tf_idf_scratch.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nltk\n",
    "import math, numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package abc to /home/hackerearth/nltk_data...\n",
      "[nltk_data]   Unzipping corpora/abc.zip.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nltk.download('abc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.corpus import abc,stopwords\n",
    "from nltk.stem import PorterStemmer \n",
    "from string import punctuation as p\n",
    "ps=PorterStemmer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['rural.txt', 'science.txt']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "abc.fileids()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=abc.words('rural.txt')\n",
    "science=abc.words('science.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['PM', 'denies', 'knowledge', 'of', 'AWB', 'kickbacks', ...]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rural"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Cystic', 'fibrosis', 'affects', '30', ',', '000', ...]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "science"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "345580\n",
      "421231\n"
     ]
    }
   ],
   "source": [
    "print(len(rural))\n",
    "print(len(science))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     /home/hackerearth/nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nltk.download('stopwords')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Removing Stop words "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "stop_words=set(stopwords.words(\"english\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=[word for word in rural if word not in stop_words]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "science=[word for word in science if word not in stop_words]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "225374\n",
      "275580\n"
     ]
    }
   ],
   "source": [
    "print(len(rural))\n",
    "print(len(science))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Removing Punctuation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def punctuation(data):\n",
    "    punct_num = '''!()-[]{};:\\n,<>./?@#$%^&\\*_~0123456789=|'\"'''\n",
    "    for i in range(len(data)):\n",
    "        no_punct = \"\"\n",
    "        for char in data[i]:\n",
    "            if char not in punct_num :\n",
    "                no_punct = no_punct + char  \n",
    "        if len(no_punct)==0:\n",
    "            no_punct=' '\n",
    "        data[i]=no_punct\n",
    "    return(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=punctuation(rural)\n",
    "science=punctuation(science)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Removing Whitespaces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=list(filter(lambda a:a!=' ',rural))\n",
    "science=list(filter(lambda a:a!=' ',science))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Lower Case "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=[ w.lower() for w in rural]\n",
    "science=[w.lower() for w in science]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Stemming Word "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural=[ps.stem(w) for w in rural]\n",
    "science=[ps.stem(w) for w in science]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['pm', 'deni', 'knowledg', ..., 'minut', 'anoth', 'said'],\n",
       "      dtype='<U16')"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(rural).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using Library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer as tfidf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "vector=tfidf()\n",
    "response=vector.fit_transform(rural,science)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  (0, 6513)\t1.0\n",
      "  (1, 2329)\t1.0\n",
      "  (2, 4740)\t1.0\n",
      "  (3, 607)\t1.0\n",
      "  (4, 4696)\t1.0\n",
      "  (5, 8580)\t1.0\n",
      "  (6, 6671)\t1.0\n",
      "  (7, 5513)\t1.0\n",
      "  (8, 2329)\t1.0\n",
      "  (9, 4732)\t1.0\n",
      "  (10, 607)\t1.0\n",
      "  (11, 6285)\t1.0\n",
      "  (12, 4696)\t1.0\n",
      "  (13, 4459)\t1.0\n",
      "  (14, 2374)\t1.0\n",
      "  (15, 9644)\t1.0\n",
      "  (16, 9451)\t1.0\n",
      "  (17, 3054)\t1.0\n",
      "  (18, 505)\t1.0\n",
      "  (19, 4680)\t1.0\n",
      "  (20, 3475)\t1.0\n",
      "  (21, 4343)\t1.0\n",
      "  (22, 4459)\t1.0\n",
      "  (23, 9451)\t1.0\n",
      "  (24, 7433)\t1.0\n",
      "  :\t:\n",
      "  (184058, 7973)\t1.0\n",
      "  (184059, 2105)\t1.0\n",
      "  (184060, 9688)\t1.0\n",
      "  (184061, 5298)\t1.0\n",
      "  (184063, 2731)\t1.0\n",
      "  (184064, 1610)\t1.0\n",
      "  (184065, 5857)\t1.0\n",
      "  (184066, 2731)\t1.0\n",
      "  (184067, 1610)\t1.0\n",
      "  (184068, 6036)\t1.0\n",
      "  (184069, 3793)\t1.0\n",
      "  (184070, 7422)\t1.0\n",
      "  (184071, 3838)\t1.0\n",
      "  (184072, 916)\t1.0\n",
      "  (184073, 1610)\t1.0\n",
      "  (184074, 7730)\t1.0\n",
      "  (184075, 8673)\t1.0\n",
      "  (184076, 3677)\t1.0\n",
      "  (184077, 3596)\t1.0\n",
      "  (184078, 2574)\t1.0\n",
      "  (184079, 4396)\t1.0\n",
      "  (184080, 4838)\t1.0\n",
      "  (184081, 5523)\t1.0\n",
      "  (184082, 366)\t1.0\n",
      "  (184083, 7422)\t1.0\n"
     ]
    }
   ],
   "source": [
    "print(response)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['aac', 'aami', 'aaron', ..., 'zone', 'zoo', 'zulu'], dtype='<U16')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(vector.get_feature_names()).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Building Vocabulary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.probability import FreqDist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def buildVocabulary(data):\n",
    "    all_words = []\n",
    "    for word in data:\n",
    "        all_words.append(word)\n",
    "    wordlist = nltk.FreqDist(all_words)\n",
    "    return wordlist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "rural_vocab=buildVocabulary(rural)\n",
    "science_vocab=buildVocabulary(science)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "FreqDist({'say': 3419, 'the': 2519, 'said': 2088, 'australia': 1586, 'year': 1342, 'farmer': 1270, 'new': 1160, 'industri': 979, 'price': 975, 'water': 960, ...})"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rural_vocab"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tf - Idf representation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def tf(vocabulary,data):\n",
    "    tf={}\n",
    "    n=len(data)\n",
    "    for (word,count) in vocabulary.items():\n",
    "        tf[word]=count/n\n",
    "    return tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "tf_rural=tf(rural_vocab,rural)\n",
    "tf_science=tf(science_vocab,science)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'pm': 0.00010864605288889855,\n",
       " 'deni': 0.000222724408422242,\n",
       " 'knowledg': 9.77814476000087e-05,\n",
       " 'awb': 0.0029660372438669305,\n",
       " 'kickback': 0.00019013059255557245,\n",
       " 'the': 0.013683970361356771,\n",
       " 'prime': 0.0005052041459333782,\n",
       " 'minist': 0.001722039938289042,\n",
       " 'knew': 5.432302644444927e-05,\n",
       " 'pay': 0.0009615175680667521,\n",
       " 'iraq': 0.0005540948697333826,\n",
       " 'despit': 0.0007496577649334,\n",
       " 'write': 5.97553290888942e-05,\n",
       " 'wheat': 0.003433215271289194,\n",
       " 'export': 0.0037156950088003305,\n",
       " 'ask': 0.0006518763173333912,\n",
       " 'kept': 0.0001195106581777884,\n",
       " 'fulli': 5.432302644444927e-05,\n",
       " 'inform': 0.00042915190891114926,\n",
       " 'sale': 0.001423263292844571,\n",
       " 'letter': 0.00010864605288889855,\n",
       " 'john': 0.0008854653310445232,\n",
       " 'howard': 0.0003205058560222507,\n",
       " 'deputi': 0.00020642750048890723,\n",
       " 'mark': 0.000521501053866713,\n",
       " 'vail': 0.00036939657982225507,\n",
       " 'releas': 0.0005649594750222725,\n",
       " 'cole': 0.000820277699311184,\n",
       " 'inquiri': 0.0012657265161556681,\n",
       " 'oil': 0.0012657265161556681,\n",
       " 'food': 0.0015808000695334739,\n",
       " 'program': 0.0008691684231111884,\n",
       " 'in': 0.0010647313183112057,\n",
       " 'one': 0.0031072771126224984,\n",
       " 'mr': 0.00222724408422242,\n",
       " 'manag': 0.0020534103996001827,\n",
       " 'director': 0.0006790378305556159,\n",
       " 'andrew': 0.0003368027639555855,\n",
       " 'lindberg': 9.234914495556377e-05,\n",
       " 'remain': 0.000450881119488929,\n",
       " 'close': 0.0006029855935333869,\n",
       " 'contact': 9.234914495556377e-05,\n",
       " 'govern': 0.005117229091067122,\n",
       " 'opposit': 0.0003150735533778058,\n",
       " 'gavan': 4.889072380000435e-05,\n",
       " 'o': 0.00028791204015558115,\n",
       " 'connor': 0.00011407835553334348,\n",
       " 'say': 0.018573042741357206,\n",
       " 'sent': 0.00020642750048890723,\n",
       " 'time': 0.0025749114534668956,\n",
       " 'though': 0.0002444536190000217,\n",
       " 'jordanian': 2.172921057777971e-05,\n",
       " 'truck': 0.00040742269833336954,\n",
       " 'compani': 0.0032702461919558463,\n",
       " 'he': 0.0014612894113556854,\n",
       " 'longer': 0.00027161513222224634,\n",
       " 'wipe': 0.0001575367766889029,\n",
       " 'hand': 0.00040742269833336954,\n",
       " 'illicit': 1.0864605288889855e-05,\n",
       " 'payment': 0.00030964125073336087,\n",
       " 'total': 0.00034766736924447534,\n",
       " 'million': 0.003139870928489168,\n",
       " 'respons': 0.000521501053866713,\n",
       " 'must': 0.0003259381586666956,\n",
       " 'lay': 6.518763173333913e-05,\n",
       " 'may': 0.0013309141478890071,\n",
       " 'squar': 9.77814476000087e-05,\n",
       " 'feet': 4.345842115555942e-05,\n",
       " 'coalit': 0.00013037526346667825,\n",
       " 'trade': 0.002243540992155755,\n",
       " 'agricultur': 0.0024119423741335477,\n",
       " 'said': 0.011342647921601008,\n",
       " 'but': 0.002629234479911345,\n",
       " 'show': 0.0011679450685556593,\n",
       " 'inquir': 1.0864605288889855e-05,\n",
       " 'futur': 0.0013091849373112275,\n",
       " 'prove': 0.00023358901371113187,\n",
       " 'it': 0.002835661980400252,\n",
       " 'would': 0.002466265400577997,\n",
       " 'astonish': 5.432302644444927e-06,\n",
       " 'i': 0.004791290932400426,\n",
       " 'done': 0.0004943395406444884,\n",
       " 'anyth': 0.00030420894808891595,\n",
       " 'possibl': 0.000450881119488929,\n",
       " 'could': 0.002455400795289107,\n",
