'beautiful': 399,\n",
       "  'music': 687,\n",
       "  'otherwise': 388,\n",
       "  'shallow': 409,\n",
       "  'contrived': 375,\n",
       "  'melodrama': 559,\n",
       "  'coming': 483,\n",
       "  'mile': 91,\n",
       "  'away': 919,\n",
       "  'then': 1399,\n",
       "  'id': 315,\n",
       "  'send': 110,\n",
       "  'me': 1696,\n",
       "  'bank': 68,\n",
       "  'account': 87,\n",
       "  'number': 278,\n",
       "  'access': 42,\n",
       "  'codes': 11,\n",
       "  'deposit': 4,\n",
       "  'special': 791,\n",
       "  'prize': 34,\n",
       "  'into': 5089,\n",
       "  'get': 2520,\n",
       "  'dumbed': 74,\n",
       "  'narration': 81,\n",
       "  'tim': 105,\n",
       "  'allen': 248,\n",
       "  'telegraphs': 18,\n",
       "  'should': 2264,\n",
       "  'feeling': 792,\n",
       "  'every': 1843,\n",
       "  'second': 799,\n",
       "  'bit': 1137,\n",
       "  'average': 321,\n",
       "  'iii': 47,\n",
       "  'iv': 28,\n",
       "  'vwhat': 1,\n",
       "  'possibly': 213,\n",
       "  'rocky': 79,\n",
       "  'vs': 139,\n",
       "  'myers': 52,\n",
       "  'essential': 106,\n",
       "  'jolt': 23,\n",
       "  'satisfaction': 57,\n",
       "  'missing': 397,\n",
       "  'chadwick': 7,\n",
       "  'boseman': 5,\n",
       "  'delivers': 355,\n",
       "  'precise': 32,\n",
       "  'impersonation': 26,\n",
       "  'browns': 37,\n",
       "  'gravelly': 2,\n",
       "  'voice': 246,\n",
       "  'eccentric': 65,\n",
       "  'moves': 223,\n",
       "  'lends': 34,\n",
       "  'emotional': 897,\n",
       "  'weight': 241,\n",
       "  'few': 1633,\n",
       "  'scenes': 1356,\n",
       "  'deserve': 278,\n",
       "  'however': 546,\n",
       "  'role': 468,\n",
       "  'feels': 3078,\n",
       "  'series': 1241,\n",
       "  'bullet': 58,\n",
       "  'points': 360,\n",
       "  'organic': 36,\n",
       "  'fluffed': 3,\n",
       "  'satirical': 95,\n",
       "  'targets': 62,\n",
       "  'shortfall': 4,\n",
       "  'required': 90,\n",
       "  'rom': 199,\n",
       "  'com': 246,\n",
       "  'charm': 497,\n",
       "  'cheap': 464,\n",
       "  'gags': 420,\n",
       "  'thoroughly': 214,\n",
       "  'unpleasant': 236,\n",
       "  'hero': 390,\n",
       "  'situate': 2,\n",
       "  'lose': 167,\n",
       "  'friends': 258,\n",
       "  'several': 302,\n",
       "  'floors': 8,\n",
       "  'below': 110,\n",
       "  'desired': 61,\n",
       "  'apartment': 22,\n",
       "  'esque': 66,\n",
       "  'heights': 63,\n",
       "  'though': 1665,\n",
       "  'richly': 24,\n",
       "  'appointed': 12,\n",
       "  'well': 2739,\n",
       "  'meant': 342,\n",
       "  'deeply': 258,\n",
       "  'flawed': 145,\n",
       "  'james': 389,\n",
       "  'titular': 74,\n",
       "  'objet': 3,\n",
       "  'dart': 7,\n",
       "  'bright': 189,\n",
       "  'ambitious': 282,\n",
       "  'terminally': 39,\n",
       "  'self': 1421,\n",
       "  'conscious': 123,\n",
       "  'comedy': 3783,\n",
       "  'hoary': 33,\n",
       "  'conceits': 24,\n",
       "  'ancient': 71,\n",
       "  'vintage': 38,\n",
       "  'dubious': 104,\n",
       "  'worth': 755,\n",
       "  'mukerjis': 1,\n",
       "  'performance': 1168,\n",
       "  'previously': 54,\n",
       "  'mentioned': 13,\n",
       "  'mercy': 25,\n",
       "  'her': 2159,\n",
       "  'alternates': 16,\n",
       "  'between': 1817,\n",
       "  'being': 2156,\n",
       "  'subject': 633,\n",
       "  'subplot': 72,\n",
       "  'according': 38,\n",
       "  'whims': 9,\n",
       "  'greedy': 16,\n",
       "  'pronged': 2,\n",
       "  'biopic': 311,\n",
       "  'while': 2834,\n",
       "  'amusing': 447,\n",
       "  'knuckleheaded': 6,\n",
       "  'tone': 735,\n",
       "  'prevents': 47,\n",
       "  'from': 7826,\n",
       "  'generating': 36,\n",
       "  'wit': 349,\n",
       "  'rabbit': 59,\n",
       "  'identity': 213,\n",
       "  'crisis': 142,\n",
       "  'literally': 181,\n",
       "  'different': 466,\n",
       "  'inside': 287,\n",
       "  'same': 1479,\n",
       "  'never': 3737,\n",
       "  'decide': 160,\n",
       "  'wants': 884,\n",
       "  'shame': 337,\n",
       "  'lavish': 46,\n",
       "  'studded': 43,\n",
       "  'murder': 183,\n",
       "  'mystery': 446,\n",
       "  'takeaway': 20,\n",
       "  'shouldnt': 179,\n",
       "  'kenneth': 24,\n",
       "  'branaghs': 13,\n",
       "  'facial': 31,\n",
       "  'hair': 148,\n",
       "  'ends': 813,\n",
       "  'shot': 558,\n",
       "  'him': 1005,\n",
       "  'board': 87,\n",
       "  'riding': 65,\n",
       "  'rolling': 118,\n",
       "  'wave': 87,\n",
       "  'far': 1795,\n",
       "  'ocean': 31,\n",
       "  'hour': 869,\n",
       "  'journey': 260,\n",
       "  'image': 114,\n",
       "  'play': 752,\n",
       "  'fine': 570,\n",
       "  'digital': 144,\n",
       "  'platform': 14,\n",
       "  'belong': 33,\n",
       "  'screen': 1484,\n",
       "  'despite': 1539,\n",
       "  'solid': 330,\n",
       "  'daring': 102,\n",
       "  'unsatisfying': 180,\n",
       "  'attempt': 673,\n",
       "  'explore': 98,\n",
       "  'terrorist': 38,\n",
       "  'mind': 610,\n",
       "  'paul': 293,\n",
       "  'anderson': 119,\n",
       "  'shoots': 47,\n",
       "  'look': 1344,\n",
       "  'cool': 354,\n",
       "  'he': 2982,\n",
       "  'grasp': 94,\n",
       "  'modulate': 4,\n",
       "  'visuals': 309,\n",
       "  'carry': 168,\n",
       "  'deepening': 6,\n",
       "  'nightmare': 111,\n",
       "  'running': 619,\n",
       "  'time': 3953,\n",
       "  'drags': 120,\n",
       "  'spielberg': 106,\n",
       "  'winds': 176,\n",
       "  'uninspiring': 45,\n",
       "  'tedious': 643,\n",
       "  'sure': 663,\n",
       "  'many': 2355,\n",
       "  'troubled': 80,\n",
       "  'teens': 125,\n",
       "  'today': 136,\n",
       "  'go': 1629,\n",
       "  'violent': 353,\n",
       "  'extremes': 33,\n",
       "  'something': 2254,\n",
       "  'teen': 411,\n",
       "  'pay': 222,\n",
       "  'money': 468,\n",
       "  'needless': 48,\n",
       "  'sequel': 1036,\n",
       "  'serves': 174,\n",
       "  'only': 4501,\n",
       "  'deepen': 14,\n",
       "  'appreciation': 33,\n",
       "  'originals': 68,\n",
       "  'tightwire': 1,\n",
       "  'act': 701,\n",
       "  'suggests': 210,\n",
       "  'screenwriter': 228,\n",
       "  'taylor': 73,\n",
       "  'sheridans': 7,\n",
       "  'moral': 332,\n",
       "  'ambivalence': 19,\n",
       "  'about': 7405,\n",
       "  'american': 852,\n",
       "  'power': 488,\n",
       "  'ended': 90,\n",
       "  'obama': 10,\n",
       "  'presidency': 5,\n",
       "  'problem': 839,\n",
       "  'master': 131,\n",
       "  'extend': 15,\n",
       "  'expand': 23,\n",
       "  'andersons': 49,\n",
       "  'artistic': 168,\n",
       "  'lohan': 35,\n",
       "  'shes': 278,\n",
       "  'better': 2615,\n",
       "  'interesting': 1495,\n",
       "  'material': 1153,\n",
       "  'queen': 134,\n",
       "  'sees': 75,\n",
       "  'herself': 129,\n",
       "  'magic': 426,\n",
       "  'mirror': 64,\n",
       "  'nice': 405,\n",
       "  'twist': 325,\n",
       "  'slightly': 304,\n",
       "  'altered': 17,\n",
       "  'version': 792,\n",
       "  'trying': 1095,\n",
       "  'hard': 1976,\n",
       "  'khaou': 3,\n",
       "  'insists': 45,\n",
       "  'sticking': 45,\n",
       "  'glacial': 28,\n",
       "  'pace': 310,\n",
       "  'lip': 47,\n",
       "  'trembling': 5,\n",
       "  'repression': 17,\n",
       "  'when': 3750,\n",
       "  'gone': 426,\n",
       "  'long': 2154,\n",
       "  'confirmation': 11,\n",
       "  'affirmation': 8,\n",
       "  'challenges': 27,\n",
       "  'deepens': 1,\n",
       "  'record': 79,\n",
       "  'relentlessly': 146,\n",
       "  'exhaustive': 9,\n",
       "  'simple': 318,\n",
       "  'unoriginal': 106,\n",
       "  'inconsequential': 90,\n",
       "  'because': 1576,\n",
       "  'said': 388,\n",
       "  'seems': 2148,\n",
       "  'very': 2572,\n",
       "  'disconnected': 35,\n",
       "  'bid': 32,\n",
       "  'cockeyed': 7,\n",
       "  'beneath': 183,\n",
       "  'synthetic': 40,\n",
       "  'sheen': 72,\n",
       "  'bland': 680,\n",
       "  'musical': 401,\n",
       "  'cues': 50,\n",
       "  'stock': 99,\n",
       "  'patter': 10,\n",
       "  'nothing': 2307,\n",
       "  'oddly': 195,\n",
       "  'niccol': 10,\n",
       "  'seem': 1037,\n",
       "  'clue': 62,\n",
       "  'mine': 80,\n",
       "  'dread': 83,\n",
       "  'anxiety': 38,\n",
       "  'greenly': 1,\n",
       "  'flickering': 9,\n",
       "  'readout': 3,\n",
       "  'human': 701,\n",
       "  'lifespan': 2,\n",
       "  'dumb': 558,\n",
       "  'ideas': 779,\n",
       "  'one': 8122,\n",
       "  'looks': 829,\n",
       "  'great': 1742,\n",
       "  'reheated': 21,\n",
       "  'familiar': 698,\n",
       "  'cautious': 18,\n",
       "  'where': 1500,\n",
       "  'affleck': 73,\n",
       "  'needed': 351,\n",
       "  'loose': 145,\n",
       "  'instead': 1265,\n",
       "  'played': 357,\n",
       "  'safe': 260,\n",
       "  'zohan': 23,\n",
       "  'apatow': 94,\n",
       "  'label': 36,\n",
       "  'got': 636,\n",
       "  'sandler': 234,\n",
       "  'guts': 60,\n",
       "  'sort': 720,\n",
       "  'cuckoos': 4,\n",
       "  'nest': 18,\n",
       "  'meets': 188,\n",
       "  'martian': 15,\n",
       "  'probably': 845,\n",
       "  'sounded': 14,\n",
       "  'paper': 193,\n",
       "  'lynch': 39,\n",
       "  'clueless': 61,\n",
       "  'characters': 3603,\n",
       "  'making': 1175,\n",
       "  'sense': 1606,\n",
       "  'goes': 1017,\n",
       "  'storyline': 222,\n",
       "  'clear': 435,\n",
       "  'precursor': 7,\n",
       "  'milos': 3,\n",
       "  'formans': 1,\n",
       "  'larry': 71,\n",
       "  'flynt': 2,\n",
       "  'depth': 377,\n",
       "  'ambiguity': 66,\n",
       "  'brought': 157,\n",
       "  'amiable': 77,\n",
       "  'monster': 277,\n",
       "  'protagonist': 196,\n",
       "  'kick': 190,\n",
       "  'irritus': 5,\n",
       "  'fourth': 145,\n",
       "  'kept': 133,\n",
       "  'verge': 18,\n",
       "  'parody': 238,\n",
       "  'bardems': 7,\n",
       "  'soulful': 26,\n",
       "  'understated': 29,\n",
       "  'melanie': 5,\n",
       "  'girl': 441,\n",
       "  'gifts': 33,\n",
       "  'neither': 716,\n",
       "  'nor': 692,\n",
       "  'alive': 141,\n",
       "  'somewhere': 308,\n",
       "  'snail': 18,\n",
       "  'paced': 178,\n",
       "  'take': 1406,\n",
       "  'place': 848,\n",
       "  's': 1075,\n",
       "  'since': 575,\n",
       "  'inspired': 312,\n",
       "  'those': 1787,\n",
       "  'hollywood': 941,\n",
       "  'melodramas': 13,\n",
       "  'company': 175,\n",
       "  'employs': 28,\n",
       "  'entire': 405,\n",
       "  'town': 140,\n",
       "  'places': 152,\n",
       "  'free': 409,\n",
       "  'corruption': 50,\n",
       "  'church': 37,\n",
       "  'local': 85,\n",
       "  'diner': 7,\n",
       "  'tintin': 12,\n",
       "  'proves': 457,\n",
       "  'less': 2103,\n",
       "  'inviting': 27,\n",
       "  'wildly': 164,\n",
       "  'overworked': 21,\n",
       "  'hardly': 332,\n",
       "  'moment': 539,\n",
       "  'downtime': 9,\n",
       "  'chance': 264,\n",
       "  'catch': 109,\n",
       "  'breath': 40,\n",
       "  'picture': 1058,\n",
       "  'breaks': 79,\n",
       "  'ground': 275,\n",
       "  'verve': 45,\n",
       "  'actors': 1193,\n",
       "  'fancy': 69,\n",
       "  'footwork': 6,\n",
       "  'battle': 296,\n",
       "  'blades': 23,\n",
       "  'fall': 325,\n",
       "  'nab': 2,\n",
       "  'medal': 9,\n",
       "  'fluff': 99,\n",
       "  'overstuffed': 120,\n",
       "  'heavy': 582,\n",
       "  'themes': 263,\n",
       "  'convoluted': 251,\n",
       "  'lines': 367,\n",
       "  'barely': 571,\n",
       "  'manages': 466,\n",
       "  'squeeze': 37,\n",
       "  'requisite': 53,\n",
       "  'canine': 20,\n",
       "  'flatulence': 23,\n",
       "  'within': 297,\n",
       "  'vacuum': 49,\n",
       "  'malaise': 15,\n",
       "  'performers': 140,\n",
       "  'struggle': 150,\n",
       "  'valiantly': 16,\n",
       "  'breathe': 65,\n",
       "  'roles': 204,\n",
       "  'concepts': 69,\n",
       "  'execution': 343,\n",
       "  'sloppy': 232,\n",
       "  'full': 2773,\n",
       "  'inconsistencies': 27,\n",
       "  'holes': 174,\n",
       "  'situations': 172,\n",
       "  'teeter': 4,\n",
       "  'achieve': 122,\n",
       "  'sections': 17,\n",
       "  'heartwarming': 55,\n",
       "  'feel': 1939,\n",
       "  'lukewarm': 51,\n",
       "  'fetched': 64,\n",
       "  'inappropriately': 6,\n",
       "  'comical': 45,\n",
       "  'connect': 134,\n",
       "  'emotionally': 319,\n",
       "  'evans': 25,\n",
       "  'bcs': 1,\n",
       "  'belle': 10,\n",
       "  'shuffle': 19,\n",
       "  'turgid': 92,\n",
       "  'romantic': 991,\n",
       "  'pretty': 1283,\n",
       "  'theyve': 172,\n",
       "  'chemical': 8,\n",
       "  'spark': 173,\n",
       "  'banana': 10,\n",
       "  'fails': 1211,\n",
       "  'countless': 74,\n",
       "  'fronts': 24,\n",
       "  'bob': 78,\n",
       "  'top': 519,\n",
       "  'sophomoric': 55,\n",
       "  'humor': 945,\n",
       "  'broad': 224,\n",
       "  'generate': 130,\n",
       "  'laughter': 105,\n",
       "  'single': 495,\n",
       "  'ensemble': 172,\n",
       "  'sufficiently': 45,\n",
       "  'developed': 105,\n",
       "  'viewers': 659,\n",
       "  'care': 672,\n",
       "  'awkward': 276,\n",
       "  'marriage': 117,\n",
       "  'intellectual': 113,\n",
       "  'exercise': 566,\n",
       "  'suspenseful': 91,\n",
       "  'creature': 82,\n",
       "  'feature': 872,\n",
       "  'remember': 243,\n",
       "  'hoping': 233,\n",
       "  'predictable': 1016,\n",
       "  'kiddie': 65,\n",
       "  'matinee': 36,\n",
       "  'posing': 41,\n",
       "  'savvy': 36,\n",
       "  'hes': 886,\n",
       "  'rightly': 9,\n",
       "  'seen': 1082,\n",
       "  'carrying': 26,\n",
       "  'anime': 45,\n",
       "  'torch': 8,\n",
       "  'retired': 15,\n",
       "  'yonebayashi': 1,\n",
       "  'certain': 335,\n",
       "  'edge': 249,\n",
       "  'offbeat': 51,\n",
       "  'experimentation': 12,\n",
       "  'perhaps': 761,\n",
       "  'spirited': 171,\n",
       "  'neighbour': 3,\n",
       "  'totoro': 1,\n",
       "  'international': 88,\n",
       "  'hits': 292,\n",
       "  'final': 705,\n",
       "  'house': 513,\n",
       "  'clock': 38,\n",
       "  'walls': 48,\n",
       "  'stumbles': 122,\n",
       "  'telegraphed': 43,\n",
       "  'jolts': 51,\n",
       "  'busy': 177,\n",
       "  'effects': 949,\n",
       "  'action': 2973,\n",
       "  'completely': 669,\n",
       "  'losing': 89,\n",
       "  'sight': 198,\n",
       "  'trauma': 57,\n",
       "  'grief': 84,\n",
       "  'give': 836,\n",
       "  'core': 257,\n",
       "  'bumbling': 31,\n",
       "  'redundant': 111,\n",
       "  'replication': 1,\n",
       "  'disagreeably': 3,\n",
       "  'heavily': 161,\n",
       "  'trafficked': 1,\n",