       " 'preserv': 4.889072380000435e-05,\n",
       " 'australia': 0.008615631994089655,\n",
       " 'valuabl': 6.518763173333913e-05,\n",
       " 'market': 0.004172008430933704,\n",
       " 'email': 5.432302644444927e-05,\n",
       " 'question': 0.000445448816844484,\n",
       " 'today': 0.0012765911214445578,\n",
       " 'peter': 0.0012928880293778928,\n",
       " 'geari': 2.172921057777971e-05,\n",
       " 'receiv': 0.0007550900675778449,\n",
       " 'indic': 0.00028791204015558115,\n",
       " 'iraqi': 0.0001955628952000174,\n",
       " 'grain': 0.0029225788227113707,\n",
       " 'board': 0.0006192825014667217,\n",
       " 'approach': 0.0002009951978444623,\n",
       " 'provid': 0.0007550900675778449,\n",
       " 'servic': 0.0010918928315334303,\n",
       " 'forward': 0.0003368027639555855,\n",
       " 'two': 0.00186327980704461,\n",
       " 'colleagu': 7.605223702222899e-05,\n",
       " 'rememb': 5.97553290888942e-05,\n",
       " 'read': 7.605223702222899e-05,\n",
       " 'although': 0.00041285500097781446,\n",
       " 'skim': 5.432302644444927e-06,\n",
       " 'support': 0.0012331327002889985,\n",
       " 'still': 0.0013254818452445622,\n",
       " 'plenti': 0.0001195106581777884,\n",
       " 'among': 0.0002172921057777971,\n",
       " 'grower': 0.005138958301644902,\n",
       " 'central': 0.0006464440146889464,\n",
       " 'western': 0.002009951978444623,\n",
       " 'new': 0.0063014710675561155,\n",
       " 'south': 0.004927098498511549,\n",
       " 'wale': 0.0020371134916668475,\n",
       " 'revel': 1.6296907933334782e-05,\n",
       " 'produc': 0.002754177440733578,\n",
       " 'broadli': 2.172921057777971e-05,\n",
       " 'attempt': 0.0002553182242889116,\n",
       " 'get': 0.0028736880989113666,\n",
       " 'best': 0.0005703917776667173,\n",
       " 'price': 0.005296495078333804,\n",
       " 'product': 0.0024391038873557724,\n",
       " 'think': 0.0022924317159557592,\n",
       " 'ploy': 5.432302644444927e-06,\n",
       " 'oversea': 0.00041828730362225943,\n",
       " 'interest': 0.0007768192781556246,\n",
       " 'tri': 0.0011570804632667696,\n",
       " 'singl': 0.0009506529627778623,\n",
       " 'desk': 0.0007931161860889594,\n",
       " 'put': 0.001271158818800113,\n",
       " 'asid': 3.2593815866669564e-05,\n",
       " 'stori': 0.00014667217140001303,\n",
       " 'go': 0.0031018448099780537,\n",
       " 'round': 0.0002390213163555768,\n",
       " 'commiss': 0.0005377979618000478,\n",
       " 'everyth': 0.0001575367766889029,\n",
       " 'way': 0.0015427739510223593,\n",
       " 'peopl': 0.0026238021772669,\n",
       " 'got': 0.0013852371743334564,\n",
       " 'thing': 0.0009506529627778623,\n",
       " 'busi': 0.001129918950044545,\n",
       " 'middl': 0.000222724408422242,\n",
       " 'east': 0.0010158405945112014,\n",
       " 'asian': 8.14845396666739e-05,\n",
       " 'countri': 0.0019230351361335042,\n",
       " 'actual': 0.0007605223702222898,\n",
       " 'pretti': 0.0003259381586666956,\n",
       " 'reason': 0.0003856934877555898,\n",
       " 'system': 0.0015373416483779145,\n",
       " 'give': 0.0007931161860889594,\n",
       " 'fair': 0.0001955628952000174,\n",
       " 'moment': 0.0005377979618000478,\n",
       " 'averag': 0.0006247148041111666,\n",
       " 'perform': 0.0002172921057777971,\n",
       " 'fairli': 0.00018469828991112753,\n",
       " 'well': 0.0013309141478890071,\n",
       " 'anoth': 0.0009832467786445318,\n",
       " 'biggest': 0.0008637361204667434,\n",
       " 'someon': 0.00010864605288889855,\n",
       " 'els': 0.00010864605288889855,\n",
       " 'take': 0.0015536385563112492,\n",
       " 'whether': 0.000744225462288955,\n",
       " 'multin': 1.0864605288889855e-05,\n",
       " 'much': 0.0010755959236000956,\n",
       " 'foothold': 2.7161513222224636e-05,\n",
       " 'advantag': 0.0002172921057777971,\n",
       " 'analyst': 0.0005758240803111623,\n",
       " 'predict': 0.000516068751222268,\n",
       " 'drop': 0.0008854653310445232,\n",
       " 'tonn': 0.001504747832511245,\n",
       " 'back': 0.0016133938854001435,\n",
       " 'malcolm': 0.00014667217140001303,\n",
       " 'bartholomaeu': 4.345842115555942e-05,\n",
       " 'pool': 0.0005323656591556029,\n",
       " 'return': 0.0005486625670889377,\n",
       " 'alreadi': 0.0007605223702222898,\n",
       " 'year': 0.007290150148845092,\n",
       " 'past': 0.0007496577649334,\n",
       " 'five': 0.0008583038178222985,\n",
       " 'premium': 0.000222724408422242,\n",
       " 'achiev': 0.0001629690793333478,\n",
       " 'monopoli': 0.0001955628952000174,\n",
       " 'sever': 0.00047261033006670866,\n",
       " 'erod': 5.432302644444927e-06,\n",
       " 'sa': 0.00027161513222224634,\n",
       " 'farmer': 0.006899024358445058,\n",
       " 'help': 0.0021620564524890812,\n",
       " 'fire': 0.0007279285543556203,\n",
       " 'ravag': 1.0864605288889855e-05,\n",
       " 'neighbour': 9.234914495556377e-05,\n",
       " 'donat': 0.00015210447404445798,\n",
       " 'truckload': 1.0864605288889855e-05,\n",
       " 'hay': 0.00036939657982225507,\n",
       " 'across': 0.0012602942135112232,\n",
       " 'border': 0.0001629690793333478,\n",
       " 'wake': 0.000222724408422242,\n",
       " 'grampian': 5.432302644444927e-05,\n",
       " 'bushfir': 0.00029334434280002607,\n",
       " 'day': 0.0014721540166445753,\n",
       " 'agist': 6.518763173333913e-05,\n",
       " 'cattl': 0.002145759544555746,\n",
       " 'they': 0.0009397883574889724,\n",
       " 'begin': 0.00030420894808891595,\n",
       " 'fodder': 0.00011407835553334348,\n",
       " 'drive': 0.00027161513222224634,\n",
       " 'coordin': 0.00012494296082223334,\n",
       " 'conner': 5.432302644444927e-06,\n",
       " 'overwhelm': 3.8026118511114494e-05,\n",
       " 'all': 0.0001195106581777884,\n",
       " 'week': 0.002156624149844636,\n",
       " 'gone': 0.00029334434280002607,\n",
       " 'place': 0.0008257100019556289,\n",
       " 'load': 0.00018469828991112753,\n",
       " 'we': 0.0029443080332891506,\n",
       " 'man': 0.0001195106581777884,\n",
       " 'full': 0.0003911257904000348,\n",
       " 'rest': 0.00014667217140001303,\n",
       " 'straight': 6.518763173333913e-05,\n",
       " 'move': 0.0009234914495556376,\n",
       " 'floodwat': 5.432302644444927e-05,\n",
       " 'highway': 0.0001195106581777884,\n",
       " 'a': 0.002558614545533561,\n",
       " 'major': 0.0014341278981334607,\n",
       " 'northern': 0.001341778753177897,\n",
       " 'territori': 0.0009506529627778623,\n",
       " 'block': 0.0001792659872666826,\n",
       " 'victoria': 0.001347211055822342,\n",
       " 'river': 0.0012222680950001086,\n",
       " 'cut': 0.0011244866474001,\n",
       " 'also': 0.0021403272419113013,\n",
       " 'flood': 0.0003802611851111449,\n",
       " 'remot': 0.0004617457247778188,\n",
       " 'pigeon': 1.6296907933334782e-05,\n",
       " 'hole': 3.2593815866669564e-05,\n",
       " 'aborigin': 0.0002770474348666913,\n",
       " 'commun': 0.0013961017796223464,\n",
       " 'stockman': 3.2593815866669564e-05,\n",
       " 'simon': 5.97553290888942e-05,\n",
       " 'ham': 1.0864605288889855e-05,\n",
       " 'describ': 8.14845396666739e-05,\n",
       " 'reloc': 9.77814476000087e-05,\n",
       " 'hundr': 0.0004345842115555942,\n",
       " 'higher': 0.0007224962517111753,\n",
       " 'ground': 0.00041828730362225943,\n",
       " 'vehicl': 8.691684231111884e-05,\n",
       " 'camp': 5.432302644444927e-05,\n",
       " 'set': 0.0012657265161556681,\n",
       " 'belong': 3.2593815866669564e-05,\n",
       " 'coupl': 0.00027161513222224634,\n",
       " 'boat': 0.00028247973751113623,\n",
       " 'onto': 0.000222724408422242,\n",
       " 'metr': 0.00012494296082223334,\n",
       " 'sandston': 5.432302644444927e-06,\n",
       " 'ridg': 1.6296907933334782e-05,\n",
       " 'height': 2.7161513222224636e-05,\n",
       " 'might': 0.0005106364485778232,\n",
       " 'littl': 0.0006029855935333869,\n",
       " 'start': 0.001271158818800113,\n",
       " 'sold': 0.0004997718432889333,\n",
       " 'm': 0.00013037526346667825,\n",
       " 'tasmania': 0.0008474392125334087,\n",
       " 'main': 0.00033137046131114054,\n",
       " 'chang': 0.0022218117815779753,\n",
       " 'second': 0.0002607505269333565,\n",
       " 'three': 0.0012331327002889985,\n",
       " 'former': 0.0005486625670889377,\n",
       " 'state': 0.0033897568501336348,\n",
       " 'own': 0.0002607505269333565,\n",
       " 'tasmanian': 0.0005323656591556029,\n",
       " 'elev': 2.7161513222224636e-05,\n",
       " 'local': 0.0011081897394667653,\n",
       " 'agribusi': 0.00012494296082223334,\n",
       " 'robert': 0.00018469828991112753,\n",
       " 'limit': 0.0005377979618000478,\n",
       " 'deal': 0.001119054344755655,\n",
       " 'includ': 0.000901762238977858,\n",
       " 'silo': 0.00010321375024445361,\n",
       " 'launceston': 4.345842115555942e-05,\n",
       " 'devonport': 5.432302644444927e-06,\n",
       " 'powranna': 1.0864605288889855e-05,\n",
       " 'macleod': 5.432302644444927e-06,\n",
       " 'bid': 0.00034766736924447534,\n",
       " 'first': 0.0016731492144890377,\n",
       " 'privatis': 9.234914495556377e-05,\n",
       " 'disappoint': 0.0002770474348666913,\n",
       " 'success': 0.0003748288824667,\n",
       " 'point': 0.0004345842115555942,\n",
       " 'wine': 0.0013906694769779014,\n",
       " 'worker': 0.0009289237522000826,\n",
       " 'stop': 0.0007279285543556203,\n",
       " 'work': 0.0021348949392668566,\n",
       " 'hardi': 2.172921057777971e-05,\n",
       " 'stanley': 3.2593815866669564e-05,\n",
       " 'wineri': 0.00014667217140001303,\n",
       " 'walk': 0.00011407835553334348,\n",
       " 'job': 0.0006355794094000565,\n",
       " 'forti': 3.2593815866669564e-05,\n",
       " 'staff': 0.00033137046131114054,\n",
       " 'morn': 0.00029334434280002607,\n",
       " 'disput': 0.00021185980313335217,\n",
       " 'enterpris': 8.691684231111884e-05,\n",
       " 'bargain': 7.061993437778406e-05,\n",
       " 'agreement': 0.0005595271723778275,\n",
       " 'stopwork': 5.432302644444927e-06,\n",
       " 'come': 0.0019393320440668391,\n",
       " 'region': 0.002009951978444623,\n",
       " 'grape': 0.0008420069098889637,\n",
       " 'crush': 9.234914495556377e-05,\n",
       " 'took': 0.0001575367766889029,\n",
       " 'matter': 0.00018469828991112753,\n",
       " 'industri': 0.005318224288911584,\n",
       " 'relat': 0.00033137046131114054,\n",
       " 'irc': 5.432302644444927e-06,\n",
       " 'friday': 0.00014123986875556812,\n",
       " 'wool': 0.0022218117815779753,\n",
       " 'bodi': 0.0005921209882444971,\n",
       " 'eye': 0.00012494296082223334,\n",
       " 'sportswear': 1.6296907933334782e-05,\n",
       " 'billion': 0.0005921209882444971,\n",
       " 'global': 0.0005703917776667173,\n",
       " 'target': 0.00034223506660003043,\n",
       " 'promot': 0.0002009951978444623,\n",
       " 'australian': 0.0052041459333782404,\n",
       " 'innov': 0.0003259381586666956,\n",
       " 'awi': 0.00034766736924447534,\n",
       " 'blend': 0.00010321375024445361,\n",
       " 'shirt': 5.432302644444927e-06,\n",
       " 'casual': 2.7161513222224636e-05,\n",
       " 'wear': 4.345842115555942e-05,\n",
       " 'manufactur': 0.00021185980313335217,\n",
       " 'largest': 0.0004943395406444884,\n",
       " 'held': 0.00030964125073336087,\n",
       " 'germani': 6.518763173333913e-05,\n",
       " 'len': 4.889072380000435e-05,\n",
       " 'stephen': 0.00010864605288889855,\n",
       " 'pricier': 5.432302644444927e-06,\n",
       " 'end': 0.0009886790812889768,\n",
       " 'shopper': 7.605223702222899e-05,\n",
       " 'will': 0.00013037526346667825,\n",
       " 'sport': 0.00011407835553334348,\n",
       " 'apparel': 1.0864605288889855e-05,\n",
       " 'sector': 0.001053866713022316,\n",
       " 'certainli': 0.0006464440146889464,\n",
       " 'fastest': 1.0864605288889855e-05,\n",
       " 'grow': 0.0011244866474001,\n",
       " 'world': 0.0013580756611112318,\n",
       " 'secret': 0.00013580756611112317,\n",
       " 'big': 0.0011842419764889942,\n",
       " 'share': 0.0006953347384889507,\n",
       " 'level': 0.0010701636209556507,\n",
       " 'particularli': 0.0004943395406444884,\n",
       " 'merino': 0.00019013059255557245,\n",
       " 'realli': 0.0011570804632667696,\n",
       " 'almost': 0.0005866886856000521,\n",
       " 'radar': 5.432302644444927e-06,\n",
       " 'organis': 0.0007061993437778405,\n",
       " 'step': 0.0003150735533778058,\n",
       " 'ban': 0.0006410117120445014,\n",
       " 'rodeo': 0.000222724408422242,\n",
       " 'campaign': 0.0005758240803111623,\n",
       " 'sustain': 0.0002661828295778014,\n",
       " 'attack': 0.00017383368462223767,\n",
       " 'anim': 0.0012548619108667782,\n",
       " 'right': 0.0009615175680667521,\n",
       " 'activist': 4.345842115555942e-05,\n",
       " 'destroy': 0.0002498859216444667,\n",
       " 'weekend': 0.00042915190891114926,\n",
       " 'event': 0.0003259381586666956,\n",
       " 'buck': 4.889072380000435e-05,\n",
       " 'hors': 0.00027161513222224634,\n",
       " 'break': 0.00036939657982225507,\n",
       " 'leg': 0.00014667217140001303,\n",
       " 'ulverston': 1.6296907933334782e-05,\n",
       " 'saturday': 5.97553290888942e-05,\n",
       " 'ago': 0.0005758240803111623,\n",
       " 'bull': 0.00019013059255557245,\n",
       " 'appar': 3.2593815866669564e-05,\n",
       " 'ride': 7.061993437778406e-05,\n",
       " 'competit': 0.0007768192781556246,\n",
       " 'carrick': 5.432302644444927e-06,\n",
       " 'owner': 0.00034766736924447534,\n",
       " 'brian': 0.0001575367766889029,\n",
       " 'fish': 0.0009832467786445318,\n",
       " 'unfortun': 0.00014667217140001303,\n",
       " 'accid': 3.2593815866669564e-05,\n",
       " 'welfar': 0.0002661828295778014,\n",
       " 'issu': 0.0010973251341778753,\n",
       " 'emma': 2.172921057777971e-05,\n",
       " 'haswel': 1.0864605288889855e-05,\n",
       " 'against': 1.6296907933334782e-05,\n",
       " 'cruelti': 7.061993437778406e-05,\n",
       " 'effort': 0.00034223506660003043,\n",
       " 'snail': 7.605223702222899e-05,\n",
       " 'use': 0.0024391038873557724,\n",
       " 'cancer': 0.00018469828991112753,\n",
       " 'research': 0.0016568523065557028,\n",
       " 'investig': 0.0005921209882444971,\n",
       " 'sea': 0.000298776645444471,\n",
       " 'eventu': 8.14845396666739e-05,\n",
       " 'treat': 0.00013580756611112317,\n",
       " 'until': 5.432302644444927e-05,\n",
       " 'popul': 0.00020642750048890723,\n",
       " 'virtual': 8.14845396666739e-05,\n",
       " 'unexplor': 5.432302644444927e-06,\n",
       " 'flinder': 7.061993437778406e-05,\n",
       " 'univers': 0.0005052041459333782,\n",
       " 'hope': 0.0011625127659112145,\n",
       " 'discov': 0.000222724408422242,\n",
       " 'benefici': 2.7161513222224636e-05,\n",
       " 'compound': 4.345842115555942e-05,\n",
       " 'predatori': 1.0864605288889855e-05,\n",
       " 'known': 0.00028247973751113623,\n",
       " 'whelk': 1.0864605288889855e-05,\n",
       " 'marin': 0.00021185980313335217,\n",
       " 'biologist': 1.6296907933334782e-05,\n",
       " 'dr': 0.0007279285543556203,\n",
       " 'kirsten': 5.432302644444927e-06,\n",
       " 'benkendorff': 5.432302644444927e-06,\n",
       " 'follow': 0.0004563134221333739,\n",
       " 'clinic': 7.061993437778406e-05,\n",
       " 'trial': 0.000450881119488929,\n",
       " 'anti': 6.518763173333913e-05,\n",
       " 'properti': 0.0008854653310445232,\n",
       " 'look': 0.0023467547424002085,\n",
       " 'current': 0.0009289237522000826,\n",
       " 'harvest': 0.0012277003976445535,\n",
       " 'consid': 0.000668173225266726,\n",
       " 'resourc': 0.0009343560548445275,\n",
       " 'lot': 0.0012874557267334478,\n",
       " 'potenti': 0.0003911257904000348,\n",
       " 'econom': 0.000450881119488929,\n",
       " 'benefit': 0.000597553290888942,\n",
       " 'side': 0.00022815671106668695,\n",
       " 'processor': 0.0005595271723778275,\n",
       " 'fail': 0.0004019903956889246,\n",
       " 'meet': 0.001119054344755655,\n",
       " 'kangaroo': 0.0001195106581777884,\n",
       " 'meat': 0.0010864605288889854,\n",
       " 'demand': 0.001037569805088981,\n",
       " 'intern': 0.0008637361204667434,\n",
       " 'pressur': 0.0005052041459333782,\n",
       " 'high': 0.0012548619108667782,\n",
       " 'quota': 7.061993437778406e-05,\n",
       " 'met': 0.00019013059255557245,\n",
       " 'leav': 0.000445448816844484,\n",
       " 'gap': 4.345842115555942e-05,\n",
       " 'phil': 8.14845396666739e-05,\n",
       " 'franolick': 5.432302644444927e-06,\n",
       " 'king': 0.00016840138197779276,\n",
       " 'perth': 0.000222724408422242,\n",
       " 'european': 0.0002553182242889116,\n",
       " 'attract': 0.00036396427717781015,\n",
       " 'low': 0.0007659546728667348,\n",
       " 'fat': 6.518763173333913e-05,\n",
       " 'content': 9.234914495556377e-05,\n",
       " 'russian': 2.7161513222224636e-05,\n",
       " 'consum': 0.0006410117120445014,\n",
       " 'larg': 0.0006192825014667217,\n",
       " 'amount': 0.0004400165142000391,\n",
       " 'sausag': 5.432302644444927e-06,\n",
       " 'minc': 5.432302644444927e-06,\n",
       " 'as': 0.00030420894808891595,\n",
       " 'far': 0.0009071945416223028,\n",
       " 'know': 0.000901762238977858,\n",
       " 'everi': 0.00041285500097781446,\n",
       " 'skerrick': 5.432302644444927e-06,\n",
       " 'human': 0.0002661828295778014,\n",
       " 'consumpt': 0.00013580756611112317,\n",
       " 'goe': 0.00016840138197779276,\n",
       " 'warn': 0.000526933356511158,\n",
       " 'snake': 4.345842115555942e-05,\n",
       " 'alert': 5.432302644444927e-05,\n",
       " 'with': 0.000450881119488929,\n",
       " 'temperatur': 0.0003368027639555855,\n",
       " 'soar': 9.77814476000087e-05,\n",
       " 'summer': 0.0003585319745333652,\n",
       " 'southern': 0.0007876838834445144,\n",
       " 'keep': 0.0007768192781556246,\n",
       " 'open': 0.0007876838834445144,\n",
       " 'paddock': 0.0002172921057777971,\n",
       " 'especi': 0.0001575367766889029,\n",
       " 'area': 0.0019284674387779492,\n",
       " 'hit': 0.0008583038178222985,\n",
       " 'forc': 0.0009506529627778623,\n",
       " 'usual': 0.00022815671106668695,\n",
       " 'habitat': 4.345842115555942e-05,\n",
       " 'maria': 2.172921057777971e-05,\n",
       " 'dovey': 5.432302644444927e-06,\n",
       " 'reptil': 2.7161513222224636e-05,\n",
       " 'rescu': 6.518763173333913e-05,\n",
       " 'word': 9.234914495556377e-05,\n",
       " 'advic': 0.00013037526346667825,\n",
       " 'ani': 4.889072380000435e-05,\n",
       " 'regularli': 2.172921057777971e-05,\n",
       " 'outdoor': 1.0864605288889855e-05,\n",
       " 'bush': 0.0002770474348666913,\n",
       " 'land': 0.0009615175680667521,\n",
       " 'alway': 0.00034223506660003043,\n",
       " 'carri': 0.00034766736924447534,\n",
       " 'least': 0.0003802611851111449,\n",
       " 'bandag': 5.432302644444927e-06,\n",
       " 'mobil': 8.691684231111884e-05,\n",
       " 'phone': 0.0001955628952000174,\n",
       " 'aerial': 5.432302644444927e-05,\n",
       " 'spray': 0.00013580756611112317,\n",
       " 'control': 0.0007116316464222855,\n",
       " 'locust': 0.0002009951978444623,\n",
       " 'threat': 0.00028791204015558115,\n",
       " 'begun': 0.00014667217140001303,\n",
       " 'plagu': 0.00011407835553334348,\n",
       " 'made': 0.0009071945416223028,\n",
       " 'continu': 0.001336346450533452,\n",
       " 'rain': 0.0013852371743334564,\n",
       " 'warm': 0.00014123986875556812,\n",
       " 'weather': 0.000521501053866713,\n",
       " 'assist': 0.0006518763173333912,\n",
       " 'breed': 0.0004019903956889246,\n",
       " 'condit': 0.0011462158579778796,\n",
       " 'pastur': 0.00019013059255557245,\n",
       " 'crop': 0.002824797375111362,\n",
       " 'pest': 0.0002661828295778014,\n",
       " 'injuri': 3.8026118511114494e-05,\n",
       " 'baton': 1.6296907933334782e-05,\n",
       " 'runner': 1.6296907933334782e-05,\n",
       " 'queensland': 0.00261293757197801,\n",
       " 'knee': 4.889072380000435e-05,\n",
       " 'run': 0.0007876838834445144,\n",
       " 'queen': 2.7161513222224636e-05,\n",
       " 'relay': 1.6296907933334782e-05,\n",
       " 'lead': 0.0006301471067556116,\n",
       " 'commonwealth': 0.00014123986875556812,\n",
       " 'game': 0.0001195106581777884,\n",
       " 'stanthorp': 2.7161513222224636e-05,\n",
       " 'bill': 0.0005432302644444927,\n",
       " 'bonner': 5.432302644444927e-06,\n",
       " 'travel': 0.0002607505269333565,\n",
       " 'coff': 1.6296907933334782e-05,\n",
       " 'harbour': 4.889072380000435e-05,\n",
       " 'and': 0.0007659546728667348,\n",
       " 'prevent': 0.0002553182242889116,\n",
       " 'make': 0.0020588427022446274,\n",
       " 'dash': 2.172921057777971e-05,\n",
       " 'about': 0.00011407835553334348,\n",
       " 'jump': 0.00029334434280002607,\n",
       " 'motorbik': 1.6296907933334782e-05,\n",
       " 'twist': 5.432302644444927e-06,\n",
       " 'ligament': 5.432302644444927e-06,\n",
       " 'abl': 0.0007333608570000652,\n",