       "  'municipal': 1,\n",
       "  'baby': 172,\n",
       "  'pool': 45,\n",
       "  'sketch': 108,\n",
       "  'salvage': 47,\n",
       "  'needlessly': 62,\n",
       "  'knots': 14,\n",
       "  'title': 931,\n",
       "  'says': 233,\n",
       "  'youre': 1154,\n",
       "  'unlucky': 11,\n",
       "  'seeing': 356,\n",
       "  'youll': 665,\n",
       "  'repeating': 37,\n",
       "  'minutes': 1626,\n",
       "  'avoid': 176,\n",
       "  'steven': 119,\n",
       "  'deknight': 2,\n",
       "  'ignores': 39,\n",
       "  'tangible': 16,\n",
       "  'details': 236,\n",
       "  'preposterous': 203,\n",
       "  'charmer': 9,\n",
       "  'crafting': 31,\n",
       "  'fixated': 20,\n",
       "  'intrigue': 123,\n",
       "  'reason': 601,\n",
       "  'anyone': 813,\n",
       "  'watches': 23,\n",
       "  'pacific': 41,\n",
       "  'rim': 32,\n",
       "  'robots': 53,\n",
       "  'fighting': 146,\n",
       "  'monsters': 162,\n",
       "  'jorge': 5,\n",
       "  'r': 153,\n",
       "  'gutierrez': 4,\n",
       "  'subsumes': 2,\n",
       "  'films': 3158,\n",
       "  'darker': 110,\n",
       "  'farrago': 17,\n",
       "  'kids': 886,\n",
       "  'tropes': 187,\n",
       "  'annoying': 291,\n",
       "  'attempts': 394,\n",
       "  'hipness': 7,\n",
       "  'director': 2674,\n",
       "  'robert': 288,\n",
       "  'luketic': 5,\n",
       "  'helming': 11,\n",
       "  'frothy': 25,\n",
       "  'romp': 133,\n",
       "  'silliness': 161,\n",
       "  'remaking': 30,\n",
       "  'ingmar': 12,\n",
       "  'bergmans': 13,\n",
       "  'katherine': 32,\n",
       "  'heigl': 47,\n",
       "  'liv': 7,\n",
       "  'ullman': 1,\n",
       "  'mamma': 24,\n",
       "  'mia': 42,\n",
       "  'spicy': 8,\n",
       "  'mlange': 9,\n",
       "  'italian': 56,\n",
       "  'stereoty': 5,\n",
       "  'dummy': 9,\n",
       "  'also': 1500,\n",
       "  'best': 1867,\n",
       "  'editor': 47,\n",
       "  'critic': 70,\n",
       "  'over': 2790,\n",
       "  'triangles': 4,\n",
       "  'chances': 72,\n",
       "  'downey': 50,\n",
       "  'jr': 72,\n",
       "  'discovers': 14,\n",
       "  'girlfriends': 18,\n",
       "  'father': 165,\n",
       "  'gross': 221,\n",
       "  'characterisations': 14,\n",
       "  'weak': 367,\n",
       "  'plotting': 257,\n",
       "  'virtually': 154,\n",
       "  'hurling': 5,\n",
       "  'popcorn': 153,\n",
       "  'rachel': 49,\n",
       "  'tries': 835,\n",
       "  'wistful': 17,\n",
       "  'missed': 279,\n",
       "  'opportunity': 410,\n",
       "  'curiosity': 82,\n",
       "  'rushed': 111,\n",
       "  'theaters': 121,\n",
       "  'casually': 23,\n",
       "  'stitched': 48,\n",
       "  'utterly': 386,\n",
       "  'forgettable': 558,\n",
       "  'chemistry': 330,\n",
       "  'leads': 393,\n",
       "  'dramatic': 593,\n",
       "  'tension': 459,\n",
       "  'badly': 303,\n",
       "  ...}}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_word_counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "35c65d89",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5249869"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_words_in_class = sum(class_word_counts['fresh'].values())+sum(class_word_counts['rotten'].values())\n",
    "total_words_in_class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "1629995f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#creat conditional probability\n",
    "cond_prob={}\n",
    "cond_prob['fresh']={}\n",
    "cond_prob['rotten']={}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "a9859999",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2938168\n",
      "2311701\n"
     ]
    }
   ],
   "source": [
    "total_fresh_counts=sum(class_word_counts['fresh'].values())\n",
    "total_word_counts=sum(class_word_counts['rotten'].values())\n",
    "print(total_fresh_counts)\n",
    "print(total_word_counts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "6902be0d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#conditional probability\n",
    "for word in vocabulary:\n",
    "    if word in class_word_counts['fresh']:\n",
    "        word_count = class_word_counts['fresh'][word]\n",
    "        cond_prob['fresh'][word] = (word_count+0.001) / total_fresh_counts\n",
    "    if word in class_word_counts['rotten']:\n",
    "        word_count = class_word_counts['rotten'][word]\n",
    "        cond_prob['rotten'][word] = (word_count+0.001) / total_word_counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "bbd24751",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5549976661157859\n",
      "0.4450023338842141\n"
     ]
    }
   ],
   "source": [
    "prior_probab_fresh=class_count['fresh']/len(train_df_Review)\n",
    "prior_probab_rotten=class_count['rotten']/len(train_df_Review)\n",
    "print(prior_probab_fresh)\n",
    "print(prior_probab_rotten)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "80986eb2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#development dataset\n",
    "dev_data_Review=dev_data.Review.tolist()\n",
    "dev_data_Freshness=dev_data.Freshness.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "62f04ee5",
   "metadata": {},
   "outputs": [],
   "source": [
    "#conditional probability \n",
    "\n",
    "new_dev_Freshness=[]\n",
    "for i in dev_data_Review:\n",
    "    words = i.split()\n",
    "    # Initialize conditional probabilities\n",
    "    positive_prob = 0.000000000000000001       #smoothing\n",
    "    negative_prob = 0.000000000000000001\n",
    "    #print(words)\n",
    "    # Calculate conditional probabilities for each word in the sentence\n",
    "    for word in words:\n",
    "    #    # Check if word is in positive word dictionary\n",
    "        if word in cond_prob['fresh']:\n",
    "            positive_prob *= float(cond_prob['fresh'][word])\n",
    "    #    # Check if word is in negative word dictionary\n",
    "        if word in cond_prob['rotten']:\n",
    "            negative_prob *= float(cond_prob['rotten'][word])\n",
    "\n",
    "    \n",
    "    if positive_prob>negative_prob:\n",
    "        new_dev_Freshness.append('fresh')\n",
    "    else:\n",
    "        new_dev_Freshness.append('rotten')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "91b1834a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of dev dataset using conditional probability:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7184685126675432"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accu_count=float(0)\n",
    "for i in range(len(new_dev_Freshness)):\n",
    "    if new_dev_Freshness[i]==dev_data_Freshness[i]:\n",
    "        #print(l[i])\n",
    "        accu_count=accu_count+1\n",
    "accuracy=accu_count/len(dev_data_Freshness)\n",
    "print(\"Accuracy of dev dataset using conditional probability:\")\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ba2f0b7a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.051902818908433714\n"
     ]
    }
   ],
   "source": [
    "#probability of occurance of the in sentence\n",
    "prob=class_word_counts['fresh']['the']+class_word_counts['rotten']['the']\n",
    "total=sum(class_word_counts['fresh'].values())+sum(class_word_counts['rotten'].values())\n",
    "print(prob/total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "4bc2af2a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fresh: 0.05169990313692069\n",
      "rotten: 0.052160725370625355\n"
     ]
    }
   ],
   "source": [
    "#p['the'|'Positive]\n",
    "print(\"fresh:\",cond_prob['fresh']['the'])\n",
    "print('rotten:',cond_prob['rotten']['the'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "70843a20",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['are', 'theirs', 'that', 'is', 'which', 'me', 'his', 'were', 'her', 'he', 'at', 'itself', 'you', 'yourselves', 'these', 'by', 'it', 'ourselves', 'who', 'on', 'i', 'she', 'they', 'and', 'be', 'whom', 'this', 'herself', 'a', 'what', 'as', 'to', 'whose', 'for', 'them', 'yours', 'was', 'hers', 'myself', 'yourself', 'himself', 'an', 'those', 'with', 'in', 'the', 'will', 'mine', 'ours', 'of', 'from', 'him', 'themselves', 'us', 'we', 'its', 'has']\n"
     ]
    }
   ],
   "source": [
    "pronouns = {'i', 'me', 'you', 'he', 'him', 'she', 'her', 'it', 'we', 'us', 'they', 'them', 'mine', 'yours', 'his', 'hers', 'its', 'ours', 'theirs', 'myself', 'yourself', 'himself', 'herself', 'itself', 'ourselves', 'yourselves', 'themselves', 'this', 'that', 'these', 'those', 'who', 'whom', 'whose', 'which', 'what', 'a', 'an', 'and', 'are', 'as', 'at', 'be', 'by', 'for', 'from', 'has', 'he', 'in', 'is', 'it', 'its', 'of', 'on', 'that', 'the', 'to', 'was', 'were', 'will', 'with'}\n",
    "pronouns=list(pronouns)\n",
    "print(pronouns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "ad853896",
   "metadata": {},
   "outputs": [],
   "source": [
    "#creat new vocabulary,class word counts,conditional probability to delete pronoun values from them\n",
    "vocabulary1=vocabulary.copy()\n",
    "class_word_counts1=class_word_counts\n",
    "cond_prob1=cond_prob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "298c9be1",
   "metadata": {},
   "outputs": [],
   "source": [
    "for x in pronouns:\n",
    "    if x in vocabulary1:\n",
    "        vocabulary1.remove(x)\n",
    "vocabulary1=sorted(vocabulary1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "0c9d6ba0",
   "metadata": {},
   "outputs": [],
   "source": [
    "for x in pronouns:\n",
    "    if x in class_word_counts1['fresh']:\n",
    "        del class_word_counts1['fresh'][x]\n",
    "    if x in class_word_counts1['rotten']:\n",
    "        del class_word_counts1['rotten'][x]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "d00308a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "for x in pronouns:\n",
    "    if x in cond_prob1['fresh']:\n",
    "        del cond_prob1['fresh'][x]\n",
    "    if x in cond_prob1['rotten']:\n",
    "        del cond_prob1['rotten'][x]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "fe535ade",
   "metadata": {},
   "outputs": [],
   "source": [
    "#conditional probability \n",
    "\n",
    "new_dev_Freshness=[]\n",
    "for i in dev_data_Review:\n",
    "    words = i.split()\n",
    "    # Initialize conditional probabilities\n",
    "    positive_prob = 0.000000000000000001        #smoothing\n",
    "    negative_prob = 0.000000000000000001\n",
    "    #print(words)\n",
    "    # Calculate conditional probabilities for each word in the sentence\n",
    "    for word in words:\n",
    "    #    # Check if word is in positive word dictionary\n",
    "        if word in cond_prob1['fresh']:\n",
    "            positive_prob *= float(cond_prob1['fresh'][word])\n",
    "    #    # Check if word is in negative word dictionary\n",
    "        if word in cond_prob1['rotten']:\n",
    "            negative_prob *= float(cond_prob1['rotten'][word])\n",
    "\n",
    "    \n",
    "    if positive_prob>negative_prob:\n",
    "        new_dev_Freshness.append('fresh')\n",
    "    else:\n",
    "        new_dev_Freshness.append('rotten')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "6a095fe9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of dev dataset using conditional probability:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7160939615754459"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#after removing pronoun words\n",
    "accu_count=float(0)\n",
    "for i in range(len(new_dev_Freshness)):\n",
    "    if new_dev_Freshness[i]==dev_data_Freshness[i]:\n",
    "        #print(l[i])\n",
    "        accu_count=accu_count+1\n",
    "accuracy=accu_count/len(dev_data_Freshness)\n",
    "print(\"Accuracy of dev dataset using conditional probability:\")\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "88744fa3",
   "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>Freshness</th>\n",
       "      <th>Review</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>104419</th>\n",
       "      <td>rotten</td>\n",
       "      <td>sofia coppolas  lost in translation was such ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121937</th>\n",
       "      <td>rotten</td>\n",
       "      <td>the raven however is very serious and the dis...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>280266</th>\n",
       "      <td>fresh</td>\n",
       "      <td>the best horror films have always reflected t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228692</th>\n",
       "      <td>fresh</td>\n",
       "      <td>a big fat value meal stuffed with profane rav...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>189229</th>\n",
       "      <td>fresh</td>\n",
       "      <td>as a fragmentary middle piece of something it...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>360468</th>\n",
       "      <td>rotten</td>\n",
       "      <td>the problem with revolver is that it is ritch...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>352358</th>\n",
       "      <td>rotten</td>\n",
       "      <td>long winded ponderous and without much in the...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>335105</th>\n",
       "      <td>fresh</td>\n",
       "      <td>watching ra dickey become one of the biggest ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>441188</th>\n",
       "      <td>rotten</td>\n",
       "      <td>the film has no shape or narrative force it m...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170980</th>\n",
       "      <td>rotten</td>\n",
       "      <td>it winds up being less than the sum of its parts</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>50958 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Freshness                                             Review\n",
       "104419    rotten   sofia coppolas  lost in translation was such ...\n",
       "121937    rotten   the raven however is very serious and the dis...\n",
       "280266     fresh   the best horror films have always reflected t...\n",
       "228692     fresh   a big fat value meal stuffed with profane rav...\n",
       "189229     fresh   as a fragmentary middle piece of something it...\n",
       "...          ...                                                ...\n",
       "360468    rotten   the problem with revolver is that it is ritch...\n",
       "352358    rotten   long winded ponderous and without much in the...\n",
       "335105     fresh   watching ra dickey become one of the biggest ...\n",
       "441188    rotten   the film has no shape or narrative force it m...\n",
       "170980    rotten   it winds up being less than the sum of its parts\n",
       "\n",
       "[50958 rows x 2 columns]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "694b186a",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_data_Review=test_data.Review.tolist()\n",
    "test_data_Freshness=test_data.Freshness.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "15e074c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "#conditional probability \n",
    "\n",
    "new_test_Freshness=[]\n",
    "for i in test_data_Review:\n",
    "    words = i.split()\n",
    "    # Initialize conditional probabilities\n",
    "    positive_prob = 0.000000000000000001     #smoothing\n",
    "    negative_prob = 0.000000000000000001\n",
    "    #print(words)\n",
    "    # Calculate conditional probabilities for each word in the sentence\n",