       " 'sinc': 0.0007387931596445101,\n",
       " 'promis': 0.00022815671106668695,\n",
       " 'adrenalin': 5.432302644444927e-06,\n",
       " 'rush': 4.889072380000435e-05,\n",
       " 'nz': 9.77814476000087e-05,\n",
       " 'appl': 0.0005106364485778232,\n",
       " 'debat': 0.0004019903956889246,\n",
       " 'zealand': 0.0005106364485778232,\n",
       " 'newspap': 4.889072380000435e-05,\n",
       " 'member': 0.0006301471067556116,\n",
       " 'voic': 7.605223702222899e-05,\n",
       " 'concern': 0.0016894461224223724,\n",
       " 'regard': 0.00015210447404445798,\n",
       " 'legitimaci': 1.0864605288889855e-05,\n",
       " 'scienc': 0.00023358901371113187,\n",
       " 'draft': 0.00029334434280002607,\n",
       " 'import': 0.0018252536885334956,\n",
       " 'risk': 0.0005758240803111623,\n",
       " 'analysi': 0.00014123986875556812,\n",
       " 'last': 0.002401077768844658,\n",
       " 'month': 0.0019990873731557334,\n",
       " 'biosecur': 0.0002661828295778014,\n",
       " 'ira': 2.172921057777971e-05,\n",
       " 'determin': 0.0001195106581777884,\n",
       " 'fireblight': 2.7161513222224636e-05,\n",
       " 'diseas': 0.0011679450685556593,\n",
       " 'under': 0.0002444536190000217,\n",
       " 'rule': 0.00041828730362225943,\n",
       " 'refus': 0.00014123986875556812,\n",
       " 'scientif': 0.00014123986875556812,\n",
       " 'evid': 0.0003748288824667,\n",
       " 'substanti': 0.0001195106581777884,\n",
       " 'accord': 0.00019013059255557245,\n",
       " 'profession': 0.00015210447404445798,\n",
       " 'matur': 2.7161513222224636e-05,\n",
       " 'pose': 9.234914495556377e-05,\n",
       " 'howev': 0.0001955628952000174,\n",
       " 'spokesperson': 3.8026118511114494e-05,\n",
       " 'base': 0.0007931161860889594,\n",
       " 'access': 0.0007387931596445101,\n",
       " 'group': 0.0016514200039112578,\n",
       " 'allison': 5.432302644444927e-06,\n",
       " 'mathemat': 1.0864605288889855e-05,\n",
       " 'model': 0.0001792659872666826,\n",
       " 'formula': 2.7161513222224636e-05,\n",
       " 'assess': 0.00028791204015558115,\n",
       " 'report': 0.0015970969774668086,\n",
       " 'publish': 6.518763173333913e-05,\n",
       " 'decemb': 0.00013037526346667825,\n",
       " 'anyon': 7.061993437778406e-05,\n",
       " 'compet': 0.00028791204015558115,\n",
       " 'reproduc': 1.0864605288889855e-05,\n",
       " 'given': 0.0005486625670889377,\n",
       " 'chairman': 0.0007985484887334043,\n",
       " 'task': 0.0001955628952000174,\n",
       " 'corboy': 5.432302644444927e-06,\n",
       " 'surpris': 0.0002770474348666913,\n",
       " 'exercis': 8.691684231111884e-05,\n",
       " 'present': 0.00030420894808891595,\n",
       " 'younger': 5.432302644444927e-05,\n",
       " 'tree': 0.0005540948697333826,\n",
       " 'pear': 0.0001195106581777884,\n",
       " 'plot': 2.7161513222224636e-05,\n",
       " 'young': 0.0006410117120445014,\n",
       " 'william': 8.14845396666739e-05,\n",
       " 'fruit': 0.0017492014515112667,\n",
       " 'plant': 0.0014775863192890203,\n",
       " 'six': 0.0006518763173333912,\n",
       " 'old': 0.0003205058560222507,\n",
       " 'part': 0.0009886790812889768,\n",
       " 'experiment': 1.6296907933334782e-05,\n",
       " 'project': 0.001032137502444536,\n",
       " 'fund': 0.0010810282262445406,\n",
       " 'spc': 1.6296907933334782e-05,\n",
       " 'ardmona': 1.6296907933334782e-05,\n",
       " 'can': 4.345842115555942e-05,\n",
       " 'associ': 0.001640555398622368,\n",
       " 'horticultur': 0.00041285500097781446,\n",
       " 'lenn': 5.432302644444927e-06,\n",
       " 'famili': 0.0006192825014667217,\n",
       " 'orchard': 0.0001955628952000174,\n",
       " 'locat': 4.345842115555942e-05,\n",
       " 'consult': 0.0003205058560222507,\n",
       " 'ba': 5.432302644444927e-06,\n",
       " 'van': 3.2593815866669564e-05,\n",
       " 'den': 5.432302644444927e-06,\n",
       " 'aim': 0.00028247973751113623,\n",
       " 'increas': 0.0018904413202668346,\n",
       " 'number': 0.0014341278981334607,\n",
       " 'grown': 0.00028247973751113623,\n",
       " 'encourag': 0.00040742269833336954,\n",
       " 'earlier': 0.0004617457247778188,\n",
       " 'feder': 0.0037645857326003345,\n",
       " 'nation': 0.0032104908628669522,\n",
       " 'mp': 0.0001792659872666826,\n",
       " 'park': 0.00027161513222224634,\n",
       " 'cobb': 2.172921057777971e-05,\n",
       " 'brought': 0.00019013059255557245,\n",
       " 'thi': 0.000516068751222268,\n",
       " 'examin': 0.00017383368462223767,\n",
       " 'role': 0.00017383368462223767,\n",
       " 'on': 0.00012494296082223334,\n",
       " 'top': 0.0005758240803111623,\n",
       " 'parti': 0.000445448816844484,\n",
       " 'reinvent': 1.0864605288889855e-05,\n",
       " 'becom': 0.0006192825014667217,\n",
       " 'pro': 1.0864605288889855e-05,\n",
       " 'activ': 0.0001629690793333478,\n",
       " 'differ': 0.0005703917776667173,\n",
       " 'liber': 0.0002009951978444623,\n",
       " 'repres': 0.00033137046131114054,\n",
       " 'better': 0.0010212728971556464,\n",
       " 'death': 0.0002009951978444623,\n",
       " 'believ': 0.0010104082918667564,\n",
       " 'due': 0.0008746007257556333,\n",
       " 'lack': 0.0005106364485778232,\n",
       " 'oxygen': 3.2593815866669564e-05,\n",
       " 'depart': 0.000890897633688968,\n",
       " 'primari': 0.00046717802742226374,\n",
       " 'kill': 0.0002444536190000217,\n",
       " 'lachlan': 5.432302644444927e-05,\n",
       " 'upstream': 1.6296907933334782e-05,\n",
       " 'lake': 0.00010864605288889855,\n",
       " 'cargelligo': 5.432302644444927e-06,\n",
       " 'weir': 3.8026118511114494e-05,\n",
       " 'between': 1.6296907933334782e-05,\n",
       " 'nativ': 0.0003856934877555898,\n",
       " 'die': 0.00021185980313335217,\n",
       " 'incid': 7.605223702222899e-05,\n",
       " 'district': 0.0003259381586666956,\n",
       " 'fisheri': 0.0002661828295778014,\n",
       " 'offic': 0.0004780426327111536,\n",
       " 'phillip': 6.518763173333913e-05,\n",
       " 'mccarthi': 5.432302644444927e-06,\n",
       " 'murray': 0.0010755959236000956,\n",
       " 'cod': 3.8026118511114494e-05,\n",
       " 'bream': 5.432302644444927e-06,\n",
       " 'golden': 4.889072380000435e-05,\n",
       " 'silver': 6.518763173333913e-05,\n",
       " 'perch': 5.432302644444927e-06,\n",
       " 'introduc': 0.0003150735533778058,\n",
       " 'speci': 0.0001629690793333478,\n",
       " 'suggest': 0.0004563134221333739,\n",
       " 'result': 0.0007007670411333957,\n",
       " 'water': 0.00521501053866713,\n",
       " 'flow': 0.0003965580930444797,\n",
       " 'chemic': 0.00036396427717781015,\n",
       " 'poison': 7.605223702222899e-05,\n",
       " 'veget': 0.000668173225266726,\n",
       " 'label': 0.0003911257904000348,\n",
       " 'ausveg': 0.00011407835553334348,\n",
       " 'law': 0.000516068751222268,\n",
       " 'joke': 5.432302644444927e-06,\n",
       " 'nobodi': 2.7161513222224636e-05,\n",
       " 'enforc': 8.14845396666739e-05,\n",
       " 'non': 0.00015210447404445798,\n",
       " 'exist': 0.00028247973751113623,\n",
       " 'fool': 5.432302644444927e-06,\n",
       " 'buy': 0.0009126268442667478,\n",
       " 'fals': 3.2593815866669564e-05,\n",
       " 'poor': 0.00028791204015558115,\n",
       " 'red': 0.0002172921057777971,\n",
       " 'tape': 1.6296907933334782e-05,\n",
       " 'slow': 0.0002553182242889116,\n",
       " 'west': 0.0011733773712001043,\n",
       " 'kimberley': 0.0001955628952000174,\n",
       " 'wa': 0.0008474392125334087,\n",
       " 'bowl': 1.6296907933334782e-05,\n",
       " 'soil': 0.0001629690793333478,\n",
       " 'avail': 0.0004943395406444884,\n",
       " 'sunshin': 2.172921057777971e-05,\n",
       " 'face': 0.000597553290888942,\n",
       " 'hurdl': 3.2593815866669564e-05,\n",
       " 'expand': 0.00014123986875556812,\n",
       " 'sweet': 3.8026118511114494e-05,\n",
       " 'potato': 0.0001955628952000174,\n",
       " 'paul': 0.00020642750048890723,\n",
       " 'higgin': 3.8026118511114494e-05,\n",
       " 'reli': 0.0001195106581777884,\n",
       " 'gener': 0.0009615175680667521,\n",
       " 'broom': 7.605223702222899e-05,\n",
       " 'horticulturalist': 4.889072380000435e-05,\n",
       " 'connect': 9.77814476000087e-05,\n",
       " 'grid': 3.2593815866669564e-05,\n",
       " 'appli': 0.00022815671106668695,\n",
       " 'nine': 0.00014667217140001303,\n",
       " 'kilometr': 0.00048347493535559854,\n",
       " 'extens': 0.00013580756611112317,\n",
       " 'bring': 0.00041285500097781446,\n",
       " 'line': 0.0005486625670889377,\n",
       " 'power': 0.0006301471067556116,\n",
       " 'custom': 0.0003150735533778058,\n",
       " 'cost': 0.0019230351361335042,\n",
       " 'regardless': 2.7161513222224636e-05,\n",
       " 'distanc': 8.691684231111884e-05,\n",
       " 'acquisit': 2.172921057777971e-05,\n",
       " 'caus': 0.0008257100019556289,\n",
       " 'headach': 2.7161513222224636e-05,\n",
       " 'purchas': 0.00017383368462223767,\n",
       " 'adjoin': 1.0864605288889855e-05,\n",
       " 'june': 0.00018469828991112753,\n",
       " 'titl': 0.00021185980313335217,\n",
       " 'improv': 0.0008039807913778492,\n",
       " 'clear': 0.000521501053866713,\n",
       " 'resolv': 0.0001575367766889029,\n",
       " 'survey': 0.000521501053866713,\n",
       " 'bed': 5.432302644444927e-05,\n",
       " 'moni': 5.432302644444927e-06,\n",
       " 'paid': 0.00040742269833336954,\n",
       " 'plan': 0.00186327980704461,\n",
       " 'infrastructur': 0.00034766736924447534,\n",
       " 'alannah': 5.432302644444927e-06,\n",
       " 'mctiernan': 1.0864605288889855e-05,\n",
       " 'seven': 0.00028791204015558115,\n",
       " 'delay': 0.00042371960626670435,\n",
       " 'regrett': 1.0864605288889855e-05,\n",
       " 'awar': 0.0001955628952000174,\n",
       " 'hold': 0.0005052041459333782,\n",
       " 'correctli': 1.6296907933334782e-05,\n",
       " 'zone': 9.234914495556377e-05,\n",
       " 'ms': 0.00014123986875556812,\n",
       " 'favour': 9.234914495556377e-05,\n",
       " 'wait': 0.00023358901371113187,\n",
       " 'offici': 0.00029334434280002607,\n",
       " 'plough': 8.691684231111884e-05,\n",
       " 'pindan': 5.432302644444927e-06,\n",
       " 'richmond': 3.8026118511114494e-05,\n",
       " 'stay': 0.00023358901371113187,\n",
       " 'fisher': 0.00013580756611112317,\n",
       " 'decid': 0.0004019903956889246,\n",
       " 'commerci': 0.00040742269833336954,\n",
       " 'recreat': 3.2593815866669564e-05,\n",
       " 'dead': 7.061993437778406e-05,\n",
       " 'eel': 5.432302644444927e-06,\n",
       " 'near': 0.0007876838834445144,\n",
       " 'wardel': 5.432302644444927e-06,\n",
       " 'thursday': 2.7161513222224636e-05,\n",
       " 'alarmingli': 5.432302644444927e-06,\n",
       " 'act': 0.00041828730362225943,\n",
       " 'anthoni': 2.172921057777971e-05,\n",
       " 'hirst': 5.432302644444927e-06,\n",
       " 'inspector': 4.345842115555942e-05,\n",
       " 'sampl': 7.605223702222899e-05,\n",
       " 'tomorrow': 0.0001792659872666826,\n",
       " 'rural': 0.0020425457943112927,\n",
       " 'woman': 4.345842115555942e-05,\n",
       " 'finalist': 2.172921057777971e-05,\n",
       " 'next': 0.0016568523065557028,\n",
       " 'seem': 0.000298776645444471,\n",
       " 'like': 0.0019230351361335042,\n",
       " 'etern': 1.0864605288889855e-05,\n",
       " 'rosebank': 5.432302644444927e-06,\n",
       " 'catherin': 1.6296907933334782e-05,\n",
       " 'ford': 2.172921057777971e-05,\n",
       " 'quest': 5.432302644444927e-06,\n",
       " 'find': 0.0009560852654223073,\n",
       " 'macadamia': 5.432302644444927e-05,\n",
       " 'win': 0.0003368027639555855,\n",
       " 'guid': 1.0864605288889855e-05,\n",
       " 'admit': 0.00014123986875556812,\n",
       " 'amaz': 6.518763173333913e-05,\n",
       " 'even': 0.0008365746072445189,\n",
       " 'dse': 5.432302644444927e-06,\n",
       " 'proactiv': 2.7161513222224636e-05,\n",
       " 'burn': 0.0001629690793333478,\n",
       " 'yesterday': 0.000673605527911171,\n",
       " 'heard': 0.00020642750048890723,\n",
       " 'cfa': 2.172921057777971e-05,\n",
       " 'feel': 0.0003150735533778058,\n",
       " 'intens': 5.432302644444927e-05,\n",
       " 'lessen': 5.432302644444927e-06,\n",
       " 'fuel': 0.0009452206601334173,\n",
       " 'reduct': 0.0001575367766889029,\n",
       " 'charley': 5.432302644444927e-06,\n",
       " 'defegley': 5.432302644444927e-06,\n",
       " 'ararat': 1.6296907933334782e-05,\n",
       " 'vff': 0.00011407835553334348,\n",
       " 'branch': 0.00010321375024445361,\n",
       " 'secretari': 0.00022815671106668695,\n",
       " 'division': 5.432302644444927e-06,\n",
       " 'command': 2.172921057777971e-05,\n",
       " 'felt': 5.432302644444927e-05,\n",
       " 'enough': 0.0007496577649334,\n",
       " 'contribut': 0.0002661828295778014,\n",
       " 'destruct': 9.77814476000087e-05,\n",
       " 'horsham': 1.6296907933334782e-05,\n",
       " 'geoff': 8.14845396666739e-05,\n",
       " 'evan': 4.889072380000435e-05,\n",
       " 'agenc': 0.0001575367766889029,\n",
       " 'polici': 0.0004563134221333739,\n",
       " 'hamper': 3.2593815866669564e-05,\n",
       " 'landown': 3.8026118511114494e-05,\n",
       " 'keen': 0.00010864605288889855,\n",
       " 'besid': 1.0864605288889855e-05,\n",
       " 'allow': 0.000744225462288955,\n",
       " 'small': 0.0006573086199778362,\n",
       " 'window': 3.8026118511114494e-05,\n",
       " 'opportun': 0.00042371960626670435,\n",
       " 'strong': 0.0005486625670889377,\n",
       " 'divis': 9.234914495556377e-05,\n",
       " 'within': 0.000521501053866713,\n",
       " 'less': 0.0006084178961778319,\n",
       " 'hectar': 0.000450881119488929,\n",
       " 'burnt': 7.061993437778406e-05,\n",
       " 'ag': 5.432302644444927e-05,\n",
       " 'speedi': 1.0864605288889855e-05,\n",
       " 'aid': 0.0002009951978444623,\n",
       " 'announc': 0.0007768192781556246,\n",
       " 'commit': 0.0002172921057777971,\n",
       " 'dollar': 0.0008528715151778536,\n",
       " 'recoveri': 0.00018469828991112753,\n",
       " 'blaze': 9.234914495556377e-05,\n",
       " 'ministeri': 4.889072380000435e-05,\n",
       " 'need': 0.0026346667825557896,\n",
       " 'hardship': 2.172921057777971e-05,\n",
       " 'suppli': 0.0011842419764889942,\n",
       " 'firefight': 0.0001575367766889029,\n",
       " 'victorian': 0.0007822515808000696,\n",
       " 'haulag': 2.7161513222224636e-05,\n",
       " 'contractor': 0.00015210447404445798,\n",
       " 'grassroot': 5.432302644444927e-06,\n",
       " 'affect': 0.0009343560548445275,\n",
       " 'semi': 4.889072380000435e-05,\n",
       " 'collect': 0.00019013059255557245,\n",
       " 'moyston': 5.432302644444927e-06,\n",
       " 'warrnambool': 2.7161513222224636e-05,\n",
       " 'surviv': 0.00020642750048890723,\n",
       " 'bruce': 0.00011407835553334348,\n",
       " 'mckay': 2.7161513222224636e-05,\n",
       " 'farm': 0.003042089480889159,\n",
       " 'sheep': 0.0013798048716890114,\n",
       " 'fix': 0.0001955628952000174,\n",
       " 'fenc': 0.0001195106581777884,\n",
       " 'willaura': 5.432302644444927e-06,\n",
       " 'tim': 0.00012494296082223334,\n",
       " 'wrench': 1.0864605288889855e-05,\n",
       " 'coop': 5.432302644444927e-06,\n",
       " 'came': 0.0002009951978444623,\n",
       " 'around': 0.0016568523065557028,\n",
       " 'haulier': 5.432302644444927e-06,\n",
       " 'scott': 0.0001629690793333478,\n",
       " 'wilson': 7.605223702222899e-05,\n",
       " 'koroit': 5.432302644444927e-06,\n",
       " 'went': 0.0001955628952000174,\n",
       " 'neil': 0.00010321375024445361,\n",
       " 'jame': 0.0001792659872666826,\n",
       " 'dpi': 4.889072380000435e-05,\n",
       " 'charli': 3.2593815866669564e-05,\n",
       " 'defegeley': 5.432302644444927e-06,\n",
       " 'blame': 0.0004400165142000391,\n",
       " 'hike': 8.14845396666739e-05,\n",
       " 'fresh': 0.0001955628952000174,\n",
       " 'fluctuat': 3.8026118511114494e-05,\n",
       " 'per': 0.003449512179222529,\n",
       " 'cent': 0.003731991916733665,\n",
       " 'latest': 0.00048347493535559854,\n",
       " 'rabobank': 7.605223702222899e-05,\n",
       " 'milk': 0.00046717802742226374,\n",
       " 'ben': 9.234914495556377e-05,\n",
       " 'russel': 9.234914495556377e-05,\n",
       " 'money': 0.0008420069098889637,\n",
       " 'case': 0.0006627409226222811,\n",
       " 'pass': 0.0002607505269333565,\n",
       " 'contrari': 1.0864605288889855e-05,\n",
       " 'see': 0.0014395602007779057,\n",
       " 'retail': 0.0003150735533778058,\n",
       " 'fall': 0.0008854653310445232,\n",
       " 'tradit': 0.0003205058560222507,\n",
       " 'egypt': 7.605223702222899e-05,\n",
       " 'pakistan': 2.7161513222224636e-05,\n",
       " 'china': 0.0004997718432889333,\n",
       " 'india': 0.000298776645444471,\n",
       " 'deliv': 0.00030420894808891595,\n",
       " 'expect': 0.0018741444123335,\n",
       " 'contract': 0.0007007670411333957,\n",
       " 'record': 0.0011570804632667696,\n",
       " 'tender': 0.00027161513222224634,\n",
       " 'fierc': 3.2593815866669564e-05,\n",
       " 'trader': 0.0001629690793333478,\n",
       " 'lloyd': 4.345842115555942e-05,\n",
       " 'georg': 0.0001195106581777884,\n",
       " 'rang': 0.0003856934877555898,\n",
       " 'factor': 0.00018469828991112753,\n",
       " 'period': 0.0002390213163555768,\n",
       " 'signific': 0.0004997718432889333,\n",
       " 'tonnag': 4.345842115555942e-05,\n",
       " 'recent': 0.0004997718432889333,\n",
       " 'quantiti': 3.8026118511114494e-05,\n",
       " 'whole': 0.0003802611851111449,\n",
       " 'bunch': 2.172921057777971e-05,\n",
       " 'broader': 4.345842115555942e-05,\n",
       " 'wherebi': 1.6296907933334782e-05,\n",
       " 'sell': 0.0010267051998000913,\n",
       " 'quiet': 2.7161513222224636e-05,\n",
       " 'david': 0.000521501053866713,\n",
       " 'johnson': 8.691684231111884e-05,\n",
       " 'impact': 0.0009071945416223028,\n",
       " 'lost': 0.0003748288824667,\n",
       " 'submit': 6.518763173333913e-05,\n",
       " 'particip': 9.234914495556377e-05,\n",
       " 'absolut': 0.00013037526346667825,\n",
       " 'probabl': 0.0008148453966667391,\n",
       " 'rel': 9.234914495556377e-05,\n",
       " 'compar': 0.000222724408422242,\n",
       " 'structur': 0.0001575367766889029,\n",
       " 'previou': 0.00021185980313335217,\n",
       " 'fact': 0.00033137046131114054,\n",
       " 'extrem': 0.0002607505269333565,\n",
       " 'comment': 0.00013037526346667825,\n",
       " 'season': 0.0017654983594446014,\n",
       " 'estim': 0.0004943395406444884,\n",
       " 'lift': 0.000298776645444471,\n",
       " 'hard': 0.00048347493535559854,\n",
       " 'noodl': 5.432302644444927e-06,\n",
       " 'grade': 9.77814476000087e-05,\n",
       " 'there': 0.0016188261880445884,\n",
       " 'mix': 0.0002172921057777971,\n",
       " 'outlook': 0.000298776645444471,\n",
       " 'soy': 2.7161513222224636e-05,\n",
       " 'bean': 6.518763173333913e-05,\n",
       " 'us': 0.003378892244844745,\n",
       " 'ukrain': 5.432302644444927e-06,\n",
       " 'brazil': 0.00012494296082223334,\n",
       " 'mean': 0.0007550900675778449,\n",
       " 'canola': 0.0004997718432889333,\n",
       " 'altern': 0.00035309967188892026,\n",
       " 'seed': 0.00041828730362225943,\n",
       " 'syme': 1.6296907933334782e-05,\n",
       " 'malt': 4.345842115555942e-05,\n",
       " 'barley': 0.00029334434280002607,\n",
       " 'chines': 0.0001792659872666826,\n",
       " 'soon': 0.0002607505269333565,\n",
       " 'pick': 0.00028247973751113623,\n",
       " 'at': 0.0002661828295778014,\n",
       " 'stock': 0.001336346450533452,\n",
       " 'canadian': 0.00010864605288889855,\n",
       " 'offer': 0.000901762238977858,\n",
       " 'marketplac': 3.8026118511114494e-05,\n",
       " 'good': 0.0014775863192890203,\n",
       " 'posit': 0.0006301471067556116,\n",
       " 'cours': 0.0003585319745333652,\n",
       " 'europ': 0.00030420894808891595,\n",
       " 'august': 0.00013580756611112317,\n",
       " 'away': 0.0003856934877555898,\n",
       " 'yet': 0.0003585319745333652,\n",
       " 'onc': 6.518763173333913e-05,\n",
       " 'emerg': 0.000222724408422242,\n",
       " ...}"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_rural"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'cystic': 4.0789133774762405e-05,\n",
       " 'fibrosi': 4.5321259749736004e-05,\n",
       " 'affect': 0.0007160759040458289,\n",
       " 'children': 0.0007387365339206969,\n",
       " 'young': 0.0004396162195724392,\n",
       " 'adult': 0.0003671022039728616,\n",
       " 'us': 0.0023068521212615627,\n",
       " 'alon': 0.00011783527534931361,\n",
       " 'inhal': 5.8917637674656804e-05,\n",
       " 'mist': 4.5321259749736004e-05,\n",
       " 'salt': 8.157826754952481e-05,\n",
       " 'water': 0.001667822358790285,\n",
       " 'reduc': 0.0006163691325964096,\n",
       " 'pu': 9.064251949947201e-06,\n",
       " 'infect': 0.0007659292897705384,\n",
       " 'fill': 0.00016768866107402322,\n",
       " 'airway': 2.7192755849841602e-05,\n",
       " 'suffer': 0.000249266928623548,\n",