    "    for word in words:\n",
    "    #    # Check if word is in positive word dictionary\n",
    "        if word in cond_prob['fresh']:\n",
    "            positive_prob *= float(cond_prob['fresh'][word])\n",
    "    #    # Check if word is in negative word dictionary\n",
    "        if word in cond_prob['rotten']:\n",
    "            negative_prob *= float(cond_prob['rotten'][word])\n",
    "\n",
    "    \n",
    "    if positive_prob<negative_prob:\n",
    "        new_test_Freshness.append('rotten')\n",
    "    else:\n",
    "        new_test_Freshness.append('fresh')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "f6bce8f5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of test dataset using conditional probability:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7157659248793123"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accu_count=float(0)\n",
    "for i in range(len(new_test_Freshness)):\n",
    "    if new_test_Freshness[i]==test_data_Freshness[i]:\n",
    "        #print(l[i])\n",
    "        accu_count=accu_count+1\n",
    "accuracy=accu_count/len(test_data_Freshness)\n",
    "print(\"Accuracy of test dataset using conditional probability:\")\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ff3f0ab9",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "fadb24c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "#prediction of test dataset using naive base theorem\n",
    "#Navies bayies theorem\n",
    "new_test_Freshness1=[]\n",
    "for i in test_data_Review:\n",
    "    words = i.split()\n",
    "    # Initialize conditional probabilities\n",
    "    positive_prob = 0.000000000000000001     #smoothing\n",
    "    negative_prob = 0.000000000000000001\n",
    "    #print(words)\n",
    "    # Calculate conditional probabilities for each word in the sentence\n",
    "    for word in words:\n",
    "    #    # Check if word is in positive word dictionary\n",
    "        if word in cond_prob['fresh']:\n",
    "            positive_prob *= float(cond_prob['fresh'][word])\n",
    "    #    # Check if word is in negative word dictionary\n",
    "        if word in cond_prob['rotten']:\n",
    "            negative_prob *= float(cond_prob['rotten'][word])\n",
    "    #    # If word is not in either dictionary, ignore it\n",
    "    # Normalize probabilities\n",
    "    prob_sentenc = positive_prob * prior_probab_fresh + negative_prob * prior_probab_rotten\n",
    "    Navie_positive_prob = positive_prob * prior_probab_fresh / prob_sentenc   \n",
    "    Navie_negative_prob = negative_prob * prior_probab_rotten / prob_sentenc    \n",
    "    \n",
    "    if Navie_positive_prob>Navie_negative_prob:\n",
    "        new_test_Freshness1.append('fresh')\n",
    "    else:\n",
    "        new_test_Freshness1.append('rotten')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "9068d994",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of test dataset using Naive Bayes theorem:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7172377251854468"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accu_count=float(0)\n",
    "for i in range(len(new_test_Freshness1)):\n",
    "    if new_test_Freshness1[i]==test_data_Freshness[i]:\n",
    "        #print(l[i])\n",
    "        accu_count=accu_count+1\n",
    "accuracy=accu_count/len(test_data_Freshness)\n",
    "print(\"Accuracy of test dataset using Naive Bayes theorem:\")\n",
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "a1abe176",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'fresh': {}, 'rotten': {}}"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#create probability dictionary for naive bayes theorem probability\n",
    "probability={}\n",
    "probability['fresh']={}\n",
    "probability['rotten']={}\n",
    "probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "25333c1e",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in train_df_Review:\n",
    "    words = i.split() \n",
    "    positive_prob=0.000000000000000001 #smoothing in case word not in conditional probability\n",
    "    negative_prob=0.000000000000000001\n",
    "    for word in words:\n",
    "        if word in cond_prob['fresh']:\n",
    "            positive_prob+= cond_prob['fresh'][word]\n",
    "        if word in cond_prob['rotten']:\n",
    "            negative_prob+= cond_prob['rotten'][word]\n",
    "            \n",
    "        prob_sentenc = positive_prob * prior_probab_fresh + negative_prob * prior_probab_rotten\n",
    "        Navie_positive_prob = positive_prob * prior_probab_fresh / prob_sentenc   \n",
    "        Navie_negative_prob = negative_prob * prior_probab_rotten / prob_sentenc\n",
    "        probability['fresh'][word] = Navie_positive_prob\n",
    "        probability['rotten'][word] = Navie_negative_prob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "125ec292",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'bendingers': 0.5833156892940583,\n",
       " 'film': 0.5955593771872225,\n",
       " 'is': 0.5560055306914136,\n",
       " 'a': 0.5549976661157859,\n",
       " 'mess': 0.1840665148510517,\n",
       " 'thats': 0.5196191506577378,\n",
       " 'really': 0.4535708679928346,\n",
       " 'all': 0.5065117034771176,\n",
       " 'there': 0.5231693548883931,\n",
       " 'to': 0.5316613614856638,\n",
       " 'it': 0.5788754218362174,\n",
       " 'marked': 0.5248535992887666,\n",
       " 'by': 0.5117498036552036,\n",
       " 'tossed': 0.5704453095617021,\n",
       " 'off': 0.48420241657634655,\n",
       " 'feeling': 0.456534639599976,\n",
       " 'distinct': 0.5948836437219106,\n",
       " 'impression': 0.5940545235533414,\n",
       " 'that': 0.5465417228661428,\n",
       " 'americas': 0.7223610462338631,\n",
       " 'most': 0.6616456815876458,\n",
       " 'viable': 0.5415898054741641,\n",
       " 'auteur': 0.5735815204119108,\n",
       " 'brats': 0.5283809628439745,\n",
       " 'had': 0.5014556818724999,\n",
       " 'taken': 0.5569577155221914,\n",
       " 'an': 0.4633729123099568,\n",
       " 'unfamiliar': 0.5476613419728029,\n",
       " 'path': 0.5100064832034656,\n",
       " 'with': 0.47066369378237344,\n",
       " 'minimal': 0.5334250969769686,\n",
       " 'investment': 0.5182131274184163,\n",
       " 'for': 0.5836609623081563,\n",
       " 'uncertain': 0.4986400668413912,\n",
       " 'gain': 0.6516306736268034,\n",
       " 'how': 0.5391496585446579,\n",
       " 'would': 0.5524365611364785,\n",
       " 'story': 0.5472911533535769,\n",
       " 'written': 0.5968365058661196,\n",
       " 'before': 0.5223189390345225,\n",
       " 'we': 0.488706325813146,\n",
       " 'made': 0.5455043429764839,\n",
       " 'the': 0.5550446536395446,\n",
       " 'moon': 0.5167497263144166,\n",
       " 'so': 0.5117894723342569,\n",
       " 'complicated': 0.6009946103558148,\n",
       " 'didnt': 0.5384590226448157,\n",
       " 'work': 0.5257732322345707,\n",
       " 'in': 0.5550446536395446,\n",
       " 'and': 0.30109205590075017,\n",
       " 'doesnt': 0.5466089438341265,\n",
       " 'now': 0.49655343991148704,\n",
       " 'classic': 0.6040535383215467,\n",
       " 'moments': 0.559975178527462,\n",
       " 'are': 0.5231693548883931,\n",
       " 'crammed': 0.4985278558311854,\n",
       " 'as': 0.5075673293470114,\n",
       " 'if': 0.5252670340354111,\n",
       " 'might': 0.49477206750073655,\n",
       " 'not': 0.5446405105552523,\n",
       " 'get': 0.5066826044026206,\n",
       " 'another': 0.47175692855045187,\n",
       " 'shot': 0.6021023319907269,\n",
       " 'at': 0.6568145559951151,\n",
       " 'this': 0.5201890372403432,\n",
       " 'its': 0.6395230409635729,\n",
       " 'very': 0.5093389046979357,\n",
       " 'sweet': 0.6617435132563634,\n",
       " 'charming': 0.5196649286166661,\n",
       " 'should': 0.4767263024085172,\n",
       " 'be': 0.520193549395545,\n",
       " 'thankful': 0.4914718005785429,\n",
       " 'isnt': 0.4268009940408407,\n",
       " 'childhood': 0.4914127522713388,\n",
       " 'ruining': 0.39271073167005693,\n",
       " 'disaster': 0.6400636490740341,\n",
       " 'mulholland': 0.436997596783757,\n",
       " 'drive': 0.5120505657868967,\n",
       " 'maddeningly': 0.7361735180613537,\n",
       " 'abstract': 0.6038905365618007,\n",
       " 'inscrutable': 0.517838309313701,\n",
       " 'on': 0.5108455853940074,\n",
       " 'first': 0.6223485271862143,\n",
       " 'viewing': 0.5553113730273381,\n",
       " 'but': 0.5530523308631778,\n",
       " 'hypnotic': 0.5452140151297372,\n",
       " 'haunting': 0.5943594882474836,\n",
       " 'every': 0.5199687038058755,\n",
       " 'after': 0.5369744869299207,\n",
       " 'you': 0.5369744869299207,\n",
       " 'fixate': 0.5148931104094983,\n",
       " 'funny': 0.5690677319035287,\n",
       " 'things': 0.55450063322697,\n",
       " 'like': 0.4917568380638473,\n",
       " 'plot': 0.5344781982177795,\n",
       " 'character': 0.48492431143225573,\n",
       " 'movies': 0.5084595812049867,\n",
       " 'anyway': 0.5410529972527521,\n",
       " 'too': 0.507831672804219,\n",
       " 'bad': 0.5243130821596096,\n",
       " 'jason': 0.5600074409990384,\n",
       " 'reitman': 0.6010027457265792,\n",
       " 'does': 0.5120083036867618,\n",
       " 'again': 0.5701200218184784,\n",
       " 'although': 0.5516542296509724,\n",
       " 'full': 0.5362069796657684,\n",
       " 'of': 0.5955593771872225,\n",
       " 'product': 0.4682175247662475,\n",
       " 'placements': 0.5128351657091064,\n",
       " 'dark': 0.6110239655743661,\n",
       " 'comedy': 0.5127327827859773,\n",
       " 'surprisingly': 0.7673670276706273,\n",
       " 'enjoyable': 0.518206678745716,\n",
       " 'highlighted': 0.5244215518693042,\n",
       " 'anna': 0.5342606473432497,\n",
       " 'kendrick': 0.5546804447935033,\n",
       " 'who': 0.5295413104454879,\n",
       " 'me': 0.5299196975273488,\n",
       " 'laugh': 0.5218405307105819,\n",
       " 'out': 0.5080880931800369,\n",
       " 'loud': 0.44329100443965025,\n",
       " 'willis': 0.6093652753975216,\n",
       " 'appears': 0.5950307091503899,\n",
       " 'have': 0.5000701909041246,\n",
       " 'finally': 0.5260274109606526,\n",
       " 'grown': 0.5210871043384838,\n",
       " 'bored': 0.5033792935967568,\n",
       " 'his': 0.5419166454422908,\n",
       " 'own': 0.5942081302035879,\n",
       " 'shtick': 0.5439616537510211,\n",
       " 'malkovich': 0.6490222837652332,\n",
       " 'spends': 0.48699309445019656,\n",
       " 'movie': 0.47066369378237344,\n",
       " 'looking': 0.55485256487179,\n",
       " 'humiliated': 0.5582356962836068,\n",
       " 'love': 0.6683164029502698,\n",
       " 'my': 0.5126645607488988,\n",
       " 'eye': 0.5278423854559039,\n",
       " 'i': 0.5255321264060965,\n",
       " 'liked': 0.5326931238412803,\n",
       " 'see': 0.5072474353067989,\n",
       " 'little': 0.47324872675846885,\n",
       " 'more': 0.5163627912605976,\n",
       " 'cris': 0.5724793563868902,\n",
       " 'abilities': 0.5515861910926212,\n",
       " 'future': 0.5821844948993755,\n",
       " 'than': 0.5092691104296516,\n",
       " 'just': 0.47010736303312534,\n",
       " 'sappy': 0.44102830935568144,\n",
       " 'ambitious': 0.6618329176621796,\n",
       " 'passionate': 0.5893912905877617,\n",
       " 'grief': 0.6093567415996677,\n",
       " 'stricken': 0.5889532104269326,\n",
       " 'art': 0.6592546301628884,\n",
       " 'marc': 0.5404840700600776,\n",
       " 'forsteris': 0.49059396966425595,\n",
       " 'constrained': 0.5469229832796593,\n",
       " 'reality': 0.6089269159839571,\n",
       " 'left': 0.4597834337241333,\n",
       " 'lacks': 0.5450354635021584,\n",
       " 'much': 0.5674678510262142,\n",
       " 'satisfying': 0.5012932341636808,\n",
       " 'climax': 0.6451088111195903,\n",
       " 'or': 0.5008497765582366,\n",
       " 'anything': 0.520938612883451,\n",
       " 'proper': 0.6246744310241806,\n",
       " 'ending': 0.5416501196281207,\n",
       " 'target': 0.5068151074906975,\n",
       " 'audience': 0.520193549395545,\n",
       " 'overly': 0.5105588815436227,\n",
       " 'horny': 0.5028365601789553,\n",
       " 'teenage': 0.7576680536423862,\n",
       " 'boys': 0.4821347181310056,\n",
       " 'want': 0.5548363521757534,\n",
       " 'female': 0.5222311619368465,\n",
       " 'nudity': 0.5333514026899474,\n",
       " 'graphic': 0.5715494450555998,\n",
       " 'violence': 0.5696142429651059,\n",
       " 'no': 0.4457786879361116,\n",
       " 'real': 0.5186776330319487,\n",
       " 'ray': 0.5941450504224565,\n",
       " 'winstones': 0.6017440427933297,\n",
       " 'irascible': 0.5269180240134872,\n",
       " 'english': 0.5426136877376018,\n",
       " 'copper': 0.5342752119748112,\n",
       " 'blows': 0.5300424541789258,\n",
       " 'steam': 0.5539745357898486,\n",
       " 'ears': 0.5258420680354614,\n",
       " 'ashley': 0.5180816260556383,\n",
       " 'jensen': 0.574236024958806,\n",
       " 'chews': 0.5337537138545693,\n",
       " 'whatever': 0.4992143415562727,\n",
       " 'scenery': 0.40155977306970286,\n",
       " 'hasnt': 0.5440473416503984,\n",
       " 'already': 0.508905628849362,\n",
       " 'been': 0.4534320142818215,\n",
       " 'trampled': 0.4946533768839195,\n",
       " 'pacing': 0.5837096580346121,\n",
       " 'slow': 0.5198573441986379,\n",
       " 'intensity': 0.49223533397707114,\n",
       " 'needs': 0.4354787561441201,\n",
       " 'cranked': 0.5064459115493782,\n",
       " 'up': 0.48693530885868547,\n",
       " 'notch': 0.6393084249619594,\n",
       " 'two': 0.5195960879639253,\n",
       " 'has': 0.5114025673786263,\n",
       " 'emotion': 0.5869631011949717,\n",
       " 'your': 0.5391894207002578,\n",
       " 'typical': 0.5617817167558833,\n",
       " 'horror': 0.5057473505167812,\n",
       " 'stories': 0.6268556833118892,\n",
       " 'problems': 0.5190337877195891,\n",
       " 'arthur': 0.5335115996616498,\n",
       " 'mostly': 0.5346465758202116,\n",
       " 'ones': 0.5173812837544622,\n",
       " 'scale': 0.6147479882250335,\n",
       " 'set': 0.5135167092960857,\n",
       " 'pieces': 0.4987080894360247,\n",
       " 'big': 0.514298756585201,\n",
       " 'becomes': 0.5103814135734189,\n",
       " 'labored': 0.4898556786720629,\n",
       " 'cast': 0.5522341724239926,\n",
       " 'they': 0.5128959697485922,\n",
       " 'can': 0.5077219094972987,\n",
       " 'end': 0.488706325813146,\n",
       " 'dead': 0.521058878406895,\n",
       " 'man': 0.652976311037656,\n",
       " 'down': 0.49165761219462717,\n",
       " 'plays': 0.5481915426553908,\n",
       " 'confused': 0.5470655898996424,\n",
       " 'noble': 0.4805186896693247,\n",
       " 'misfire': 0.20400002757505883,\n",
       " 'energetic': 0.5766744304604504,\n",
       " 'slapstick': 0.4782568895234308,\n",
       " 'aimed': 0.5707556735915703,\n",
       " 'indiscriminating': 0.5873539745502732,\n",
       " 'youngsters': 0.5087050459978756,\n",
       " 'will': 0.520193549395545,\n",
       " 'buy': 0.38014841765474855,\n",
       " 'lots': 0.5327384949363021,\n",
       " 'crood': 0.5065832287060213,\n",
       " 'toys': 0.6536049885125758,\n",
       " 'one': 0.5352741267867055,\n",
       " 'van': 0.5830708515503861,\n",
       " 'dammes': 0.5505635833444827,\n",
       " 'best': 0.7474122059442175,\n",
       " 'flicksgood': 0.6291964453581458,\n",
       " 'good': 0.5405664766548519,\n",
       " 'skirmishes': 0.4783022468852141,\n",
       " 'masterfully': 0.4921816250398756,\n",
       " 'directed': 0.6399751938658169,\n",
       " 'xavier': 0.5650628353107885,\n",
       " 'beauvois': 0.618061013789255,\n",
       " 'hands': 0.6131281816304075,\n",
       " 'euthanasia': 0.5142763750933189,\n",
       " 'ever': 0.5432066386135713,\n",
       " 'films': 0.5073450388071507,\n",
       " 'humor': 0.49783367610919027,\n",
       " 'sauce': 0.5433002261223091,\n",
       " 'elevate': 0.6102362361817802,\n",
       " 'from': 0.5042975634113942,\n",
       " 'whining': 0.5058973406341029,\n",
       " 'self': 0.5418096415498528,\n",
       " 'absorbed': 0.5035361817119398,\n",
       " 'unattached': 0.5345588468042041,\n",
       " 'women': 0.5042882691116867,\n",
       " 'populate': 0.5727658095462702,\n",
       " 'many': 0.5471757607606745,\n",
       " 'books': 0.5285681445154767,\n",
       " 'today': 0.5459585212480589,\n",
       " 'dialogue': 0.44947255863039604,\n",
       " 'flat': 0.5344830272076444,\n",
       " 'animation': 0.622129288906061,\n",
       " 'workmanlike': 0.4942770075861149,\n",
       " 'quirky': 0.6500040955859373,\n",
       " 'pacy': 0.5424441801242961,\n",
       " 'enough': 0.5157821389651227,\n",
       " 'keep': 0.5062803859136302,\n",
       " 'younger': 0.5381585974117673,\n",
       " 'children': 0.5204212968789748,\n",
       " 'quiet': 0.5911384892063736,\n",
       " 'easily': 0.5046341357367302,\n",
       " 'ryan': 0.5644881018511194,\n",
       " 'reynolds': 0.5107374755414028,\n",
       " 'performance': 0.5774018900700802,\n",
       " 'since': 0.5227805668889781,\n",
       " 'buried': 0.2858440762664661,\n",
       " 'ben': 0.5299210653553309,\n",
       " 'mendelsohn': 0.5514959777913344,\n",
       " 'portrays': 0.5330070713362324,\n",
       " 'flawed': 0.5129711199533777,\n",
       " 'helpless': 0.5766718689290993,\n",
       " 'individuals': 0.5462526653851556,\n",
       " 'incredibly': 0.52875973984812,\n",
       " 'well': 0.5267438064223785,\n",
       " 'when': 0.5101917472133083,\n",
       " 'take': 0.5552571297983195,\n",
       " 'into': 0.5080368456710679,\n",
       " 'consideration': 0.5187837166505921,\n",
       " 'mississippi': 0.5350895789989812,\n",
       " 'grind': 0.5011500475705339,\n",
       " 'addictive': 0.5862617628287536,\n",
       " 'gambling': 0.37058439979799646,\n",
       " 'storyline': 0.5274683642087498,\n",
       " 'involving': 0.530354600924434,\n",
       " 'visually': 0.6391701466169946,\n",
       " 'stunning': 0.7175254025241246,\n",
       " 'ive': 0.5476870470558396,\n",
       " 'seen': 0.5559614613127126,\n",
       " 'dearth': 0.7118199237610261,\n",
       " 'new': 0.5862958523384592,\n",
       " 'multiplex': 0.4721567234115334,\n",
       " 'dwayne': 0.46644328882293645,\n",
       " 'johnsons': 0.5621261166500485,\n",
       " 'draw': 0.5172459270654793,\n",
       " 'may': 0.5974316455762392,\n",
       " 'make': 0.5082684732358267,\n",
       " 'lucrative': 0.4789643945101106,\n",
       " 'opening': 0.5568982166268766,\n",
       " 'weekend': 0.5509895249124783,\n",
       " 'beyond': 0.5618390461196204,\n",
       " 'insidious': 0.49876694934423277,\n",
       " 'chapter': 0.5893811098641554,\n",
       " 'lack': 0.5514683413184099,\n",
       " 'unique': 0.58802634272426,\n",
       " 'charm': 0.6589352850729695,\n",
       " 'both': 0.5280065686241522,\n",
       " 'predecessors': 0.5023989298939766,\n",
       " 'still': 0.5642670587845275,\n",
       " 'scary': 0.48308538736467077,\n",
       " 'fun': 0.6139915355070525,\n",
       " 'thrill': 0.5708245112174555,\n",
       " 'ride': 0.49820997655290206,\n",
       " 'delivers': 0.599230777850772,\n",
       " 'some': 0.5373625982073136,\n",
       " 'serious': 0.6149876152628102,\n",
       " 'blood': 0.4639280970324037,\n",
       " 'pressure': 0.5271389186060286,\n",
       " 'raising': 0.4885266804943554,\n",
       " 'tension': 0.5480333681339102,\n",
       " 'intriguing': 0.5343712270068887,\n",
       " 'tale': 0.5448662008756352,\n",
       " 'about': 0.5659560222761216,\n",
       " 'individual': 0.6136823890033715,\n",
       " 'whose': 0.5219138553016069,\n",
       " 'mounting': 0.7519866092504977,\n",
       " 'debt': 0.5375056102561659,\n",
       " 'leads': 0.5099727969832509,\n",
       " 'unfortunate': 0.5591625757715899,\n",
       " 'circumstances': 0.5309960055275127,\n",
       " 'thai': 0.8153213019247376,\n",
       " 'arthouse': 0.5941630137889539,\n",
       " 'director': 0.5307796450120462,\n",
       " 'apichatpong': 0.5558521302280902,\n",
       " 'weerasethakul': 0.46606997470389744,\n",
       " 'uncle': 0.6053717324289147,\n",
       " 'boonmee': 0.5424765099366403,\n",
       " 'won': 0.6128066734725587,\n",
       " 'palme': 0.5543973083978773,\n",
       " 'dor': 0.6827595971488996,\n",
       " 'cannes': 0.7528982304200461,\n",
       " 'few': 0.5596267097834561,\n",
       " 'years': 0.5842400576266258,\n",
       " 'ago': 0.4771191419433331,\n",
       " 'comes': 0.5580944950465957,\n",
       " 'strange': 0.6575707244238459,\n",
       " 'swooning': 0.5655825785838668,\n",
       " 'dream': 0.6833433354420231,\n",
       " 'cemetery': 0.47946143765214433,\n",
       " 'splendour': 0.5330287859076875,\n",
       " 'engaging': 0.5304926896982816,\n",
       " 'exciting': 0.5861314287185274,\n",
       " 'sci': 0.5357124412309743,\n",
       " 'fi': 0.5377651141681123,\n",
       " 'flick': 0.6015033422914356,\n",
       " 'bring': 0.5439995085215698,\n",
       " 'year': 0.5500003258602112,\n",
       " 'old': 0.5407781679568725,\n",
       " 'us': 0.6267158337373051,\n",
       " 'stepford': 0.2464873154384267,\n",
       " 'threat': 0.6784234890040484,\n",
       " 'tradition': 0.500180568621139,\n",
       " 'minded': 0.588926177402072,\n",
       " 'men': 0.5602228113819037,\n",
       " 'corporate': 0.7265088468469991,\n",
       " 'mother': 0.5337475340788659,\n",
       " 'hens': 0.47645857195132946,\n",
       " 'such': 0.5802484925902006,\n",
       " 'martha': 0.5585990209878599,\n",
       " 'stewart': 0.5966570461040617,\n",
       " 'bake': 0.5948068527614196,\n",
       " 'their': 0.5140499524016063,\n",
       " 'cakes': 0.5276125267948473,\n",
       " 'eat': 0.4946751464537348,\n",
       " 'them': 0.5620951116393459,\n",
       " 'while': 0.5788754218362174,\n",
       " 'find': 0.49865923735978257,\n",
       " 'jarmuschs': 0.7354557220480135,\n",
       " 'call': 0.5292305121815153,\n",
       " 'poor': 0.5183076928779962,\n",
       " 'pretentious': 0.3557020010031129,\n",
       " 'others': 0.5719324235761248,\n",
       " 'adore': 0.5679436365395493,\n",
       " 'artistic': 0.7490933395485326,\n",
       " 'sensibilities': 0.513907998115339,\n",
       " 'war': 0.6250338447370785,\n",
       " 'stalemate': 0.500999377409535,\n",
       " 'aleksandr': 0.9999999999994111,\n",
       " 'burovs': 0.6764031573285699,\n",
       " 'bleached': 0.49175495484521375,\n",
       " 'images': 0.6931382121083862,\n",
       " 'creating': 0.5683241321582306,\n",
       " 'alien': 0.47746769250293297,\n",
       " 'landscape': 0.6252235344410776,\n",
       " 'which': 0.5550446536395446,\n",
       " 'colour': 0.507876752479724,\n",
       " 'rare': 0.487492777631537,\n",
       " 'compassion': 0.5998044765950511,\n",
       " 'rarely': 0.5573793844144488,\n",
       " 'combat': 0.5508215404652865,\n",
       " 'seemed': 0.559111094980755,\n",
       " 'savage': 0.6394827674339955,\n",
       " 'futile': 0.5127873866301091,\n",
       " 'nick': 0.52754510313747,\n",
       " 'morans': 0.999999999999215,\n",
       " 'based': 0.46626015462733045,\n",
       " 'stage': 0.5078467973373376,\n",
       " 'play': 0.5348687450232438,\n",
       " 'he': 0.49235276961449026,\n",
       " 'wrote': 0.5179880240895258,\n",
       " 'james': 0.5944162163878182,\n",
       " 'hicks': 0.5269268041355772,\n",
       " 'eccentric': 0.5286219427844246,\n",
       " 'sometimes': 0.5353407869599689,\n",
       " 'underpowered': 0.49900969891415003,\n",
       " 'always': 0.5283721204487363,\n",
       " 'watchable': 0.5191406543230765,\n",
       " 'early': 0.5159648674124402,\n",
       " 's': 0.629059852032295,\n",
       " 'prehistory': 0.5786140686357789,\n",
       " 'pop': 0.604356966328503,\n",
       " 'culture': 0.7187423391276914,\n",
       " 'garden': 0.5347698507406312,\n",
       " 'state': 0.4723224724544654,\n",
       " 'braff': 0.5108766603688008,\n",
       " 'takes': 0.5773429152806578,\n",
       " 'trials': 0.5157942948782978,\n",
       " 'tribulations': 0.5432402131023595,\n",
       " 'bloom': 0.4894313918865223,\n",
       " 'family': 0.6409289593906101,\n",
       " 'life': 0.6721604955699734,\n",
       " 'ties': 0.5806816838528467,\n",
       " 'happily': 0.6379158860285955,\n",
       " 'finale': 0.5736360553552377,\n",
       " 'wish': 0.5233816371798703,\n",
       " 'was': 0.528524006922286,\n",
       " 'here': 0.5315297704025518,\n",
       " 'light': 0.5591938602407945,\n",
       " 'entertainment': 0.6467213734047186,\n",
       " 'forces': 0.595285981199381,\n",
       " 'itself': 0.6003361095821916,\n",
       " 'upon': 0.47866123571198155,\n",
       " 'part': 0.6500717008685502,\n",
       " 'screenwriters': 0.5134615445926277,\n",
       " 'never': 0.544565398722835,\n",
       " 'lose': 0.5033005054069531,\n",
       " 'sight': 0.5146218016079754,\n",
       " 'moral': 0.5275535821673294,\n",
       " 'else': 0.456631344745963,\n",
       " 'gets': 0.5737023787901795,\n",
       " 'mix': 0.5492175960196398,\n",
       " 'shrek': 0.4836464018504648,\n",
       " 'must': 0.6258000515442406,\n",
       " 'heart': 0.5331405094583256,\n",
       " 'soul': 0.5350483300136804,\n",
       " 'myers': 0.4884378101082425,\n",
       " 'master': 0.6210941132999557,\n",
       " 'makes': 0.5420201081675806,\n",
       " 'seem': 0.5440806907375825,\n",
       " 'easy': 0.5784043714508376,\n",
       " 'being': 0.6191408503812453,\n",
       " 'green': 0.5434134585935188,\n",
       " 'disappointing': 0.4774000296689199,\n",
       " 'actor': 0.6298478356219205,\n",
       " 'peter': 0.5264073744336979,\n",
       " 'berg': 0.5532223841289519,\n",
       " 'thinks': 0.26062537897829374,\n",
       " 'clever': 0.6511252377575517,\n",
       " 'witty': 0.6422182797700202,\n",
       " 'whole': 0.555472696832242,\n",
       " 'headache': 0.5476534048304746,\n",
       " 'inducing': 0.3690190395336126,\n",
       " 'ordeal': 0.536444806669042,\n",
       " 'hopefully': 0.5169360807862224,\n",
       " 'put': 0.4769315581821281,\n",
       " 'found': 0.44517968194725566,\n",
       " 'footage': 0.6327179326919021,\n",
       " 'sub': 0.5224312891394246,\n",
       " 'genre': 0.5195162580494573,\n",
       " 'kind': 0.5162795338937016,\n",
       " 'people': 0.5042942783201264,\n",
       " 'discover': 0.49769805135913375,\n",
       " 'hilarious': 0.6997058945245154,\n",
       " 'ways': 0.5306848711010168,\n",
       " 'mock': 0.5978857991709186,\n",
       " 'come': 0.5482610011302984,\n",
       " 'shallows': 0.7057982021563171,\n",
       " 'often': 0.5127600625009096,\n",
       " 'ankle': 0.6319497507923378,\n",
       " 'deep': 0.5229614410884835,\n",
       " 'unable': 0.27605725244838375,\n",
       " 'past': 0.6222633085099547,\n",
       " 'snout': 0.534934187253256,\n",
       " 'theres': 0.46101299766831216,\n",
       " 'impressive': 0.7127376416574271,\n",
       " 'production': 0.5073312812721362,\n",
       " 'values': 0.6036112675537159,\n",
       " 'effective': 0.6754311634171025,\n",
       " 'star': 0.5182936573813898,\n",
       " 'turn': 0.5081883035687322,\n",
       " 'burtons': 0.5576032368541075,\n",
       " 'favorite': 0.48186122087063843,\n",
       " 'johnny': 0.5181045153189581,\n",
       " 'depp': 0.5389864690653207,\n",
       " 'together': 0.526642072928157,\n",
       " 'remarkably': 0.5744601578641305,\n",
       " 'dull': 0.11784785265108257,\n",
       " 'non': 0.5741770937035913,\n",
       " 'starter': 0.46636225142768584,\n",
       " 'could': 0.48996412078959156,\n",
       " 'least': 0.5191769037039644,\n",
       " 'b': 0.5447035620281938,\n",
       " 'grade': 0.49630245841158194,\n",
       " 'conviction': 0.6370343164385739,\n",
       " 'what': 0.5549976661157859,\n",
       " 'were': 0.2814091615369472,\n",
       " 'high': 0.5823737734062264,\n",
       " 'school': 0.5527617313229133,\n",
       " 'geek': 0.6137686959101006,\n",
       " 'porn': 0.49936886787671686,\n",
       " 'moved': 0.46292839372049965,\n",
       " 'next': 0.5483543035524548,\n",
       " 'door': 0.5141000378117268,\n",
       " 'answer': 0.4779654465167433,\n",
       " 'surely': 0.5965084135238136,\n",
       " 'outwit': 0.5796036422201366,\n",
       " 'cut': 0.524577441687721,\n",
       " 'rate': 0.5925591153470257,\n",
       " 'risky': 0.5790664645351011,\n",
       " 'business': 0.5154607986476877,\n",
       " 'match': 0.6703257604464612,\n",
       " 'bar': 0.5195003008124661,\n",
       " 'royale': 0.5801178361215112,\n",
       " 'neither': 0.5327461218060265,\n",
       " 'let': 0.5017758583891804,\n",
       " 'franchise': 0.5510728886752481,\n",
       " 'fall': 0.5554567355574853,\n",
       " 'unduly': 0.4878285239779893,\n",
       " 'far': 0.5075673293470114,\n",
       " 'mark': 0.517321701532575,\n",
       " 'appealingly': 0.46824946232423953,\n",
       " 'cheesy': 0.5291514223691504,\n",
       " 'tribute': 0.5566289671053152,\n",
       " 'hope': 0.5517989348454724,\n",
       " 'springs': 0.47546758933451333,\n",
       " 'eternal': 0.4749748001650996,\n",
       " 'hopelessly': 0.21484092646708947,\n",
       " 'inept': 0.5555394358272452,\n",
       " 'photogenic': 0.5325366020796581,\n",
       " 'charismatic': 0.6052255551594441,\n",
       " 'earth': 0.5510667241899048,\n",
       " 'miss': 0.5206817155197742,\n",
       " 'quick': 0.5346549144270022,\n",
       " 'show': 0.5080979514232346,\n",
       " 'characters': 0.5310357640947867,\n",
       " 'skin': 0.5732563976488297,\n",
       " 'less': 0.5038507425011782,\n",
       " 'inclined': 0.5207920881068543,\n",
       " 'explore': 0.48053913190830755,\n",
       " 'lies': 0.595679172063428,\n",
       " 'beneath': 0.6481042899620113,\n",
       " 'those': 0.5773206032312694,\n",
       " 'saw': 0.4894676098613687,\n",
       " 'animated': 0.5756873906719496,\n",
       " 'version': 0.45494889710655834,\n",
       " 'musical': 0.6327186313165964,\n",
       " 'numbers': 0.5736756501408334,\n",
       " 'yet': 0.6486512598779097,\n",
       " 'favreau': 0.531875273742824,\n",
       " 'manages': 0.5557444243470421,\n",
       " 'create': 0.5929134417806734,\n",
       " 'beautiful': 0.574149940994424,\n",
       " 'balanced': 0.5657789516631148,\n",
       " 'review': 0.5408462778062515,\n",
       " 'spanish': 0.5451716674060038,\n",
       " 'child': 0.5768576972850118,\n",
       " 'michael': 0.5980892495458172,\n",
       " 'haneke': 0.49585532406853855,\n",
       " 'eli': 0.5056548985409117,\n",
       " 'roth': 0.5055757426968781,\n",
       " 'meaning': 0.5445118481656862,\n",
       " 'basically': 0.6697418002157685,\n",
       " 'arty': 0.5107662316441314,\n",
       " 'torture': 0.4464563041861981,\n",
       " 'tranquilizers': 0.5543305617007922,\n",
       " 'hur': 0.5507665626764925,\n",
       " 'epic': 0.6307281308298197,\n",
       " 'memory': 0.49617911449168783,\n",
       " 'charlton': 0.5175735486284989,\n",
       " 'heston': 0.5175822190880331,\n",
       " 'chariot': 0.5030800857285231,\n",
       " 'race': 0.526752630669231,\n",
       " 'remains': 0.5576483098381959,\n",