       " 'although': 0.0005257266130969376,\n",
       " 'side': 0.00040789133774762405,\n",
       " 'effect': 0.0015635834613658922,\n",
       " 'includ': 0.0018219146419393873,\n",
       " 'nasti': 2.2660629874868002e-05,\n",
       " 'cough': 2.2660629874868002e-05,\n",
       " 'fit': 0.00022660629874868003,\n",
       " 'harsh': 1.3596377924920801e-05,\n",
       " 'tast': 0.000135963779249208,\n",
       " 'that': 0.0006480940144212248,\n",
       " 'conclus': 0.0001722207870489968,\n",
       " 'two': 0.002474540782335586,\n",
       " 'studi': 0.00498533857247096,\n",
       " 'publish': 0.0013007201548174232,\n",
       " 'week': 0.0009744070846193241,\n",
       " 'issu': 0.0011919491314180569,\n",
       " 'the': 0.01380938784574456,\n",
       " 'new': 0.0046318327464230195,\n",
       " 'england': 0.0001722207870489968,\n",
       " 'journal': 0.0018672359016891234,\n",
       " 'medicin': 0.0004033592117726504,\n",
       " 'they': 0.0020485209406880672,\n",
       " 'found': 0.0035577188903542765,\n",
       " 'content': 0.00019941354289883842,\n",
       " 'improv': 0.00046680897542228085,\n",
       " 'lung': 0.00022207417277370642,\n",
       " 'function': 0.00040789133774762405,\n",
       " 'case': 0.0008611039352449841,\n",
       " 'produc': 0.0012871237768925024,\n",
       " 'less': 0.000988003462544245,\n",
       " 'absente': 9.064251949947201e-06,\n",
       " 'school': 0.00040789133774762405,\n",
       " 'work': 0.0025107977901353746,\n",
       " 'progress': 0.00010423889742439281,\n",
       " 'frequent': 0.00012236740132428721,\n",
       " 'fatal': 9.064251949947201e-05,\n",
       " 'genet': 0.0011964812573930304,\n",
       " 'diseas': 0.0010967744859436113,\n",
       " 'mark': 0.0003444415740979936,\n",
       " 'thicken': 2.2660629874868002e-05,\n",
       " 'mucu': 1.3596377924920801e-05,\n",
       " 'make': 0.0027600647187589226,\n",
       " 'harder': 0.00011330314937434001,\n",
       " 'clear': 0.00039429495982270325,\n",
       " 'debri': 0.0001903492909488912,\n",
       " 'bacteria': 0.0005529193689467793,\n",
       " 'solut': 0.00018581716497391762,\n",
       " 'realli': 0.0006662225183211193,\n",
       " 'open': 0.00043055196762249205,\n",
       " 'avenu': 1.8128503899894402e-05,\n",
       " 'approach': 0.000317248818248152,\n",
       " 'patient': 0.0009018930690197465,\n",
       " 'treat': 0.00042148771567254484,\n",
       " 'say': 0.024464416012907494,\n",
       " 'dr': 0.0028643036161833156,\n",
       " 'gail': 4.532125974973601e-06,\n",
       " 'weinmann': 9.064251949947201e-06,\n",
       " 'nation': 0.0011692885015431888,\n",
       " 'heart': 0.0003806985818977824,\n",
       " 'blood': 0.0006979474001459345,\n",
       " 'institut': 0.0009472143287694825,\n",
       " 'sponsor': 2.7192755849841602e-05,\n",
       " 'one': 0.0045955757386232305,\n",
       " 'elkin': 2.7192755849841602e-05,\n",
       " 'royal': 0.00020394566887381202,\n",
       " 'princ': 2.7192755849841602e-05,\n",
       " 'alfr': 2.2660629874868002e-05,\n",
       " 'hospit': 0.0003308451961730728,\n",
       " 'sydney': 0.0005347908650468849,\n",
       " 'australia': 0.001998667554963358,\n",
       " 'colleagu': 0.0011103708638685321,\n",
       " 'author': 0.0010242604703440336,\n",
       " 'team': 0.002356705506986272,\n",
       " 'volunt': 0.00020394566887381202,\n",
       " 'regularli': 9.51746454744456e-05,\n",
       " 'salti': 1.3596377924920801e-05,\n",
       " 'fewer': 0.000135963779249208,\n",
       " 'breath': 0.00016768866107402322,\n",
       " 'problem': 0.0008248469274451953,\n",
       " 'ad': 0.0004124234637225976,\n",
       " 'surfac': 0.0009018930690197465,\n",
       " 'benefici': 6.344976364963041e-05,\n",
       " 'conclud': 0.0002855239364233368,\n",
       " 'all': 0.00021754204679873282,\n",
       " 'first': 0.0022841914913866944,\n",
       " 'chemic': 0.0007251401559957761,\n",
       " 'tri': 0.0006526261403961985,\n",
       " 'passag': 4.5321259749736004e-05,\n",
       " 'much': 0.0013868305483419217,\n",
       " 'possibl': 0.001123967241793453,\n",
       " 'in': 0.0020711815705629355,\n",
       " 'second': 0.0005574514949217528,\n",
       " 'base': 0.0011103708638685321,\n",
       " 'assist': 0.00026739543252344243,\n",
       " 'profsesor': 4.532125974973601e-06,\n",
       " 'scott': 1.8128503899894402e-05,\n",
       " 'donaldson': 4.532125974973601e-06,\n",
       " 'univers': 0.004482272589248891,\n",
       " 'north': 0.000498533857247096,\n",
       " 'carolina': 5.4385511699683204e-05,\n",
       " 'chapel': 1.3596377924920801e-05,\n",
       " 'hill': 9.51746454744456e-05,\n",
       " 'sustain': 0.000135963779249208,\n",
       " 'acceler': 0.0001540922831491024,\n",
       " 'clearanc': 9.064251949947201e-06,\n",
       " 'help': 0.0015817119652657865,\n",
       " 'hydrat': 1.3596377924920801e-05,\n",
       " 'accompani': 7.70461415745512e-05,\n",
       " 'editori': 1.8128503899894402e-05,\n",
       " 'felix': 4.532125974973601e-06,\n",
       " 'ratjen': 9.064251949947201e-06,\n",
       " 'sick': 0.0001404959052241816,\n",
       " 'toronto': 3.6257007799788805e-05,\n",
       " 'canada': 0.00012236740132428721,\n",
       " 'cite': 5.4385511699683204e-05,\n",
       " 'sever': 0.0007523329118456177,\n",
       " 'unpleas': 1.3596377924920801e-05,\n",
       " 'treatment': 0.00044414834554741284,\n",
       " 'bad': 0.00024020267667360082,\n",
       " 'lengthi': 4.532125974973601e-06,\n",
       " 'minut': 0.00036257007799788803,\n",
       " 'take': 0.0012553988950676873,\n",
       " 'administ': 3.6257007799788805e-05,\n",
       " 'he': 0.0019306856653387538,\n",
       " 'may': 0.0033039198357557547,\n",
       " 'receiv': 0.000317248818248152,\n",
       " 'best': 0.00044414834554741284,\n",
       " 'long': 0.001531858579541077,\n",
       " 'term': 0.0005347908650468849,\n",
       " 'antibiot': 0.0001540922831491024,\n",
       " 'would': 0.0032268736941812033,\n",
       " 'appear': 0.0009064251949947201,\n",
       " 'peopl': 0.0036392971579038013,\n",
       " 'get': 0.0016089047211156281,\n",
       " 'better': 0.0007523329118456177,\n",
       " 'drug': 0.0009653428326693769,\n",
       " 'limit': 0.00035803795202291446,\n",
       " 'inher': 3.6257007799788805e-05,\n",
       " 'mean': 0.0010287925963190074,\n",
       " 'step': 0.0003671022039728616,\n",
       " 'scientist': 0.003856839204702534,\n",
       " 'rescu': 4.5321259749736004e-05,\n",
       " 'mous': 0.00012236740132428721,\n",
       " 'immun': 0.00046680897542228085,\n",
       " 'system': 0.002225273853712038,\n",
       " 'overwhelm': 4.0789133774762405e-05,\n",
       " 'but': 0.004228473534650369,\n",
       " 'human': 0.003240470072106124,\n",
       " 'discov': 0.0006843510222210137,\n",
       " 'invad': 5.4385511699683204e-05,\n",
       " 'whole': 0.0003081845662982048,\n",
       " 'bodi': 0.0012327382651928192,\n",
       " 'like': 0.004187684400875607,\n",
       " 'malaria': 0.000158624409124076,\n",
       " 'disabl': 4.9853385724709604e-05,\n",
       " 'prevent': 0.0003897628338477296,\n",
       " 'detect': 0.0007206080300208024,\n",
       " 'fight': 0.000249266928623548,\n",
       " 'microorgan': 4.5321259749736004e-05,\n",
       " 'australian': 0.002166356216037381,\n",
       " 'german': 0.00016315653509904962,\n",
       " 'research': 0.009395097146120274,\n",
       " 'discoveri': 0.0005846442507715944,\n",
       " 'develop': 0.002030392436788173,\n",
       " 'vaccin': 0.0003806985818977824,\n",
       " 'restor': 0.00010423889742439281,\n",
       " 'today': 0.0013596377924920801,\n",
       " 'onlin': 0.0005257266130969376,\n",
       " 'natur': 0.0013777662963919744,\n",
       " 'immunolog': 3.1724881824815205e-05,\n",
       " 'respons': 0.0008203148014702216,\n",
       " 'alert': 9.51746454744456e-05,\n",
       " 'specialis': 9.064251949947201e-05,\n",
       " 'sentri': 4.532125974973601e-06,\n",
       " 'dendrit': 3.1724881824815205e-05,\n",
       " 'cell': 0.002483605034285533,\n",
       " 'viru': 0.0008112505495202745,\n",
       " 'captur': 0.000158624409124076,\n",
       " 'display': 0.0001767529130239704,\n",
       " 'fragment': 0.00016768866107402322,\n",
       " 'antigen': 4.0789133774762405e-05,\n",
       " 'pathogen': 4.9853385724709604e-05,\n",
       " 'process': 0.0009653428326693769,\n",
       " 'call': 0.0015273264535661034,\n",
       " 'cross': 0.000158624409124076,\n",
       " 'present': 0.000702479526120908,\n",
       " 'lead': 0.0008792324391448784,\n",
       " 'jose': 1.3596377924920801e-05,\n",
       " 'villadango': 2.2660629874868002e-05,\n",
       " 'immunologist': 1.3596377924920801e-05,\n",
       " 'walter': 3.1724881824815205e-05,\n",
       " 'eliza': 4.532125974973601e-06,\n",
       " 'medic': 0.000747800785870644,\n",
       " 'melbourn': 0.0002991203143482576,\n",
       " 'sepsi': 1.8128503899894402e-05,\n",
       " 'bacteri': 5.8917637674656804e-05,\n",
       " 'overstimul': 9.064251949947201e-06,\n",
       " 'thi': 0.0023974946407610345,\n",
       " 'result': 0.0014004269262668425,\n",
       " 'critic': 0.00024020267667360082,\n",
       " 'alarm': 3.6257007799788805e-05,\n",
       " 'shut': 4.9853385724709604e-05,\n",
       " 'occur': 0.0007659292897705384,\n",
       " 'local': 0.0002628633065484688,\n",
       " 'involv': 0.000838443305370116,\n",
       " 'throughout': 0.0001404959052241816,\n",
       " 'concentr': 0.00023567055069862721,\n",
       " 'respond': 0.00031271669227317844,\n",
       " 'a': 0.0023159163732115098,\n",
       " 'type': 0.0008928288170697993,\n",
       " 'manag': 0.0004894696052971489,\n",
       " 'mice': 0.0003535058260479408,\n",
       " 'compromis': 2.7192755849841602e-05,\n",
       " 'inject': 0.0001948814169238648,\n",
       " 'live': 0.0012191418872678985,\n",
       " 'made': 0.001268995272992608,\n",
       " 'expos': 0.00022660629874868003,\n",
       " 'secondari': 4.9853385724709604e-05,\n",
       " 'laboratori': 0.00034897370007296724,\n",
       " 'subsequ': 8.157826754952481e-05,\n",
       " 'find': 0.002664890073284477,\n",
       " 'show': 0.0022615308615118266,\n",
       " 'miss': 0.00021754204679873282,\n",
       " 'link': 0.0007795256676954592,\n",
       " 'immunosupress': 4.532125974973601e-06,\n",
       " 'anim': 0.0015907762172157338,\n",
       " 'capac': 0.000135963779249208,\n",
       " 'virus': 0.000294588188373284,\n",
       " 'we': 0.0018808322796140442,\n",
       " 'abl': 0.0007749935417204857,\n",
       " 'back': 0.0010061319664441393,\n",
       " 'strategi': 0.0001540922831491024,\n",
       " 'alreadi': 0.0006299655105213304,\n",