       " 'lanthimoss': 0.8728210316242309,\n",
       " 'terrifically': 0.5415046543771426,\n",
       " 'twisted': 0.7529049540913403,\n",
       " 'satire': 0.7459496248216115,\n",
       " 'places': 0.6217605467667767,\n",
       " 'fear': 0.5186138875050325,\n",
       " 'tread': 0.5674480082033037,\n",
       " 'symbolism': 0.48457770843672604,\n",
       " 'humanized': 0.5868963794921016,\n",
       " 'top': 0.4789576624050463,\n",
       " 'flight': 0.6030396370929823,\n",
       " 'brutal': 0.5658970585110641,\n",
       " 'rapturously': 0.9999999999998187,\n",
       " 'romantic': 0.6151866238631571,\n",
       " 'same': 0.4439418295401142,\n",
       " 'time': 0.5655154275933878,\n",
       " 'despite': 0.4682320595957364,\n",
       " 'shortcomings': 0.5123677989470065,\n",
       " 'happy': 0.6388303440173763,\n",
       " 'feet': 0.6124095369727822,\n",
       " 'brings': 0.590545877285971,\n",
       " 'lot': 0.551670654070492,\n",
       " 'energy': 0.6136470246050516,\n",
       " 'creativity': 0.48740399682797564,\n",
       " 'increasingly': 0.5301932906989758,\n",
       " 'relies': 0.5346497740889052,\n",
       " 'over': 0.5869799760579812,\n",
       " 'stimulation': 0.44032464333648996,\n",
       " 'blythe': 0.5812102920801118,\n",
       " 'danner': 0.5816351632998965,\n",
       " 'lands': 0.5842385486673831,\n",
       " 'her': 0.5230680253680646,\n",
       " 'screen': 0.5792900688113782,\n",
       " 'role': 0.5339309237458462,\n",
       " 'retired': 0.5753524676021566,\n",
       " 'widow': 0.6032409952079184,\n",
       " 'decides': 0.5858650702757638,\n",
       " 'dip': 0.5243842179757459,\n",
       " 'tentative': 0.3912516503709968,\n",
       " 'toe': 0.6836112427239874,\n",
       " 'dating': 0.4912262218793229,\n",
       " 'pool': 0.5124498241875204,\n",
       " 'indie': 0.5873459806587684,\n",
       " 'ill': 0.5705172490958789,\n",
       " 'dreams': 0.5750317587429098,\n",
       " 'truth': 0.5194117163561827,\n",
       " 'beauty': 0.5136850443004181,\n",
       " 'theyre': 0.5468175075029565,\n",
       " 'necessarily': 0.5433728701334057,\n",
       " 'nine': 0.4537768099868705,\n",
       " 'committed': 0.538782415241875,\n",
       " 'relationship': 0.5614436936563973,\n",
       " 'kids': 0.4943889556046255,\n",
       " 'backseat': 0.5640663289312284,\n",
       " 'blatantly': 0.4259356491631329,\n",
       " 'hearted': 0.7591468922644623,\n",
       " 'half': 0.5194323547408413,\n",
       " 'vibrant': 0.5453979109588598,\n",
       " 'original': 0.563741249014862,\n",
       " 'wired': 0.5558300010587328,\n",
       " 'article': 0.5060908699729552,\n",
       " 'myopically': 0.5735334304975784,\n",
       " 'neglects': 0.5475582677116216,\n",
       " 'address': 0.6029010639838931,\n",
       " 'arizonas': 0.5199350054775292,\n",
       " 'troubling': 0.640944204732455,\n",
       " 'anti': 0.4923919193595002,\n",
       " 'immigration': 0.5245982437721285,\n",
       " 'legislation': 0.5607589524060806,\n",
       " 'through': 0.5427334883134418,\n",
       " 'even': 0.5281260429305502,\n",
       " 'splash': 0.5706602111458512,\n",
       " 'hindsight': 0.5781525202476762,\n",
       " 'moulin': 0.5008296727121516,\n",
       " 'rouge': 0.487658205267243,\n",
       " 'le': 0.5869231128042506,\n",
       " 'queda': 0.6043994167090714,\n",
       " 'chica': 0.5435546382255182,\n",
       " 'la': 0.6114274561442224,\n",
       " 'pantalla': 0.5976761700925862,\n",
       " 'slo': 0.6005647298315847,\n",
       " 'satura': 0.47229107327419184,\n",
       " 'desborda': 0.5636595913311582,\n",
       " 'es': 0.6247160068154567,\n",
       " 'una': 0.5635342850490606,\n",
       " 'fiesta': 0.534441946389439,\n",
       " 'de': 0.6232331721889385,\n",
       " 'color': 0.48681196451000186,\n",
       " 'msica': 0.6939281009482072,\n",
       " 'heartwarming': 0.7344194117674856,\n",
       " 'message': 0.583927029400249,\n",
       " 'combined': 0.4888308718211465,\n",
       " 'creative': 0.5453366529583568,\n",
       " 'fantastical': 0.5562486871811916,\n",
       " 'elements': 0.5433937690228027,\n",
       " 'memorable': 0.5205225125431782,\n",
       " 'contraband': 0.5197232939658467,\n",
       " 'thriller': 0.6051560501153941,\n",
       " 'offers': 0.5937423276011033,\n",
       " 'way': 0.5723196476894638,\n",
       " 'assemble': 0.6028444227512929,\n",
       " 'minute': 0.523173947281817,\n",
       " 'trailer': 0.5335860390659672,\n",
       " 'death': 0.6395230409635729,\n",
       " 'smoochy': 0.6078499066008873,\n",
       " 'odd': 0.5494481883477395,\n",
       " 'im': 0.45418074187900714,\n",
       " 'almost': 0.5579490573319681,\n",
       " 'tempted': 0.46634597334003675,\n",
       " 'recommend': 0.4180484824380536,\n",
       " 'stardust': 0.5592172540115724,\n",
       " 'note': 0.5477889913902059,\n",
       " 'pizzazz': 0.5780161735988766,\n",
       " 'zero': 0.5516872627353302,\n",
       " 'substance': 0.5126932608577216,\n",
       " 'thoughtful': 0.7928467439332569,\n",
       " 'motion': 0.49108956371936024,\n",
       " 'picture': 0.5490952538298189,\n",
       " 'ponders': 0.6106536178202117,\n",
       " 'mysteries': 0.5373122842156651,\n",
       " 'accessible': 0.46575336767544406,\n",
       " 'emotionally': 0.6931238823576547,\n",
       " 'give': 0.6267158337373051,\n",
       " 'tati': 0.5063884172212546,\n",
       " 'posthumous': 0.5002757600561786,\n",
       " 'chance': 0.5231728907778866,\n",
       " 'express': 0.5655698380090066,\n",
       " 'sad': 0.5017434701434326,\n",
       " 'chomets': 0.6624121777317264,\n",
       " 'only': 0.5310962388616282,\n",
       " 'true': 0.6082906711967314,\n",
       " 'course': 0.5616402516001338,\n",
       " 'honest': 0.595848056469553,\n",
       " 'beautifully': 0.8678655372092481,\n",
       " 'thoughtfully': 0.562906285324179,\n",
       " 'realized': 0.8340128492607886,\n",
       " 'promising': 0.5088350609982726,\n",
       " 'quite': 0.5373299349058587,\n",
       " 'ultimately': 0.6004291078820784,\n",
       " 'underwhelming': 0.462546219999249,\n",
       " 'uma': 0.4781077267217203,\n",
       " 'daquelas': 0.48308430456704515,\n",
       " 'obras': 0.5564843563431364,\n",
       " 'que': 0.611246779430524,\n",
       " 'por': 0.5302852735147726,\n",
       " 'mais': 0.5104247690687679,\n",
       " 'eu': 0.6120220709410436,\n",
       " 'tenha': 0.5289708870203051,\n",
       " 'apreciado': 0.5623810027522462,\n",
       " 'sei': 0.5626779684634243,\n",
       " 'jamais': 0.6053044868599087,\n",
       " 'voltarei': 0.5558482835223033,\n",
       " 'assistir': 0.6559442470308214,\n",
       " 'e': 0.4921318404816562,\n",
       " 'isto': 0.6561830935028045,\n",
       " 'acreditem': 0.5665577519688905,\n",
       " 'um': 0.5416141994464065,\n",
       " 'elo': 0.5672476462552274,\n",
       " 'music': 0.5037502982452632,\n",
       " 'otherwise': 0.5476440741257449,\n",
       " 'shallow': 0.584548398149331,\n",
       " 'contrived': 0.5294196522331667,\n",
       " 'melodrama': 0.5263248874661728,\n",
       " 'al': 0.6210380668289531,\n",
       " 'daradji': 0.6068956740297348,\n",
       " 'directs': 0.5478457143896664,\n",
       " 'sure': 0.5770931013765965,\n",
       " 'hand': 0.5807446913051504,\n",
       " 'avoiding': 0.6285016975066515,\n",
       " 'icy': 0.5081071459692234,\n",
       " 'realism': 0.551215628938003,\n",
       " 'chest': 0.49418256506708363,\n",
       " 'beating': 0.5767990521359464,\n",
       " 'result': 0.6168539475739133,\n",
       " 'warm': 0.5327461649738139,\n",
       " 'human': 0.6238217698125184,\n",
       " 'quietly': 0.5603009790276875,\n",
       " 'devastating': 0.8979303339952307,\n",
       " 'coming': 0.4998375725241702,\n",
       " 'mile': 0.5462350239928981,\n",
       " 'away': 0.5116705572431788,\n",
       " 'then': 0.5010619105532318,\n",
       " 'id': 0.6135157889387459,\n",
       " 'send': 0.5074680173628396,\n",
       " 'bank': 0.5244992511964766,\n",
       " 'account': 0.5510763574893827,\n",
       " 'number': 0.5160862121032033,\n",
       " 'access': 0.5072655085071903,\n",
       " 'codes': 0.4910404117933071,\n",
       " 'deposit': 0.5738776478282631,\n",
       " 'special': 0.48513766439708256,\n",
       " 'prize': 0.6135774316054118,\n",
       " 'day': 0.5703297318392598,\n",
       " 'turns': 0.5059955236069974,\n",
       " 'enjoying': 0.5587234999364191,\n",
       " 'traditional': 0.6238008477378615,\n",
       " 'admonishment': 0.6505533952606114,\n",
       " 'stop': 0.5447953981602779,\n",
       " 'wasting': 0.5410945587324691,\n",
       " 'living': 0.6089558413109498,\n",
       " 'enjoy': 0.4676346710673644,\n",
       " 'single': 0.5472231969086414,\n",
       " 'moment': 0.5683855116029864,\n",
       " 'anyone': 0.5004428448125392,\n",
       " 'interest': 0.5294865023385753,\n",
       " 'dance': 0.5124653892373503,\n",
       " 'documentary': 0.5152862854082335,\n",
       " 'dumbed': 0.5211361015758768,\n",
       " 'narration': 0.5212152302080576,\n",
       " 'tim': 0.5875625955535632,\n",
       " 'allen': 0.5092974118589729,\n",
       " 'telegraphs': 0.48235137474689305,\n",
       " 'second': 0.5456800534390674,\n",
       " 'bit': 0.5181287641969492,\n",
       " 'average': 0.5205091960223702,\n",
       " 'iii': 0.5044794399550375,\n",
       " 'iv': 0.40469410913460624,\n",
       " 'vwhat': 0.5495126097905391,\n",
       " 'possibly': 0.7098108087028394,\n",
       " 'rocky': 0.5045267330565284,\n",
       " 'vs': 0.5433577459287051,\n",
       " 'among': 0.6021206665290041,\n",
       " 'holocaust': 0.6090268820884485,\n",
       " 'agnieszka': 0.6301127028623963,\n",
       " 'hollands': 0.6566268734862973,\n",
       " 'academy': 0.465649457742731,\n",
       " 'award': 0.6605718354609159,\n",
       " 'nominated': 0.553782502628402,\n",
       " 'darkness': 0.5346985930661803,\n",
       " 'during': 0.5187524316479417,\n",
       " 'world': 0.5350308335984678,\n",
       " 'ii': 0.5254718066630927,\n",
       " 'small': 0.6066658590263466,\n",
       " 'town': 0.5695435378207844,\n",
       " 'poland': 0.5385922175944571,\n",
       " 'named': 0.52469596496572,\n",
       " 'lvov': 0.5663568493542929,\n",
       " 'ukraine': 0.6097433489011638,\n",
       " 'propulsive': 0.5765016110753391,\n",
       " 'punk': 0.6345974930994565,\n",
       " 'rock': 0.5896904970180379,\n",
       " 'soundtrack': 0.6860266417261466,\n",
       " 'rough': 0.6405434615302684,\n",
       " 'cinematography': 0.5249337772485353,\n",
       " 'dragonslayer': 0.5163211173611685,\n",
       " 'care': 0.547125714017691,\n",
       " 'scrawny': 0.5667689420950799,\n",
       " 'young': 0.5319699570619725,\n",
       " 'skating': 0.5421248279600331,\n",
       " 'nowhere': 0.47926614631552605,\n",
       " 'viewers': 0.5378500059243452,\n",
       " 'wont': 0.5190355148802557,\n",
       " 'feel': 0.5770166846632376,\n",
       " 'theyve': 0.5334851933612913,\n",
       " 'dealt': 0.5508765129984236,\n",
       " 'cruelly': 0.502584237702639,\n",
       " 'pirates': 0.5539837012021681,\n",
       " 'defining': 0.6008448067717193,\n",
       " 'wondermentbut': 0.49635530723786664,\n",
       " 'tap': 0.6687083119640321,\n",
       " 'essential': 0.5990138489442978,\n",
       " 'jolt': 0.5343050597558701,\n",
       " 'satisfaction': 0.6243874572612655,\n",
       " 'missing': 0.5342489783543869,\n",
       " 'pike': 0.5735934492965683,\n",
       " 'gives': 0.5045799316061819,\n",
       " 'thoroughly': 0.5247085582324728,\n",
       " 'convincing': 0.5514355495863258,\n",
       " 'colvin': 0.7434734285235729,\n",
       " 'drama': 0.598554108207715,\n",
       " 'though': 0.5673416718443198,\n",
       " 'dimensional': 0.5165326895286177,\n",
       " 'filmmakers': 0.4962180752270662,\n",
       " 'focused': 0.4973162124639129,\n",
       " 'dont': 0.5137442567090487,\n",
       " 'notice': 0.5920426351544573,\n",
       " 'chadwick': 0.7464180525793914,\n",
       " 'boseman': 0.8186134406171528,\n",
       " 'precise': 0.5066024422346701,\n",
       " 'impersonation': 0.46994835606344154,\n",
       " 'browns': 0.612573829432012,\n",
       " 'gravelly': 0.5767080747601818,\n",
       " 'voice': 0.6432344726898168,\n",
       " 'moves': 0.5920327582835287,\n",
       " 'lends': 0.5939037193132295,\n",
       " 'emotional': 0.5983871743002493,\n",
       " 'weight': 0.48119468896077694,\n",
       " 'scenes': 0.4925253913189252,\n",
       " 'deserve': 0.5484123958590886,\n",
       " 'however': 0.43487616618345115,\n",
       " 'feels': 0.4887594689571211,\n",
       " 'series': 0.42817727926704086,\n",
       " 'bullet': 0.5490164592545526,\n",
       " 'points': 0.5163960668206773,\n",
       " 'organic': 0.5949019374911291,\n",
       " 'fluffed': 0.5282641320215159,\n",
       " 'satirical': 0.6316235371809172,\n",
       " 'targets': 0.5295383033125051,\n",
       " 'shortfall': 0.5380771507684677,\n",
       " 'required': 0.5588583514602645,\n",
       " 'rom': 0.43644075431371493,\n",
       " 'com': 0.48944431835785646,\n",
       " 'cheap': 0.46979822835165364,\n",
       " 'gags': 0.5141504444695731,\n",
       " 'unpleasant': 0.5231057103473964,\n",
       " 'hero': 0.5934975773264002,\n",
       " 'situate': 0.5971213362726819,\n",
       " 'friends': 0.62705592263459,\n",
       " 'several': 0.6039297490580015,\n",
       " 'floors': 0.4802644328048319,\n",
       " 'below': 0.5681635064128467,\n",
       " 'desired': 0.5916698581688153,\n",
       " 'apartment': 0.6051549667525612,\n",
       " 'esque': 0.5273534368742018,\n",
       " 'heights': 0.6167037974411416,\n",
       " 'wonder': 0.5044995226390555,\n",
       " 'wheel': 0.4607098868105011,\n",
       " 'kate': 0.5332927052231461,\n",
       " 'winslet': 0.5332350411723729,\n",
       " 'pluses': 0.46880861396022233,\n",
       " 'plenty': 0.5700884383710622,\n",
       " 'laughs': 0.484496836855015,\n",
       " 'outrageous': 0.6359180500365128,\n",
       " 'situations': 0.5331587973514919,\n",
       " 'guys': 0.5572345662173707,\n",
       " 'themselves': 0.49865923735978257,\n",
       " 'lawrence': 0.4993484584003821,\n",
       " 'luxuriates': 0.5992637605229856,\n",
       " 'tremendous': 0.7537453594947889,\n",
       " 'hits': 0.5044875571689277,\n",
       " 'spot': 0.5575874432319222,\n",
       " 'entertaining': 0.7997267116702883,\n",
       " 'informative': 0.8342755323963998,\n",
       " 'deeply': 0.4731461281109546,\n",
       " 'moving': 0.5610403936120323,\n",
       " 'richly': 0.8646799873218991,\n",
       " 'appointed': 0.6032867994542856,\n",
       " 'meant': 0.3923071199389512,\n",
       " 'titular': 0.5089371627894722,\n",
       " 'objet': 0.5783826283017235,\n",
       " 'dart': 0.5512971919403531,\n",
       " 'adult': 0.5105814612216258,\n",
       " 'beginners': 0.48176318519598643,\n",
       " 'baby': 0.44139321176810115,\n",
       " 'step': 0.5258541029046441,\n",
       " 'figuring': 0.5558063319969361,\n",
       " 'kroll': 0.5384482909788046,\n",
       " 'transition': 0.5101611353035499,\n",
       " 'potential': 0.505059697330744,\n",
       " 'certainly': 0.6014132924068647,\n",
       " 'unremarkable': 0.513502740154687,\n",
       " 'rainmaker': 0.5690013451802882,\n",
       " 'other': 0.5101715368215168,\n",
       " 'astonishing': 0.5445024024143316,\n",
       " 'complex': 0.7097721294405641,\n",
       " 'dismissed': 0.5338101283679666,\n",
       " 'bright': 0.5233294735078402,\n",
       " 'terminally': 0.5357387835975428,\n",
       " 'conscious': 0.5279912021741967,\n",
       " 'hoary': 0.5546324572558726,\n",
       " 'conceits': 0.5841084426541714,\n",
       " 'ancient': 0.4235507545767011,\n",
       " 'vintage': 0.5821979702005246,\n",
       " 'dubious': 0.5295332704044546,\n",
       " 'worth': 0.4954489889220662,\n",
       " 'riotously': 0.5701718467694032,\n",
       " 'extremely': 0.489778371771794,\n",
       " 'abundantly': 0.5724882005259088,\n",
       " 'given': 0.5222888647466407,\n",
       " 'monumental': 0.6503712985209045,\n",
       " 'task': 0.47844239258592386,\n",
       " 'these': 0.5563160465239839,\n",
       " 'managed': 0.5828598737965233,\n",
       " 'tell': 0.4903017319602527,\n",
       " 'grow': 0.43314223921267264,\n",
       " 'band': 0.6909421617175329,\n",
       " 'fascinating': 0.6884056958520665,\n",
       " 'tactfully': 0.9999999999994111,\n",
       " 'charts': 0.7049939414557546,\n",
       " 'sort': 0.5941833912158472,\n",
       " 'philosophical': 0.5423831111639543,\n",
       " 'differences': 0.5545738564402453,\n",
       " 'eventually': 0.4347581359424819,\n",
       " 'undermine': 0.5248910374727759,\n",
       " 'any': 0.516319435449221,\n",
       " 'group': 0.511999681945344,\n",
       " 'endeavor': 0.523076509858099,\n",
       " ...}"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "probability['fresh'] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "5110f40d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'bendingers': 0.41668431070594164,\n",
       " 'film': 0.40444062281277743,\n",
       " 'is': 0.4439944693085865,\n",
       " 'a': 0.4450023338842141,\n",
       " 'mess': 0.8159334851489483,\n",
       " 'thats': 0.48038084934226216,\n",
       " 'really': 0.5464291320071654,\n",
       " 'all': 0.49348829652288234,\n",
       " 'there': 0.4768306451116068,\n",
       " 'to': 0.4683386385143362,\n",
       " 'it': 0.4211245781637825,\n",
       " 'marked': 0.4751464007112334,\n",
       " 'by': 0.4882501963447964,\n",
       " 'tossed': 0.42955469043829797,\n",
       " 'off': 0.5157975834236534,\n",
       " 'feeling': 0.5434653604000241,\n",
       " 'distinct': 0.4051163562780895,\n",
       " 'impression': 0.4059454764466585,\n",
       " 'that': 0.4534582771338573,\n",
       " 'americas': 0.27763895376613695,\n",
       " 'most': 0.33835431841235425,\n",
       " 'viable': 0.4584101945258358,\n",
       " 'auteur': 0.4264184795880892,\n",
       " 'brats': 0.4716190371560255,\n",
       " 'had': 0.49854431812750016,\n",
       " 'taken': 0.4430422844778085,\n",
       " 'an': 0.5366270876900431,\n",
       " 'unfamiliar': 0.45233865802719714,\n",
       " 'path': 0.4899935167965343,\n",
       " 'with': 0.5293363062176266,\n",
       " 'minimal': 0.46657490302303145,\n",
       " 'investment': 0.4817868725815837,\n",
       " 'for': 0.41633903769184377,\n",
       " 'uncertain': 0.5013599331586087,\n",
       " 'gain': 0.34836932637319656,\n",
       " 'how': 0.4608503414553421,\n",
       " 'would': 0.4475634388635215,\n",
       " 'story': 0.45270884664642314,\n",
       " 'written': 0.4031634941338805,\n",
       " 'before': 0.47768106096547747,\n",
       " 'we': 0.511293674186854,\n",
       " 'made': 0.4544956570235161,\n",
       " 'the': 0.44495534636045525,\n",
       " 'moon': 0.4832502736855833,\n",
       " 'so': 0.48821052766574313,\n",
       " 'complicated': 0.39900538964418514,\n",
       " 'didnt': 0.46154097735518435,\n",
       " 'work': 0.4742267677654292,\n",
       " 'in': 0.44495534636045525,\n",
       " 'and': 0.6989079440992497,\n",
       " 'doesnt': 0.4533910561658735,\n",
       " 'now': 0.503446560088513,\n",
       " 'classic': 0.39594646167845327,\n",
       " 'moments': 0.4400248214725379,\n",
       " 'are': 0.4768306451116068,\n",
       " 'crammed': 0.5014721441688145,\n",
       " 'as': 0.4924326706529886,\n",
       " 'if': 0.4747329659645889,\n",
       " 'might': 0.5052279324992635,\n",
       " 'not': 0.4553594894447476,\n",
       " 'get': 0.4933173955973795,\n",
       " 'another': 0.5282430714495482,\n",
       " 'shot': 0.3978976680092731,\n",
       " 'at': 0.343185444004885,\n",
       " 'this': 0.47981096275965684,\n",
       " 'its': 0.3604769590364271,\n",
       " 'very': 0.4906610953020642,\n",
       " 'sweet': 0.3382564867436367,\n",
       " 'charming': 0.48033507138333387,\n",
       " 'should': 0.5232736975914828,\n",
       " 'be': 0.47980645060445504,\n",
       " 'thankful': 0.5085281994214571,\n",
       " 'isnt': 0.5731990059591593,\n",
       " 'childhood': 0.5085872477286613,\n",
       " 'ruining': 0.6072892683299431,\n",
       " 'disaster': 0.3599363509259658,\n",
       " 'mulholland': 0.563002403216243,\n",
       " 'drive': 0.48794943421310333,\n",
       " 'maddeningly': 0.2638264819386462,\n",
       " 'abstract': 0.3961094634381994,\n",
       " 'inscrutable': 0.4821616906862989,\n",
       " 'on': 0.4891544146059927,\n",
       " 'first': 0.37765147281378575,\n",
       " 'viewing': 0.4446886269726618,\n",
       " 'but': 0.44694766913682227,\n",
       " 'hypnotic': 0.4547859848702628,\n",
       " 'haunting': 0.40564051175251636,\n",
       " 'every': 0.48003129619412455,\n",
       " 'after': 0.4630255130700792,\n",
       " 'you': 0.4630255130700792,\n",
       " 'fixate': 0.48510688959050163,\n",
       " 'funny': 0.4309322680964714,\n",
       " 'things': 0.4454993667730301,\n",
       " 'like': 0.5082431619361527,\n",
       " 'plot': 0.4655218017822205,\n",
       " 'character': 0.5150756885677443,\n",
       " 'movies': 0.49154041879501337,\n",
       " 'anyway': 0.4589470027472478,\n",
       " 'too': 0.49216832719578096,\n",
       " 'bad': 0.47568691784039047,\n",
       " 'jason': 0.43999255900096157,\n",
       " 'reitman': 0.3989972542734209,\n",
       " 'does': 0.4879916963132382,\n",
       " 'again': 0.42987997818152146,\n",
       " 'although': 0.44834577034902756,\n",
       " 'full': 0.46379302033423175,\n",
       " 'of': 0.40444062281277743,\n",
       " 'product': 0.5317824752337524,\n",
       " 'placements': 0.48716483429089363,\n",
       " 'dark': 0.3889760344256338,\n",
       " 'comedy': 0.48726721721402266,\n",
       " 'surprisingly': 0.2326329723293727,\n",
       " 'enjoyable': 0.48179332125428403,\n",
       " 'highlighted': 0.47557844813069594,\n",
       " 'anna': 0.46573935265675037,\n",
       " 'kendrick': 0.4453195552064968,\n",
       " 'who': 0.47045868955451203,\n",
       " 'me': 0.47008030247265115,\n",
       " 'laugh': 0.4781594692894181,\n",
       " 'out': 0.4919119068199631,\n",
       " 'loud': 0.5567089955603498,\n",
       " 'willis': 0.39063472460247844,\n",
       " 'appears': 0.40496929084961003,\n",
       " 'have': 0.4999298090958755,\n",
       " 'finally': 0.47397258903934736,\n",
       " 'grown': 0.4789128956615161,\n",
       " 'bored': 0.49662070640324324,\n",
       " 'his': 0.45808335455770927,\n",
       " 'own': 0.40579186979641213,\n",
       " 'shtick': 0.4560383462489788,\n",
       " 'malkovich': 0.35097771623476687,\n",
       " 'spends': 0.5130069055498034,\n",
       " 'movie': 0.5293363062176266,\n",
       " 'looking': 0.4451474351282099,\n",
       " 'humiliated': 0.44176430371639314,\n",
       " 'love': 0.3316835970497302,\n",
       " 'my': 0.48733543925110123,\n",
       " 'eye': 0.4721576145440961,\n",
       " 'i': 0.4744678735939035,\n",
       " 'liked': 0.4673068761587196,\n",
       " 'see': 0.492752564693201,\n",
       " 'little': 0.526751273241531,\n",
       " 'more': 0.48363720873940247,\n",
       " 'cris': 0.42752064361310976,\n",
       " 'abilities': 0.4484138089073789,\n",
       " 'future': 0.41781550510062443,\n",
       " 'than': 0.4907308895703483,\n",
       " 'just': 0.5298926369668746,\n",
       " 'sappy': 0.5589716906443186,\n",
       " 'ambitious': 0.3381670823378204,\n",
       " 'passionate': 0.41060870941223837,\n",
       " 'grief': 0.3906432584003323,\n",
       " 'stricken': 0.4110467895730674,\n",
       " 'art': 0.34074536983711157,\n",
       " 'marc': 0.4595159299399224,\n",
       " 'forsteris': 0.5094060303357442,\n",
       " 'constrained': 0.4530770167203406,\n",
       " 'reality': 0.39107308401604285,\n",
       " 'left': 0.5402165662758666,\n",
       " 'lacks': 0.45496453649784163,\n",
       " 'much': 0.4325321489737858,\n",
       " 'satisfying': 0.49870676583631923,\n",
       " 'climax': 0.35489118888040966,\n",
       " 'or': 0.49915022344176346,\n",
       " 'anything': 0.4790613871165489,\n",
       " 'proper': 0.37532556897581926,\n",
       " 'ending': 0.45834988037187935,\n",
       " 'target': 0.49318489250930253,\n",
       " 'audience': 0.47980645060445504,\n",
       " 'overly': 0.48944111845637733,\n",
       " 'horny': 0.49716343982104466,\n",
       " 'teenage': 0.24233194635761385,\n",
       " 'boys': 0.5178652818689945,\n",
       " 'want': 0.44516364782424656,\n",
       " 'female': 0.47776883806315357,\n",
       " 'nudity': 0.4666485973100526,\n",
       " 'graphic': 0.4284505549444003,\n",
       " 'violence': 0.4303857570348942,\n",
       " 'no': 0.5542213120638884,\n",
       " 'real': 0.48132236696805136,\n",
       " 'ray': 0.4058549495775436,\n",
       " 'winstones': 0.39825595720667023,\n",
       " 'irascible': 0.4730819759865129,\n",
       " 'english': 0.4573863122623982,\n",
       " 'copper': 0.46572478802518885,\n",
       " 'blows': 0.4699575458210742,\n",
       " 'steam': 0.4460254642101514,\n",
       " 'ears': 0.4741579319645386,\n",
       " 'ashley': 0.4819183739443617,\n",
       " 'jensen': 0.42576397504119395,\n",
       " 'chews': 0.4662462861454306,\n",
       " 'whatever': 0.5007856584437272,\n",
       " 'scenery': 0.5984402269302972,\n",
       " 'hasnt': 0.4559526583496016,\n",
       " 'already': 0.491094371150638,\n",
       " 'been': 0.5465679857181784,\n",
       " 'trampled': 0.5053466231160805,\n",
       " 'pacing': 0.41629034196538794,\n",
       " 'slow': 0.480142655801362,\n",
       " 'intensity': 0.5077646660229289,\n",
       " 'needs': 0.5645212438558799,\n",
       " 'cranked': 0.4935540884506218,\n",
       " 'up': 0.5130646911413145,\n",
       " 'notch': 0.3606915750380406,\n",
       " 'two': 0.48040391203607463,\n",
       " 'has': 0.48859743262137373,\n",
       " 'emotion': 0.4130368988050282,\n",
       " 'your': 0.46081057929974223,\n",
       " 'typical': 0.4382182832441168,\n",
       " 'horror': 0.4942526494832189,\n",
       " 'stories': 0.3731443166881108,\n",
       " 'problems': 0.48096621228041087,\n",
       " 'arthur': 0.4664884003383501,\n",
       " 'mostly': 0.4653534241797885,\n",
       " 'ones': 0.4826187162455378,\n",
       " 'scale': 0.3852520117749664,\n",
       " 'set': 0.4864832907039143,\n",
       " 'pieces': 0.5012919105639753,\n",
       " 'big': 0.4857012434147991,\n",
       " 'becomes': 0.48961858642658107,\n",
       " 'labored': 0.5101443213279372,\n",
       " 'cast': 0.4477658275760074,\n",
       " 'they': 0.48710403025140786,\n",
       " 'can': 0.4922780905027012,\n",
       " 'end': 0.511293674186854,\n",
       " 'dead': 0.4789411215931051,\n",
       " 'man': 0.347023688962344,\n",
       " 'down': 0.5083423878053729,\n",
       " 'plays': 0.45180845734460906,\n",
       " 'confused': 0.45293441010035757,\n",
       " 'noble': 0.5194813103306752,\n",
       " 'misfire': 0.7959999724249411,\n",
       " 'energetic': 0.42332556953954953,\n",
       " 'slapstick': 0.5217431104765692,\n",
       " 'aimed': 0.42924432640842963,\n",
       " 'indiscriminating': 0.41264602544972684,\n",
       " 'youngsters': 0.4912949540021245,\n",
       " 'will': 0.47980645060445504,\n",
       " 'buy': 0.6198515823452514,\n",
       " 'lots': 0.46726150506369796,\n",
       " 'crood': 0.4934167712939787,\n",
       " 'toys': 0.34639501148742435,\n",
       " 'one': 0.4647258732132945,\n",
       " 'van': 0.41692914844961393,\n",
       " 'dammes': 0.44943641665551715,\n",
       " 'best': 0.2525877940557824,\n",
       " 'flicksgood': 0.37080355464185427,\n",
       " 'good': 0.4594335233451481,\n",
       " 'skirmishes': 0.5216977531147858,\n",
       " 'masterfully': 0.5078183749601244,\n",
       " 'directed': 0.360024806134183,\n",
       " 'xavier': 0.4349371646892115,\n",
       " 'beauvois': 0.3819389862107449,\n",
       " 'hands': 0.3868718183695926,\n",
       " 'euthanasia': 0.48572362490668103,\n",
       " 'ever': 0.4567933613864287,\n",
       " 'films': 0.4926549611928493,\n",
       " 'humor': 0.5021663238908097,\n",
       " 'sauce': 0.45669977387769095,\n",
       " 'elevate': 0.38976376381821976,\n",
       " 'from': 0.49570243658860574,\n",
       " 'whining': 0.4941026593658971,\n",
       " 'self': 0.4581903584501473,\n",
       " 'absorbed': 0.4964638182880602,\n",
       " 'unattached': 0.46544115319579593,\n",
       " 'women': 0.4957117308883133,\n",
       " 'populate': 0.42723419045372985,\n",
       " 'many': 0.4528242392393255,\n",
       " 'books': 0.47143185548452327,\n",
       " 'today': 0.4540414787519412,\n",
       " 'dialogue': 0.550527441369604,\n",
       " 'flat': 0.46551697279235565,\n",
       " 'animation': 0.377870711093939,\n",
       " 'workmanlike': 0.5057229924138851,\n",
       " 'quirky': 0.3499959044140627,\n",
       " 'pacy': 0.4575558198757038,\n",
       " 'enough': 0.4842178610348772,\n",
       " 'keep': 0.4937196140863699,\n",
       " 'younger': 0.46184140258823275,\n",
       " 'children': 0.4795787031210253,\n",
       " 'quiet': 0.40886151079362637,\n",
       " 'easily': 0.49536586426326984,\n",
       " 'ryan': 0.4355118981488805,\n",
       " 'reynolds': 0.48926252445859725,\n",
       " 'performance': 0.42259810992991975,\n",
       " 'since': 0.47721943311102183,\n",
       " 'buried': 0.714155923733534,\n",
       " 'ben': 0.4700789346446692,\n",
       " 'mendelsohn': 0.44850402220866564,\n",
       " 'portrays': 0.46699292866376774,\n",
       " 'flawed': 0.48702888004662237,\n",
       " 'helpless': 0.42332813107090067,\n",
       " 'individuals': 0.45374733461484446,\n",
       " 'incredibly': 0.4712402601518801,\n",
       " 'well': 0.47325619357762155,\n",
       " 'when': 0.48980825278669166,\n",
       " 'take': 0.44474287020168046,\n",
       " 'into': 0.4919631543289321,\n",
       " 'consideration': 0.48121628334940797,\n",
       " 'mississippi': 0.46491042100101876,\n",
       " 'grind': 0.498849952429466,\n",
       " 'addictive': 0.4137382371712464,\n",
       " 'gambling': 0.6294156002020035,\n",
       " 'storyline': 0.47253163579125035,\n",
       " 'involving': 0.46964539907556596,\n",
       " 'visually': 0.3608298533830054,\n",
       " 'stunning': 0.2824745974758755,\n",
       " 'ive': 0.4523129529441606,\n",
       " 'seen': 0.4440385386872874,\n",
       " 'dearth': 0.28818007623897396,\n",
       " 'new': 0.4137041476615409,\n",
       " 'multiplex': 0.5278432765884665,\n",
       " 'dwayne': 0.5335567111770636,\n",
       " 'johnsons': 0.43787388334995153,\n",
       " 'draw': 0.48275407293452066,\n",
       " 'may': 0.40256835442376077,\n",
       " 'make': 0.4917315267641733,\n",
       " 'lucrative': 0.5210356054898894,\n",
       " 'opening': 0.44310178337312334,\n",
       " 'weekend': 0.44901047508752157,\n",
       " 'beyond': 0.4381609538803796,\n",
       " 'insidious': 0.5012330506557672,\n",
       " 'chapter': 0.4106188901358447,\n",
       " 'lack': 0.44853165868159006,\n",
       " 'unique': 0.4119736572757399,\n",
       " 'charm': 0.34106471492703044,\n",
       " 'both': 0.4719934313758479,\n",
       " 'predecessors': 0.4976010701060235,\n",