       " 'use': 0.00560170770506737,\n",
       " 'anti': 0.000158624409124076,\n",
       " 'cancer': 0.0009245536988946144,\n",
       " 'yet': 0.0005393229910218584,\n",
       " 'analys': 0.0005121302351720168,\n",
       " 'capsul': 9.51746454744456e-05,\n",
       " 'clue': 0.00012689952729926082,\n",
       " 'solar': 0.0008520396832950368,\n",
       " 'began': 0.00018581716497391762,\n",
       " 'seven': 0.00021754204679873282,\n",
       " 'year': 0.005030659832220697,\n",
       " 'quest': 9.064251949947201e-06,\n",
       " 'return': 0.000271927558498416,\n",
       " 'earth': 0.0015817119652657865,\n",
       " 'pristin': 2.2660629874868002e-05,\n",
       " 'sampl': 0.00047134110139725443,\n",
       " 'format': 0.0002130099208237592,\n",
       " 'end': 0.0005665157468717001,\n",
       " 'triumph': 4.532125974973601e-06,\n",
       " 'dark': 0.0003081845662982048,\n",
       " 'wet': 6.344976364963041e-05,\n",
       " 'utah': 2.2660629874868002e-05,\n",
       " 'desert': 9.51746454744456e-05,\n",
       " 'weekend': 6.344976364963041e-05,\n",
       " 'it': 0.003235937946131151,\n",
       " 'hard': 0.00029005606239831044,\n",
       " 'describ': 0.00048040535334720164,\n",
       " 'feel': 0.00037163432994783525,\n",
       " 'point': 0.0006888831481959872,\n",
       " 'mission': 0.00044414834554741284,\n",
       " 'washington': 9.970677144941921e-05,\n",
       " 'stardust': 4.0789133774762405e-05,\n",
       " 'professor': 0.0025697154278100315,\n",
       " 'don': 2.7192755849841602e-05,\n",
       " 'brownle': 4.5321259749736004e-05,\n",
       " 'said': 0.0004894696052971489,\n",
       " 'shortli': 4.5321259749736004e-05,\n",
       " 'probe': 0.00018581716497391762,\n",
       " 'touchdown': 4.532125974973601e-06,\n",
       " 'utc': 3.1724881824815205e-05,\n",
       " 'aedt': 9.064251949947201e-06,\n",
       " 'sunday': 2.2660629874868002e-05,\n",
       " 'travel': 0.0003308451961730728,\n",
       " 'almost': 0.0004033592117726504,\n",
       " 'billion': 0.00045321259749736006,\n",
       " 'mile': 0.000135963779249208,\n",
       " 'space': 0.0016360974769654697,\n",
       " 'visit': 0.00010877102339936641,\n",
       " 'comet': 0.000294588188373284,\n",
       " 'grab': 2.7192755849841602e-05,\n",
       " 'piec': 0.00029005606239831044,\n",
       " 'land': 0.0005030659832220696,\n",
       " 'morn': 8.61103935244984e-05,\n",
       " 'incred': 5.4385511699683204e-05,\n",
       " 'thrill': 2.2660629874868002e-05,\n",
       " 'releas': 0.0005257266130969376,\n",
       " 'preciou': 4.5321259749736004e-05,\n",
       " 'cargo': 2.7192755849841602e-05,\n",
       " 'high': 0.0012282061392178457,\n",
       " 'speed': 0.0003806985818977824,\n",
       " 'flybi': 1.8128503899894402e-05,\n",
       " 'send': 0.0001767529130239704,\n",
       " 'tyre': 2.2660629874868002e-05,\n",
       " 'size': 0.0008067184235453008,\n",
       " 'canist': 9.064251949947201e-06,\n",
       " 'soar': 2.7192755849841602e-05,\n",
       " 'blaze': 9.064251949947201e-06,\n",
       " 'kilometr': 0.0008475075573200633,\n",
       " 'per': 0.00039429495982270325,\n",
       " 'hour': 0.00045774472347233364,\n",
       " 'smash': 4.0789133774762405e-05,\n",
       " 'planet': 0.0015046658236912353,\n",
       " 'atmospher': 0.0007795256676954592,\n",
       " 'fastest': 3.1724881824815205e-05,\n",
       " 'object': 0.0006979474001459345,\n",
       " 'top': 0.00033990944812302003,\n",
       " 'record': 0.0006027727546714888,\n",
       " 'set': 0.0006118370066214361,\n",
       " 'apollo': 2.2660629874868002e-05,\n",
       " 'command': 6.344976364963041e-05,\n",
       " 'modul': 4.0789133774762405e-05,\n",
       " 'as': 0.0005211944871219641,\n",
       " 'heat': 0.0004894696052971489,\n",
       " 'shield': 9.064251949947201e-05,\n",
       " 'fri': 2.2660629874868002e-05,\n",
       " 'away': 0.000543855116996832,\n",
       " 'plume': 4.9853385724709604e-05,\n",
       " 'form': 0.001305252280792397,\n",
       " 'wake': 0.000135963779249208,\n",
       " 'look': 0.0016814187367152058,\n",
       " 'torch': 9.064251949947201e-06,\n",
       " 'visibl': 0.00022207417277370642,\n",
       " 'part': 0.0014457481860165786,\n",
       " 'western': 0.0003806985818977824,\n",
       " 'quit': 0.00042148771567254484,\n",
       " 'sure': 0.00022660629874868003,\n",
       " 'bright': 0.00018581716497391762,\n",
       " 'go': 0.0011738206275181625,\n",
       " 'think': 0.0012010133833680041,\n",
       " 'see': 0.0011738206275181625,\n",
       " 'anyth': 0.000249266928623548,\n",
       " 'slip': 4.9853385724709604e-05,\n",
       " 'outsid': 0.00018128503899894402,\n",
       " 'member': 0.00042148771567254484,\n",
       " 'spot': 0.00033990944812302003,\n",
       " 'streak': 4.532125974973601e-06,\n",
       " 'across': 0.0005075981091970433,\n",
       " 'sky': 0.0001404959052241816,\n",
       " 'after': 0.00020394566887381202,\n",
       " 'scan': 0.00029005606239831044,\n",
       " 'final': 0.0002583311805734952,\n",
       " 'mar': 0.0005030659832220696,\n",
       " 'clearli': 0.00016768866107402322,\n",
       " 'time': 0.0030773135370070746,\n",
       " 'twinkl': 9.064251949947201e-06,\n",
       " 'littl': 0.0006027727546714888,\n",
       " 'bit': 0.0002809918104483632,\n",
       " 'brighter': 5.4385511699683204e-05,\n",
       " 'move': 0.0007342044079457232,\n",
       " 'i': 0.0015499870834409714,\n",
       " 'thought': 0.0008293790534201689,\n",
       " 'mayb': 0.00012689952729926082,\n",
       " 'helicopt': 5.4385511699683204e-05,\n",
       " 'kept': 0.00011330314937434001,\n",
       " 'reddish': 4.532125974973601e-06,\n",
       " 'colour': 0.0005030659832220696,\n",
       " 'trail': 6.344976364963041e-05,\n",
       " 'glow': 0.00010423889742439281,\n",
       " 'behind': 0.00026739543252344243,\n",
       " 'iron': 0.00016768866107402322,\n",
       " 'collect': 0.00039882708579767683,\n",
       " 'spacedust': 4.532125974973601e-06,\n",
       " 'dure': 0.0001314316532742344,\n",
       " 'pass': 0.00034897370007296724,\n",
       " 'wild': 0.00021754204679873282,\n",
       " 'vilt': 9.064251949947201e-06,\n",
       " 'extend': 0.00020847779484878563,\n",
       " 'gel': 3.1724881824815205e-05,\n",
       " 'devic': 0.0005075981091970433,\n",
       " 'trap': 0.00014502803119915522,\n",
       " 'particl': 0.00046680897542228085,\n",
       " 'also': 0.004246602038550263,\n",
       " 'interstellar': 3.6257007799788805e-05,\n",
       " 'dust': 0.00047587322737222806,\n",
       " 'believ': 0.0011058387378935586,\n",
       " 'contain': 0.0006707546442960929,\n",
       " 'unalt': 4.532125974973601e-06,\n",
       " 'materi': 0.0007251401559957761,\n",
       " 'primit': 0.00011330314937434001,\n",
       " 'could': 0.004636364872397993,\n",
       " 'went': 0.0002311384247236536,\n",
       " 'edg': 0.0001903492909488912,\n",
       " 'far': 0.0007704614157455121,\n",
       " 'sun': 0.0007296722819707497,\n",
       " 'cold': 0.00023567055069862721,\n",
       " 'condit': 0.00079312204562038,\n",
       " 'confid': 0.000158624409124076,\n",
       " 'initi': 0.00029005606239831044,\n",
       " 'build': 0.0005710478728466736,\n",
       " 'block': 0.00018581716497391762,\n",
       " 'retriev': 9.51746454744456e-05,\n",
       " 'crew': 7.70461415745512e-05,\n",
       " 'taken': 0.00044414834554741284,\n",
       " 'clean': 0.00022207417277370642,\n",
       " 'room': 0.0001404959052241816,\n",
       " 'test': 0.0017176757445149945,\n",
       " 'train': 0.00023567055069862721,\n",
       " 'rang': 0.0005665157468717001,\n",
       " 'inspect': 5.8917637674656804e-05,\n",
       " 'schedul': 0.00010877102339936641,\n",
       " 'remov': 0.00027645968447338965,\n",
       " 'flown': 1.8128503899894402e-05,\n",
       " 'nasa': 0.0005982406286965152,\n",
       " 'center': 0.00027645968447338965,\n",
       " 'houston': 1.8128503899894402e-05,\n",
       " 'tuesday': 1.8128503899894402e-05,\n",
       " 'there': 0.0008429754313450897,\n",
       " 'catalogu': 3.1724881824815205e-05,\n",
       " 'begin': 0.00024473480264857443,\n",
       " 'distribut': 0.00018128503899894402,\n",
       " 'analysi': 0.0003806985818977824,\n",
       " 'gotcha': 9.064251949947201e-06,\n",
       " 'strike': 9.51746454744456e-05,\n",
       " 'imag': 0.0011149029898435056,\n",
       " 'spark': 4.9853385724709604e-05,\n",
       " 'interest': 0.0004124234637225976,\n",
       " 'internet': 0.00039429495982270325,\n",
       " 'user': 0.0001722207870489968,\n",
       " 'twentieth': 4.532125974973601e-06,\n",
       " 'decid': 0.0001495601571741288,\n",
       " 'whether': 0.0009925355885192185,\n",
       " 'websit': 0.00033537732214804645,\n",
       " 'visual': 0.000135963779249208,\n",
       " 'appeal': 6.344976364963041e-05,\n",
       " 'assess': 0.0002809918104483632,\n",
       " 'within': 0.0006254333845463569,\n",
       " 'millisecond': 5.8917637674656804e-05,\n",
       " 'suggest': 0.00192615353936378,\n",
       " 'web': 0.00016768866107402322,\n",
       " 'design': 0.0007795256676954592,\n",
       " 'good': 0.0007840577936704329,\n",
       " 'impress': 0.00012236740132428721,\n",
       " 'canadian': 0.00012689952729926082,\n",
       " 'report': 0.0023204484991864835,\n",
       " 'march': 8.157826754952481e-05,\n",
       " 'april': 0.0001314316532742344,\n",
       " 'behaviour': 0.0007704614157455121,\n",
       " 'inform': 0.001123967241793453,\n",
       " 'technolog': 0.002030392436788173,\n",
       " 'gitt': 4.532125974973601e-06,\n",
       " 'lindgaard': 4.532125974973601e-06,\n",
       " 'carleton': 4.532125974973601e-06,\n",
       " 'ottawa': 4.532125974973601e-06,\n",
       " 'confirm': 0.00041695558969757126,\n",
       " 'fickl': 4.532125974973601e-06,\n",
       " 'lot': 0.0006163691325964096,\n",
       " 'flash': 9.51746454744456e-05,\n",
       " 'ask': 0.0004894696052971489,\n",
       " 'particip': 0.00029005606239831044,\n",
       " 'rate': 0.0006616903923461456,\n",
       " 'when': 0.0006118370066214361,\n",
       " 'repeat': 0.00012236740132428721,\n",
       " 'exercis': 0.00010877102339936641,\n",
       " 'longer': 0.0004033592117726504,\n",
       " 'view': 0.00040789133774762405,\n",
       " 'period': 0.000589176376746568,\n",
       " 'consist': 0.00026739543252344243,\n",
       " 'medium': 5.8917637674656804e-05,\n",
       " 'messag': 0.00022660629874868003,\n",
       " 'come': 0.0010922423599686378,\n",
       " 'news': 0.00019941354289883842,\n",
       " 'anyon': 0.0001314316532742344,\n",
       " 'hope': 0.0006662225183211193,\n",
       " 'convey': 4.0789133774762405e-05,\n",
       " 'sue': 3.1724881824815205e-05,\n",