       " 'still': 0.4357329412154725,\n",
       " 'scary': 0.5169146126353292,\n",
       " 'fun': 0.38600846449294757,\n",
       " 'thrill': 0.42917548878254447,\n",
       " 'ride': 0.5017900234470981,\n",
       " 'delivers': 0.40076922214922794,\n",
       " 'some': 0.46263740179268636,\n",
       " 'serious': 0.3850123847371898,\n",
       " 'blood': 0.5360719029675964,\n",
       " 'pressure': 0.4728610813939713,\n",
       " 'raising': 0.5114733195056447,\n",
       " 'tension': 0.4519666318660898,\n",
       " 'intriguing': 0.4656287729931114,\n",
       " 'tale': 0.45513379912436475,\n",
       " 'about': 0.4340439777238785,\n",
       " 'individual': 0.38631761099662854,\n",
       " 'whose': 0.47808614469839306,\n",
       " 'mounting': 0.24801339074950235,\n",
       " 'debt': 0.46249438974383417,\n",
       " 'leads': 0.49002720301674907,\n",
       " 'unfortunate': 0.44083742422841005,\n",
       " 'circumstances': 0.46900399447248736,\n",
       " 'thai': 0.18467869807526233,\n",
       " 'arthouse': 0.4058369862110461,\n",
       " 'director': 0.4692203549879539,\n",
       " 'apichatpong': 0.4441478697719099,\n",
       " 'weerasethakul': 0.5339300252961027,\n",
       " 'uncle': 0.3946282675710853,\n",
       " 'boonmee': 0.4575234900633598,\n",
       " 'won': 0.38719332652744143,\n",
       " 'palme': 0.44560269160212274,\n",
       " 'dor': 0.31724040285110044,\n",
       " 'cannes': 0.24710176957995386,\n",
       " 'few': 0.4403732902165438,\n",
       " 'years': 0.41575994237337427,\n",
       " 'ago': 0.522880858056667,\n",
       " 'comes': 0.4419055049534044,\n",
       " 'strange': 0.3424292755761542,\n",
       " 'swooning': 0.43441742141613304,\n",
       " 'dream': 0.31665666455797686,\n",
       " 'cemetery': 0.5205385623478558,\n",
       " 'splendour': 0.4669712140923125,\n",
       " 'engaging': 0.4695073103017185,\n",
       " 'exciting': 0.4138685712814726,\n",
       " 'sci': 0.4642875587690258,\n",
       " 'fi': 0.46223488583188765,\n",
       " 'flick': 0.39849665770856424,\n",
       " 'bring': 0.4560004914784304,\n",
       " 'year': 0.4499996741397889,\n",
       " 'old': 0.4592218320431275,\n",
       " 'us': 0.3732841662626949,\n",
       " 'stepford': 0.7535126845615734,\n",
       " 'threat': 0.3215765109959515,\n",
       " 'tradition': 0.499819431378861,\n",
       " 'minded': 0.4110738225979281,\n",
       " 'men': 0.4397771886180963,\n",
       " 'corporate': 0.2734911531530009,\n",
       " 'mother': 0.4662524659211341,\n",
       " 'hens': 0.5235414280486705,\n",
       " 'such': 0.4197515074097995,\n",
       " 'martha': 0.44140097901214015,\n",
       " 'stewart': 0.4033429538959383,\n",
       " 'bake': 0.4051931472385804,\n",
       " 'their': 0.48595004759839355,\n",
       " 'cakes': 0.4723874732051528,\n",
       " 'eat': 0.5053248535462653,\n",
       " 'them': 0.4379048883606542,\n",
       " 'while': 0.4211245781637825,\n",
       " 'find': 0.5013407626402174,\n",
       " 'jarmuschs': 0.26454427795198643,\n",
       " 'call': 0.47076948781848466,\n",
       " 'poor': 0.4816923071220039,\n",
       " 'pretentious': 0.6442979989968871,\n",
       " 'others': 0.4280675764238752,\n",
       " 'adore': 0.4320563634604507,\n",
       " 'artistic': 0.2509066604514674,\n",
       " 'sensibilities': 0.48609200188466084,\n",
       " 'war': 0.3749661552629215,\n",
       " 'stalemate': 0.499000622590465,\n",
       " 'aleksandr': 5.888154608306692e-13,\n",
       " 'burovs': 0.32359684267143013,\n",
       " 'bleached': 0.5082450451547862,\n",
       " 'images': 0.3068617878916137,\n",
       " 'creating': 0.43167586784176937,\n",
       " 'alien': 0.522532307497067,\n",
       " 'landscape': 0.3747764655589224,\n",
       " 'which': 0.44495534636045525,\n",
       " 'colour': 0.4921232475202761,\n",
       " 'rare': 0.512507222368463,\n",
       " 'compassion': 0.40019552340494896,\n",
       " 'rarely': 0.44262061558555127,\n",
       " 'combat': 0.44917845953471347,\n",
       " 'seemed': 0.44088890501924505,\n",
       " 'savage': 0.36051723256600454,\n",
       " 'futile': 0.48721261336989086,\n",
       " 'nick': 0.47245489686252984,\n",
       " 'morans': 7.850218789678338e-13,\n",
       " 'based': 0.5337398453726696,\n",
       " 'stage': 0.4921532026626624,\n",
       " 'play': 0.4651312549767563,\n",
       " 'he': 0.5076472303855097,\n",
       " 'wrote': 0.4820119759104742,\n",
       " 'james': 0.4055837836121819,\n",
       " 'hicks': 0.47307319586442276,\n",
       " 'eccentric': 0.4713780572155753,\n",
       " 'sometimes': 0.46465921304003116,\n",
       " 'underpowered': 0.5009903010858499,\n",
       " 'always': 0.47162787955126384,\n",
       " 'watchable': 0.48085934567692334,\n",
       " 'early': 0.4840351325875599,\n",
       " 's': 0.37094014796770497,\n",
       " 'prehistory': 0.4213859313642211,\n",
       " 'pop': 0.39564303367149695,\n",
       " 'culture': 0.28125766087230875,\n",
       " 'garden': 0.46523014925936884,\n",
       " 'state': 0.5276775275455345,\n",
       " 'braff': 0.48912333963119914,\n",
       " 'takes': 0.4226570847193422,\n",
       " 'trials': 0.4842057051217022,\n",
       " 'tribulations': 0.45675978689764063,\n",
       " 'bloom': 0.5105686081134777,\n",
       " 'family': 0.35907104060939,\n",
       " 'life': 0.32783950443002646,\n",
       " 'ties': 0.4193183161471534,\n",
       " 'happily': 0.3620841139714046,\n",
       " 'finale': 0.42636394464476224,\n",
       " 'wish': 0.47661836282012987,\n",
       " 'was': 0.4714759930777141,\n",
       " 'here': 0.4684702295974482,\n",
       " 'light': 0.4408061397592055,\n",
       " 'entertainment': 0.3532786265952813,\n",
       " 'forces': 0.4047140188006189,\n",
       " 'itself': 0.3996638904178084,\n",
       " 'upon': 0.5213387642880183,\n",
       " 'part': 0.34992829913145,\n",
       " 'screenwriters': 0.4865384554073724,\n",
       " 'never': 0.4554346012771649,\n",
       " 'lose': 0.49669949459304685,\n",
       " 'sight': 0.4853781983920246,\n",
       " 'moral': 0.47244641783267055,\n",
       " 'else': 0.5433686552540371,\n",
       " 'gets': 0.42629762120982045,\n",
       " 'mix': 0.45078240398036007,\n",
       " 'shrek': 0.5163535981495352,\n",
       " 'must': 0.3741999484557595,\n",
       " 'heart': 0.4668594905416744,\n",
       " 'soul': 0.46495166998631954,\n",
       " 'myers': 0.5115621898917575,\n",
       " 'master': 0.37890588670004427,\n",
       " 'makes': 0.4579798918324195,\n",
       " 'seem': 0.45591930926241747,\n",
       " 'easy': 0.42159562854916244,\n",
       " 'being': 0.3808591496187546,\n",
       " 'green': 0.4565865414064812,\n",
       " 'disappointing': 0.5225999703310801,\n",
       " 'actor': 0.3701521643780795,\n",
       " 'peter': 0.4735926255663021,\n",
       " 'berg': 0.4467776158710481,\n",
       " 'thinks': 0.7393746210217063,\n",
       " 'clever': 0.34887476224244823,\n",
       " 'witty': 0.35778172022997984,\n",
       " 'whole': 0.444527303167758,\n",
       " 'headache': 0.4523465951695254,\n",
       " 'inducing': 0.6309809604663874,\n",
       " 'ordeal': 0.46355519333095785,\n",
       " 'hopefully': 0.48306391921377756,\n",
       " 'put': 0.5230684418178719,\n",
       " 'found': 0.5548203180527443,\n",
       " 'footage': 0.36728206730809776,\n",
       " 'sub': 0.4775687108605754,\n",
       " 'genre': 0.4804837419505427,\n",
       " 'kind': 0.4837204661062984,\n",
       " 'people': 0.4957057216798735,\n",
       " 'discover': 0.5023019486408662,\n",
       " 'hilarious': 0.30029410547548463,\n",
       " 'ways': 0.4693151288989832,\n",
       " 'mock': 0.4021142008290814,\n",
       " 'come': 0.4517389988697016,\n",
       " 'shallows': 0.294201797843683,\n",
       " 'often': 0.4872399374990905,\n",
       " 'ankle': 0.3680502492076622,\n",
       " 'deep': 0.4770385589115165,\n",
       " 'unable': 0.7239427475516163,\n",
       " 'past': 0.3777366914900452,\n",
       " 'snout': 0.465065812746744,\n",
       " 'theres': 0.5389870023316877,\n",
       " 'impressive': 0.2872623583425729,\n",
       " 'production': 0.4926687187278639,\n",
       " 'values': 0.39638873244628403,\n",
       " 'effective': 0.32456883658289754,\n",
       " 'star': 0.4817063426186101,\n",
       " 'turn': 0.49181169643126776,\n",
       " 'burtons': 0.4423967631458925,\n",
       " 'favorite': 0.5181387791293615,\n",
       " 'johnny': 0.481895484681042,\n",
       " 'depp': 0.4610135309346792,\n",
       " 'together': 0.47335792707184293,\n",
       " 'remarkably': 0.4255398421358695,\n",
       " 'dull': 0.8821521473489174,\n",
       " 'non': 0.42582290629640884,\n",
       " 'starter': 0.5336377485723142,\n",
       " 'could': 0.5100358792104084,\n",
       " 'least': 0.4808230962960356,\n",
       " 'b': 0.4552964379718062,\n",
       " 'grade': 0.5036975415884181,\n",
       " 'conviction': 0.3629656835614261,\n",
       " 'what': 0.4450023338842141,\n",
       " 'were': 0.7185908384630528,\n",
       " 'high': 0.4176262265937735,\n",
       " 'school': 0.44723826867708666,\n",
       " 'geek': 0.3862313040898995,\n",
       " 'porn': 0.5006311321232831,\n",
       " 'moved': 0.5370716062795003,\n",
       " 'next': 0.45164569644754515,\n",
       " 'door': 0.48589996218827314,\n",
       " 'answer': 0.5220345534832567,\n",
       " 'surely': 0.4034915864761864,\n",
       " 'outwit': 0.42039635777986345,\n",
       " 'cut': 0.47542255831227903,\n",
       " 'rate': 0.4074408846529743,\n",
       " 'risky': 0.42093353546489887,\n",
       " 'business': 0.48453920135231227,\n",
       " 'match': 0.3296742395535387,\n",
       " 'bar': 0.48049969918753394,\n",
       " 'royale': 0.4198821638784887,\n",
       " 'neither': 0.46725387819397346,\n",
       " 'let': 0.4982241416108197,\n",
       " 'franchise': 0.44892711132475194,\n",
       " 'fall': 0.44454326444251474,\n",
       " 'unduly': 0.5121714760220107,\n",
       " 'far': 0.4924326706529886,\n",
       " 'mark': 0.482678298467425,\n",
       " 'appealingly': 0.5317505376757604,\n",
       " 'cheesy': 0.4708485776308497,\n",
       " 'tribute': 0.4433710328946849,\n",
       " 'hope': 0.44820106515452757,\n",
       " 'springs': 0.5245324106654866,\n",
       " 'eternal': 0.5250251998349004,\n",
       " 'hopelessly': 0.7851590735329106,\n",
       " 'inept': 0.4444605641727548,\n",
       " 'photogenic': 0.4674633979203418,\n",
       " 'charismatic': 0.39477444484055585,\n",
       " 'earth': 0.4489332758100951,\n",
       " 'miss': 0.4793182844802258,\n",
       " 'quick': 0.46534508557299786,\n",
       " 'show': 0.49190204857676556,\n",
       " 'characters': 0.4689642359052134,\n",
       " 'skin': 0.4267436023511702,\n",
       " 'less': 0.49614925749882166,\n",
       " 'inclined': 0.4792079118931457,\n",
       " 'explore': 0.5194608680916923,\n",
       " 'lies': 0.4043208279365719,\n",
       " 'beneath': 0.3518957100379887,\n",
       " 'those': 0.4226793967687306,\n",
       " 'saw': 0.5105323901386312,\n",
       " 'animated': 0.42431260932805037,\n",
       " 'version': 0.5450511028934416,\n",
       " 'musical': 0.3672813686834035,\n",
       " 'numbers': 0.4263243498591667,\n",
       " 'yet': 0.3513487401220902,\n",
       " 'favreau': 0.468124726257176,\n",
       " 'manages': 0.44425557565295787,\n",
       " 'create': 0.4070865582193266,\n",
       " 'beautiful': 0.425850059005576,\n",
       " 'balanced': 0.4342210483368851,\n",
       " 'review': 0.45915372219374856,\n",
       " 'spanish': 0.4548283325939962,\n",
       " 'child': 0.4231423027149881,\n",
       " 'michael': 0.40191075045418284,\n",
       " 'haneke': 0.5041446759314615,\n",
       " 'eli': 0.49434510145908817,\n",
       " 'roth': 0.49442425730312195,\n",
       " 'meaning': 0.45548815183431374,\n",
       " 'basically': 0.3302581997842316,\n",
       " 'arty': 0.48923376835586874,\n",
       " 'torture': 0.5535436958138019,\n",
       " 'tranquilizers': 0.44566943829920785,\n",
       " 'hur': 0.44923343732350746,\n",
       " 'epic': 0.36927186917018034,\n",
       " 'memory': 0.5038208855083122,\n",
       " 'charlton': 0.48242645137150114,\n",
       " 'heston': 0.48241778091196696,\n",
       " 'chariot': 0.49691991427147686,\n",
       " 'race': 0.473247369330769,\n",
       " 'remains': 0.4423516901618041,\n",
       " 'lanthimoss': 0.1271789683757691,\n",
       " 'terrifically': 0.4584953456228575,\n",
       " 'twisted': 0.2470950459086598,\n",
       " 'satire': 0.25405037517838847,\n",
       " 'places': 0.3782394532332233,\n",
       " 'fear': 0.4813861124949676,\n",
       " 'tread': 0.4325519917966964,\n",
       " 'symbolism': 0.5154222915632739,\n",
       " 'humanized': 0.4131036205078983,\n",
       " 'top': 0.5210423375949537,\n",
       " 'flight': 0.39696036290701775,\n",
       " 'brutal': 0.434102941488936,\n",
       " 'rapturously': 1.8120534257254559e-13,\n",
       " 'romantic': 0.3848133761368428,\n",
       " 'same': 0.5560581704598859,\n",
       " 'time': 0.43448457240661226,\n",
       " 'despite': 0.5317679404042636,\n",
       " 'shortcomings': 0.48763220105299354,\n",
       " 'happy': 0.3611696559826237,\n",
       " 'feet': 0.3875904630272178,\n",
       " 'brings': 0.40945412271402903,\n",
       " 'lot': 0.448329345929508,\n",
       " 'energy': 0.38635297539494834,\n",
       " 'creativity': 0.5125960031720243,\n",
       " 'increasingly': 0.4698067093010241,\n",
       " 'relies': 0.4653502259110948,\n",
       " 'over': 0.4130200239420188,\n",
       " 'stimulation': 0.55967535666351,\n",
       " 'blythe': 0.4187897079198883,\n",
       " 'danner': 0.4183648367001035,\n",
       " 'lands': 0.415761451332617,\n",
       " 'her': 0.4769319746319353,\n",
       " 'screen': 0.42070993118862177,\n",
       " 'role': 0.4660690762541537,\n",
       " 'retired': 0.42464753239784336,\n",
       " 'widow': 0.39675900479208165,\n",
       " 'decides': 0.4141349297242362,\n",
       " 'dip': 0.4756157820242541,\n",
       " 'tentative': 0.6087483496290031,\n",
       " 'toe': 0.3163887572760127,\n",
       " 'dating': 0.508773778120677,\n",
       " 'pool': 0.4875501758124797,\n",
       " 'indie': 0.4126540193412315,\n",
       " 'ill': 0.42948275090412114,\n",
       " 'dreams': 0.4249682412570902,\n",
       " 'truth': 0.4805882836438174,\n",
       " 'beauty': 0.486314955699582,\n",
       " 'theyre': 0.4531824924970434,\n",
       " 'necessarily': 0.45662712986659415,\n",
       " 'nine': 0.5462231900131296,\n",
       " 'committed': 0.46121758475812497,\n",
       " 'relationship': 0.4385563063436026,\n",
       " 'kids': 0.5056110443953745,\n",
       " 'backseat': 0.4359336710687716,\n",
       " 'blatantly': 0.5740643508368671,\n",
       " 'hearted': 0.24085310773553767,\n",
       " 'half': 0.48056764525915885,\n",
       " 'vibrant': 0.4546020890411402,\n",
       " 'original': 0.43625875098513806,\n",
       " 'wired': 0.44416999894126724,\n",
       " 'article': 0.49390913002704484,\n",
       " 'myopically': 0.42646656950242157,\n",
       " 'neglects': 0.4524417322883783,\n",
       " 'address': 0.3970989360161069,\n",
       " 'arizonas': 0.4800649945224707,\n",
       " 'troubling': 0.35905579526754505,\n",
       " 'anti': 0.5076080806404998,\n",
       " 'immigration': 0.4754017562278714,\n",
       " 'legislation': 0.4392410475939192,\n",
       " 'through': 0.4572665116865582,\n",
       " 'even': 0.4718739570694498,\n",
       " 'splash': 0.4293397888541488,\n",
       " 'hindsight': 0.42184747975232395,\n",
       " 'moulin': 0.4991703272878484,\n",
       " 'rouge': 0.512341794732757,\n",
       " 'le': 0.4130768871957495,\n",
       " 'queda': 0.39560058329092856,\n",
       " 'chica': 0.45644536177448175,\n",
       " 'la': 0.38857254385577744,\n",
       " 'pantalla': 0.40232382990741383,\n",
       " 'slo': 0.3994352701684154,\n",
       " 'satura': 0.527708926725808,\n",
       " 'desborda': 0.43634040866884183,\n",
       " 'es': 0.37528399318454336,\n",
       " 'una': 0.4364657149509394,\n",
       " 'fiesta': 0.46555805361056113,\n",
       " 'de': 0.37676682781106147,\n",
       " 'color': 0.5131880354899981,\n",
       " 'msica': 0.30607189905179266,\n",
       " 'heartwarming': 0.26558058823251446,\n",
       " 'message': 0.416072970599751,\n",
       " 'combined': 0.5111691281788535,\n",
       " 'creative': 0.45466334704164324,\n",
       " 'fantastical': 0.4437513128188083,\n",
       " 'elements': 0.4566062309771973,\n",
       " 'memorable': 0.47947748745682195,\n",
       " 'contraband': 0.4802767060341533,\n",
       " 'thriller': 0.39484394988460597,\n",
       " 'offers': 0.40625767239889654,\n",
       " 'way': 0.4276803523105362,\n",
       " 'assemble': 0.3971555772487072,\n",
       " 'minute': 0.47682605271818296,\n",
       " 'trailer': 0.4664139609340327,\n",
       " 'death': 0.3604769590364271,\n",
       " 'smoochy': 0.39215009339911266,\n",
       " 'odd': 0.4505518116522606,\n",
       " 'im': 0.545819258120993,\n",
       " 'almost': 0.4420509426680319,\n",
       " 'tempted': 0.5336540266599632,\n",
       " 'recommend': 0.5819515175619464,\n",
       " 'stardust': 0.44078274598842754,\n",
       " 'note': 0.4522110086097942,\n",
       " 'pizzazz': 0.4219838264011234,\n",
       " 'zero': 0.44831273726466986,\n",
       " 'substance': 0.4873067391422784,\n",
       " 'thoughtful': 0.2071532560667431,\n",
       " 'motion': 0.5089104362806397,\n",
       " 'picture': 0.4509047461701811,\n",
       " 'ponders': 0.3893463821797883,\n",
       " 'mysteries': 0.4626877157843348,\n",
       " 'accessible': 0.534246632324556,\n",
       " 'emotionally': 0.3068761176423453,\n",
       " 'give': 0.3732841662626949,\n",
       " 'tati': 0.49361158277874545,\n",
       " 'posthumous': 0.4997242399438213,\n",
       " 'chance': 0.47682710922211335,\n",
       " 'express': 0.4344301619909933,\n",
       " 'sad': 0.49825652985656743,\n",
       " 'chomets': 0.33758782226827355,\n",
       " 'only': 0.46890376113837184,\n",
       " 'true': 0.39170932880326864,\n",
       " 'course': 0.43835974839986624,\n",
       " 'honest': 0.40415194353044703,\n",
       " 'beautifully': 0.13213446279075197,\n",
       " 'thoughtfully': 0.43709371467582103,\n",
       " 'realized': 0.16598715073921141,\n",
       " 'promising': 0.49116493900172753,\n",
       " 'quite': 0.46267006509414127,\n",
       " 'ultimately': 0.39957089211792174,\n",
       " 'underwhelming': 0.537453780000751,\n",
       " 'uma': 0.5218922732782798,\n",
       " 'daquelas': 0.5169156954329548,\n",
       " 'obras': 0.44351564365686363,\n",
       " 'que': 0.3887532205694759,\n",
       " 'por': 0.46971472648522733,\n",
       " 'mais': 0.48957523093123223,\n",
       " 'eu': 0.38797792905895645,\n",
       " 'tenha': 0.47102911297969485,\n",
       " 'apreciado': 0.4376189972477539,\n",
       " 'sei': 0.4373220315365757,\n",
       " 'jamais': 0.3946955131400914,\n",
       " 'voltarei': 0.4441517164776967,\n",
       " 'assistir': 0.3440557529691786,\n",
       " 'e': 0.5078681595183437,\n",
       " 'isto': 0.3438169064971957,\n",
       " 'acreditem': 0.43344224803110953,\n",
       " 'um': 0.45838580055359346,\n",
       " 'elo': 0.4327523537447725,\n",
       " 'music': 0.4962497017547368,\n",
       " 'otherwise': 0.4523559258742552,\n",
       " 'shallow': 0.41545160185066904,\n",
       " 'contrived': 0.4705803477668334,\n",
       " 'melodrama': 0.4736751125338272,\n",
       " 'al': 0.37896193317104687,\n",
       " 'daradji': 0.3931043259702651,\n",
       " 'directs': 0.4521542856103335,\n",
       " 'sure': 0.42290689862340336,\n",
       " 'hand': 0.41925530869484956,\n",
       " 'avoiding': 0.3714983024933484,\n",
       " 'icy': 0.4918928540307767,\n",
       " 'realism': 0.4487843710619971,\n",
       " 'chest': 0.5058174349329164,\n",
       " 'beating': 0.42320094786405366,\n",
       " 'result': 0.3831460524260867,\n",
       " 'warm': 0.46725383502618606,\n",
       " 'human': 0.3761782301874817,\n",
       " 'quietly': 0.43969902097231245,\n",
       " 'devastating': 0.10206966600476929,\n",
       " 'coming': 0.5001624274758298,\n",
       " 'mile': 0.45376497600710186,\n",
       " 'away': 0.4883294427568211,\n",
       " 'then': 0.49893808944676815,\n",
       " 'id': 0.386484211061254,\n",
       " 'send': 0.4925319826371603,\n",
       " 'bank': 0.47550074880352355,\n",
       " 'account': 0.44892364251061734,\n",
       " 'number': 0.48391378789679673,\n",
       " 'access': 0.4927344914928097,\n",
       " 'codes': 0.5089595882066928,\n",
       " 'deposit': 0.42612235217173705,\n",
       " 'special': 0.5148623356029174,\n",
       " 'prize': 0.3864225683945881,\n",
       " 'day': 0.4296702681607402,\n",
       " 'turns': 0.4940044763930026,\n",
       " 'enjoying': 0.44127650006358093,\n",
       " 'traditional': 0.3761991522621384,\n",
       " 'admonishment': 0.3494466047393886,\n",
       " 'stop': 0.45520460183972217,\n",
       " 'wasting': 0.45890544126753086,\n",
       " 'living': 0.3910441586890501,\n",
       " 'enjoy': 0.5323653289326357,\n",
       " 'single': 0.4527768030913586,\n",
       " 'moment': 0.4316144883970136,\n",
       " 'anyone': 0.4995571551874608,\n",
       " 'interest': 0.4705134976614247,\n",
       " 'dance': 0.48753461076264965,\n",
       " 'documentary': 0.4847137145917664,\n",
       " 'dumbed': 0.47886389842412325,\n",
       " 'narration': 0.47878476979194234,\n",
       " 'tim': 0.4124374044464369,\n",
       " 'allen': 0.49070258814102713,\n",
       " 'telegraphs': 0.517648625253107,\n",
       " 'second': 0.45431994656093255,\n",
       " 'bit': 0.48187123580305086,\n",
       " 'average': 0.4794908039776297,\n",
       " 'iii': 0.49552056004496253,\n",
       " 'iv': 0.5953058908653938,\n",
       " 'vwhat': 0.4504873902094609,\n",
       " 'possibly': 0.2901891912971607,\n",
       " 'rocky': 0.4954732669434716,\n",
       " 'vs': 0.4566422540712947,\n",
       " 'among': 0.39787933347099597,\n",
       " 'holocaust': 0.3909731179115515,\n",
       " 'agnieszka': 0.3698872971376037,\n",
       " 'hollands': 0.3433731265137026,\n",
       " 'academy': 0.534350542257269,\n",
       " 'award': 0.33942816453908403,\n",
       " 'nominated': 0.446217497371598,\n",
       " 'darkness': 0.46530140693381966,\n",
       " 'during': 0.48124756835205834,\n",
       " 'world': 0.46496916640153224,\n",
       " 'ii': 0.4745281933369074,\n",
       " 'small': 0.3933341409736534,\n",
       " 'town': 0.4304564621792155,\n",
       " 'poland': 0.46140778240554287,\n",
       " 'named': 0.4753040350342801,\n",
       " 'lvov': 0.4336431506457073,\n",
       " 'ukraine': 0.3902566510988362,\n",
       " 'propulsive': 0.4234983889246609,\n",
       " 'punk': 0.36540250690054354,\n",
       " 'rock': 0.41030950298196206,\n",
       " 'soundtrack': 0.31397335827385336,\n",
       " 'rough': 0.3594565384697316,\n",
       " 'cinematography': 0.47506622275146465,\n",
       " 'dragonslayer': 0.4836788826388316,\n",
       " 'care': 0.45287428598230905,\n",
       " 'scrawny': 0.43323105790492,\n",
       " 'young': 0.4680300429380274,\n",
       " 'skating': 0.4578751720399669,\n",
       " 'nowhere': 0.5207338536844739,\n",
       " 'viewers': 0.4621499940756548,\n",
       " 'wont': 0.4809644851197443,\n",
       " 'feel': 0.4229833153367624,\n",
       " 'theyve': 0.4665148066387087,\n",
       " 'dealt': 0.4491234870015764,\n",
       " 'cruelly': 0.49741576229736084,\n",
       " 'pirates': 0.4460162987978319,\n",
       " 'defining': 0.3991551932282807,\n",
       " 'wondermentbut': 0.5036446927621334,\n",
       " 'tap': 0.3312916880359678,\n",
       " 'essential': 0.40098615105570234,\n",
       " 'jolt': 0.46569494024412994,\n",
       " 'satisfaction': 0.37561254273873435,\n",
       " 'missing': 0.4657510216456131,\n",
       " 'pike': 0.42640655070343175,\n",
       " 'gives': 0.4954200683938181,\n",
       " 'thoroughly': 0.4752914417675272,\n",
       " 'convincing': 0.44856445041367426,\n",
       " 'colvin': 0.2565265714764271,\n",
       " 'drama': 0.40144589179228496,\n",
       " 'though': 0.4326583281556801,\n",
       " 'dimensional': 0.4834673104713823,\n",
       " 'filmmakers': 0.5037819247729339,\n",
       " 'focused': 0.5026837875360872,\n",
       " 'dont': 0.48625574329095134,\n",
       " 'notice': 0.4079573648455426,\n",
       " 'chadwick': 0.2535819474206085,\n",
       " 'boseman': 0.18138655938284726,\n",
       " 'precise': 0.49339755776533,\n",
       " 'impersonation': 0.5300516439365583,\n",
       " 'browns': 0.387426170567988,\n",
       " 'gravelly': 0.4232919252398181,\n",
       " 'voice': 0.3567655273101832,\n",
       " 'moves': 0.4079672417164712,\n",
       " 'lends': 0.4060962806867704,\n",
       " 'emotional': 0.40161282569975065,\n",
       " 'weight': 0.5188053110392231,\n",
       " 'scenes': 0.5074746086810747,\n",
       " 'deserve': 0.45158760414091137,\n",
       " 'however': 0.5651238338165488,\n",
       " 'feels': 0.5112405310428788,\n",
       " 'series': 0.5718227207329591,\n",
       " 'bullet': 0.45098354074544744,\n",
       " 'points': 0.48360393317932265,\n",
       " 'organic': 0.4050980625088709,\n",
       " 'fluffed': 0.4717358679784841,\n",
       " 'satirical': 0.36837646281908276,\n",
       " 'targets': 0.4704616966874951,\n",
       " 'shortfall': 0.46192284923153226,\n",
       " 'required': 0.4411416485397356,\n",
       " 'rom': 0.5635592456862851,\n",
       " 'com': 0.5105556816421436,\n",
       " 'cheap': 0.5302017716483464,\n",
       " 'gags': 0.4858495555304269,\n",
       " 'unpleasant': 0.4768942896526035,\n",
       " 'hero': 0.4065024226735997,\n",
       " 'situate': 0.4028786637273181,\n",
       " 'friends': 0.37294407736541013,\n",
       " 'several': 0.39607025094199855,\n",
       " 'floors': 0.5197355671951682,\n",
       " 'below': 0.43183649358715326,\n",
       " 'desired': 0.40833014183118466,\n",
       " 'apartment': 0.39484503324743875,\n",
       " 'esque': 0.4726465631257982,\n",
       " 'heights': 0.38329620255885855,\n",
       " 'wonder': 0.4955004773609445,\n",
       " 'wheel': 0.5392901131894989,\n",
       " 'kate': 0.46670729477685385,\n",
       " 'winslet': 0.4667649588276271,\n",
       " 'pluses': 0.5311913860397777,\n",
       " 'plenty': 0.42991156162893784,\n",
       " 'laughs': 0.5155031631449849,\n",
       " 'outrageous': 0.36408194996348725,\n",
       " 'situations': 0.4668412026485081,\n",
       " 'guys': 0.44276543378262945,\n",
       " 'themselves': 0.5013407626402174,\n",
       " 'lawrence': 0.500651541599618,\n",
       " 'luxuriates': 0.4007362394770144,\n",
       " 'tremendous': 0.24625464050521095,\n",
       " 'hits': 0.49551244283107226,\n",
       " 'spot': 0.44241255676807784,\n",
       " 'entertaining': 0.20027328832971178,\n",
       " 'informative': 0.16572446760360024,\n",
       " 'deeply': 0.5268538718890454,\n",
       " 'moving': 0.4389596063879677,\n",
       " 'richly': 0.13532001267810087,\n",
       " 'appointed': 0.39671320054571435,\n",
       " 'meant': 0.6076928800610488,\n",
       " 'titular': 0.49106283721052785,\n",
       " 'objet': 0.4216173716982765,\n",
       " 'dart': 0.4487028080596468,\n",
       " 'adult': 0.48941853877837416,\n",
       " 'beginners': 0.5182368148040136,\n",
       " 'baby': 0.5586067882318988,\n",
       " 'step': 0.4741458970953559,\n",
       " 'figuring': 0.44419366800306403,\n",
       " 'kroll': 0.4615517090211954,\n",
       " 'transition': 0.4898388646964502,\n",
       " 'potential': 0.49494030266925615,\n",
       " 'certainly': 0.39858670759313525,\n",
       " 'unremarkable': 0.48649725984531306,\n",
       " 'rainmaker': 0.4309986548197119,\n",
       " 'other': 0.48982846317848333,\n",
       " 'astonishing': 0.4554975975856684,\n",
       " 'complex': 0.2902278705594358,\n",
       " 'dismissed': 0.4661898716320334,\n",
       " 'bright': 0.4766705264921599,\n",
       " 'terminally': 0.46426121640245716,\n",
       " 'conscious': 0.47200879782580335,\n",
       " 'hoary': 0.4453675427441273,\n",
       " 'conceits': 0.4158915573458285,\n",
       " 'ancient': 0.5764492454232988,\n",
       " 'vintage': 0.41780202979947545,\n",
       " 'dubious': 0.4704667295955454,\n",
       " 'worth': 0.5045510110779339,\n",
       " 'riotously': 0.4298281532305967,\n",
       " 'extremely': 0.510221628228206,\n",
       " 'abundantly': 0.42751179947409124,\n",
       " 'given': 0.47771113525335923,\n",
       " 'monumental': 0.3496287014790956,\n",
       " 'task': 0.5215576074140762,\n",
       " 'these': 0.44368395347601614,\n",
       " 'managed': 0.4171401262034768,\n",
       " 'tell': 0.5096982680397473,\n",
       " 'grow': 0.5668577607873274,\n",
       " 'band': 0.30905783828246713,\n",
       " 'fascinating': 0.3115943041479336,\n",
       " 'tactfully': 5.888154608306692e-13,\n",
       " 'charts': 0.29500605854424555,\n",
       " 'sort': 0.40581660878415277,\n",
       " 'philosophical': 0.4576168888360457,\n",
       " 'differences': 0.4454261435597547,\n",
       " 'eventually': 0.5652418640575181,\n",
       " 'undermine': 0.4751089625272242,\n",
       " 'any': 0.483680564550779,\n",
       " 'group': 0.48800031805465616,\n",
       " 'endeavor': 0.4769234901419011,\n",
       " ...}"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "probability['rotten'] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "56091998",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top 10 words for class fresh\n",
      "['reassures', 'hirokazu', 'zvyagintsev', 'evenhanded', 'belleville', 'farocki', 'akhavans', 'koreedas', 'akermans', 'ades']\n",
      "top 10 words for class rotten\n",
      "['charmless', 'dallying', 'laughless', 'spritzing', 'kada', 'amateurishly', 'meretricious', 'convolutedin', 'primjera', 'rijetkih']\n"
     ]
    }
   ],
   "source": [
    "fresh_prob=probability['fresh']\n",
    "rotten_prob=probability['rotten']\n",
    "\n",
    "# sort the dictionary items by their values in descending order\n",
    "sorted_items_fresh = sorted(fresh_prob.items(), key=lambda x: x[1], reverse=True)\n",
    "\n",
    "# get the top 10 items\n",
    "top_10_fresh = sorted_items_fresh[:10]\n",
    "\n",
    "# sort the dictionary items by their values in descending order\n",
    "sorted_items_rotten = sorted(rotten_prob.items(), key=lambda x: x[1], reverse=True)\n",
    "\n",
    "# get the top 10 items\n",
    "top_10_rotten = sorted_items_rotten[:10]\n",
    "top_fresh=[]\n",
    "for key in top_10_fresh:\n",
    "        top_fresh.append(key[0])\n",
    "\n",
    "top_rotten=[]\n",
    "for key in top_10_rotten:\n",
    "        top_rotten.append(key[0])\n",
    "top_fresh\n",
    "# print the top 10 items for fresh class after removing pronouns\n",
    "print(\"top 10 words for class fresh\")\n",
    "print(top_fresh)\n",
    "# print the top 10 items for rotten class after removing pronouns\n",
    "print(\"top 10 words for class rotten\")\n",
    "print(top_rotten)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c2b12982",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}