       " 'burgess': 2.2660629874868002e-05,\n",
       " 'evalu': 9.970677144941921e-05,\n",
       " 'useabl': 9.064251949947201e-06,\n",
       " 'senior': 0.0001404959052241816,\n",
       " 'lectur': 9.51746454744456e-05,\n",
       " 'doubt': 0.00012689952729926082,\n",
       " 'quickli': 0.0002130099208237592,\n",
       " 'stay': 0.0001495601571741288,\n",
       " 'usual': 0.0002855239364233368,\n",
       " 'tie': 8.61103935244984e-05,\n",
       " 'movement': 0.0002583311805734952,\n",
       " 'interact': 0.00024473480264857443,\n",
       " 'way': 0.0018128503899894402,\n",
       " 'structur': 0.0004849374793221752,\n",
       " 'unclear': 9.970677144941921e-05,\n",
       " 'chang': 0.0019578784211885954,\n",
       " 'abil': 0.00046680897542228085,\n",
       " 'adapt': 0.00019941354289883842,\n",
       " 'alway': 0.00024020267667360082,\n",
       " 'click': 4.0789133774762405e-05,\n",
       " 'around': 0.0016315653509904962,\n",
       " 'halo': 3.1724881824815205e-05,\n",
       " 'associ': 0.0011602242495932418,\n",
       " 'psycholog': 0.00029005606239831044,\n",
       " 'bill': 0.00012236740132428721,\n",
       " 'von': 9.970677144941921e-05,\n",
       " 'hippel': 4.5321259749736004e-05,\n",
       " 'south': 0.0008882966910948257,\n",
       " 'wale': 0.0002311384247236536,\n",
       " 'read': 0.0003217809442231256,\n",
       " 'word': 0.00039882708579767683,\n",
       " 'stunningli': 4.532125974973601e-06,\n",
       " 'remark': 0.00011783527534931361,\n",
       " 'timefram': 1.8128503899894402e-05,\n",
       " 'complex': 0.00043508409359746563,\n",
       " 'stimuli': 3.1724881824815205e-05,\n",
       " 'fast': 0.0002628633065484688,\n",
       " 'hold': 0.00021754204679873282,\n",
       " 'later': 0.00043055196762249205,\n",
       " 'judgement': 4.9853385724709604e-05,\n",
       " 'emot': 0.00038523070787275604,\n",
       " 'independ': 0.000135963779249208,\n",
       " 'cognit': 0.0001722207870489968,\n",
       " 'evolutionari': 0.0002311384247236536,\n",
       " 'rapidli': 0.0001404959052241816,\n",
       " 'danger': 0.00021754204679873282,\n",
       " 'situat': 0.0001948814169238648,\n",
       " 'reflect': 0.00029005606239831044,\n",
       " 'bia': 3.1724881824815205e-05,\n",
       " 'toward': 0.0003308451961730728,\n",
       " 'someth': 0.0006254333845463569,\n",
       " 'drive': 0.0003217809442231256,\n",
       " 'quick': 5.8917637674656804e-05,\n",
       " 'sort': 0.00033537732214804645,\n",
       " 'reaction': 0.00027645968447338965,\n",
       " 'consid': 0.00043508409359746563,\n",
       " 'still': 0.0008565718092700105,\n",
       " 'jewish': 2.7192755849841602e-05,\n",
       " 'histori': 0.00048040535334720164,\n",
       " 'migrat': 9.51746454744456e-05,\n",
       " 'dna': 0.0007251401559957761,\n",
       " 'four': 0.0005574514949217528,\n",
       " 'mother': 0.00018581716497391762,\n",
       " 'ago': 0.0011013066119185848,\n",
       " 'europ': 0.00033990944812302003,\n",
       " 'ancestor': 0.00025379905459852164,\n",
       " 'ashkenazi': 1.3596377924920801e-05,\n",
       " 'jew': 3.1724881824815205e-05,\n",
       " 'aliv': 4.5321259749736004e-05,\n",
       " 'intern': 0.00101972834436906,\n",
       " 'paint': 0.00032631307019809924,\n",
       " 'vivid': 1.3596377924920801e-05,\n",
       " 'pictur': 0.00040789133774762405,\n",
       " 'evolut': 0.00038523070787275604,\n",
       " 'surviv': 0.0003671022039728616,\n",
       " 'correl': 8.61103935244984e-05,\n",
       " 'well': 0.001065049604118796,\n",
       " 'establish': 0.00022207417277370642,\n",
       " 'written': 9.51746454744456e-05,\n",
       " 'oral': 3.1724881824815205e-05,\n",
       " 'doron': 9.064251949947201e-06,\n",
       " 'behar': 3.6257007799788805e-05,\n",
       " 'technion': 4.532125974973601e-06,\n",
       " 'israel': 7.251401559957761e-05,\n",
       " 'american': 0.0003081845662982048,\n",
       " 'million': 0.0016814187367152058,\n",
       " 'descend': 3.6257007799788805e-05,\n",
       " 'women': 0.0010061319664441393,\n",
       " 'for': 0.0008792324391448784,\n",
       " 'popul': 0.0006979474001459345,\n",
       " 'subject': 0.0002855239364233368,\n",
       " 'birthplac': 4.532125974973601e-06,\n",
       " 'grandmoth': 2.2660629874868002e-05,\n",
       " 'great': 0.0004260198416475184,\n",
       " 'write': 0.0006073048806464625,\n",
       " 'mitochondri': 4.9853385724709604e-05,\n",
       " 'nucleu': 0.00010877102339936641,\n",
       " 'carri': 0.0005846442507715944,\n",
       " 'instruct': 2.2660629874868002e-05,\n",
       " 'virtual': 0.00018581716497391762,\n",
       " 'unchang': 2.2660629874868002e-05,\n",
       " 'daughter': 4.5321259749736004e-05,\n",
       " 'occasion': 4.5321259749736004e-05,\n",
       " 'mutat': 0.00024020267667360082,\n",
       " 'known': 0.0011738206275181625,\n",
       " 'molecular': 0.00018581716497391762,\n",
       " 'clock': 8.61103935244984e-05,\n",
       " 'track': 0.00037163432994783525,\n",
       " 'instanc': 0.00020394566887381202,\n",
       " 'comput': 0.001123967241793453,\n",
       " 'ancestr': 4.9853385724709604e-05,\n",
       " 'eve': 9.064251949947201e-06,\n",
       " 'africa': 0.0003444415740979936,\n",
       " 'now': 0.00024473480264857443,\n",
       " 'kind': 0.0004033592117726504,\n",
       " 'pool': 7.251401559957761e-05,\n",
       " 'near': 0.0004396162195724392,\n",
       " 'east': 0.0002583311805734952,\n",
       " 'among': 0.0006073048806464625,\n",
       " 'matern': 0.00010423889742439281,\n",
       " 'lineag': 8.157826754952481e-05,\n",
       " 'real': 0.00043508409359746563,\n",
       " 'exact': 4.9853385724709604e-05,\n",
       " 'specif': 0.00047587322737222806,\n",
       " 'marker': 7.251401559957761e-05,\n",
       " 'settl': 8.61103935244984e-05,\n",
       " 'direct': 0.0003806985818977824,\n",
       " 'then': 0.0001404959052241816,\n",
       " 'certain': 0.0004260198416475184,\n",
       " 'probabl': 0.0008293790534201689,\n",
       " 'th': 0.00032631307019809924,\n",
       " 'centuri': 0.0004260198416475184,\n",
       " 'simpli': 0.00023567055069862721,\n",
       " 'demograph': 1.8128503899894402e-05,\n",
       " 'matter': 0.00043508409359746563,\n",
       " 'start': 0.000543855116996832,\n",
       " 'expand': 0.00010877102339936641,\n",
       " 'dramat': 0.00011330314937434001,\n",
       " 'tradit': 0.00026739543252344243,\n",
       " 'famili': 0.00041695558969757126,\n",
       " 'might': 0.0010922423599686378,\n",
       " 'characteris': 6.7981889624604e-05,\n",
       " 'number': 0.0011602242495932418,\n",
       " 'gave': 0.000158624409124076,\n",
       " 'rise': 0.0003897628338477296,\n",
       " 'european': 0.0003806985818977824,\n",
       " 'explod': 9.064251949947201e-05,\n",
       " 'prior': 7.70461415745512e-05,\n",
       " 'world': 0.0016814187367152058,\n",
       " 'war': 0.00020394566887381202,\n",
       " 'ii': 7.70461415745512e-05,\n",
       " 'nazi': 4.532125974973601e-06,\n",
       " 'alli': 1.3596377924920801e-05,\n",
       " 'kill': 0.0005257266130969376,\n",
       " 'estim': 0.0004260198416475184,\n",
       " 'defin': 0.000158624409124076,\n",
       " 'common': 0.000589176376746568,\n",
       " 'northern': 0.00037616645592280883,\n",
       " 'central': 0.00031271669227317844,\n",
       " 'ancestri': 4.5321259749736004e-05,\n",
       " 'cultur': 0.0004260198416475184,\n",
       " 'yiddish': 4.532125974973601e-06,\n",
       " 'languag': 0.0003308451961730728,\n",
       " 'commun': 0.0010061319664441393,\n",
       " 'pop': 4.0789133774762405e-05,\n",
       " 'non': 0.0003444415740979936,\n",
       " 'share': 0.0002583311805734952,\n",
       " 'origin': 0.0005846442507715944,\n",
       " 'particular': 0.00040789133774762405,\n",
       " 'provid': 0.0008747003131699049,\n",
       " 'explan': 0.00016315653509904962,\n",
       " 'inherit': 7.251401559957761e-05,\n",
       " 'disproportion': 4.532125974973601e-06,\n",
       " 'descent': 4.0789133774762405e-05,\n",
       " 'breast': 0.0002809918104483632,\n",
       " 'colon': 3.1724881824815205e-05,\n",
       " 'caus': 0.0016089047211156281,\n",
       " 'nuclear': 0.000543855116996832,\n",
       " 'group': 0.0015998404691656809,\n",
       " 'these': 0.0007115437780708553,\n",
       " 'gene': 0.0015590513353909185,\n",
       " 'date': 0.0006209012585713833,\n",
       " 'bottleneck': 1.8128503899894402e-05,\n",
       " 'phenomenon': 0.00019941354289883842,\n",
       " 'squeez': 5.4385511699683204e-05,\n",
       " 'larg': 0.0009336179508445617,\n",
       " 'small': 0.0010061319664441393,\n",
       " 'skorecki': 4.532125974973601e-06,\n",
       " 'founder': 3.1724881824815205e-05,\n",
       " 'huge': 0.00033537732214804645,\n",
       " 'gain': 0.0001903492909488912,\n",
       " 'insight': 7.70461415745512e-05,\n",
       " 'secret': 0.0001404959052241816,\n",
       " 'life': 0.001146627871668321,\n",
       " 'cloud': 0.0004849374793221752,\n",
       " 'biggest': 0.00016768866107402322,\n",
       " 'climat': 0.0007749935417204857,\n",
       " 'experi': 0.000988003462544245,\n",
       " 'storm': 0.0003671022039728616,\n",
       " 'chaser': 4.532125974973601e-06,\n",
       " 'gather': 0.00020394566887381202,\n",
       " 'weather': 0.00037616645592280883,\n",
       " 'balloon': 3.6257007799788805e-05,\n",
       " 'next': 0.0007749935417204857,\n",
       " 'month': 0.0007976541715953537,\n",
       " 'largest': 0.00024020267667360082,\n",
       " 'meteorolog': 0.0001314316532742344,\n",
       " 'tropic': 0.0003081845662982048,\n",
       " 'warm': 0.0005166623611469904,\n",
       " 'darwin': 5.4385511699683204e-05,\n",
       " 'comprehens': 5.4385511699683204e-05,\n",
       " 'data': 0.000838443305370116,\n",
       " 'thundercloud': 9.064251949947201e-06,\n",
       " 'peter': 0.0001903492909488912,\n",
       " 'princip': 5.4385511699683204e-05,\n",
       " 'bureau': 9.51746454744456e-05,\n",
       " 'give': 0.0006934152741709609,\n",
       " 'model': 0.0007342044079457232,\n",
       " 'accur': 0.00018581716497391762,\n",
       " 'short': 0.0003444415740979936,\n",
       " 'forecast': 6.344976364963041e-05,\n",
       " 'predict': 0.0006163691325964096,\n",
       " 'uncertainti': 5.8917637674656804e-05,\n",
       " 'greenhous': 0.00024473480264857443,\n",
       " 'calcul': 0.00024020267667360082,\n",
       " 'understand': 0.0006662225183211193,\n",
       " 'thunderstorm': 5.8917637674656804e-05,\n",
       " 'underli': 9.51746454744456e-05,\n",
       " 'aim': 0.0001540922831491024,\n",
       " 'increas': 0.0012508667690927138,\n",
       " 'evolv': 0.0004622768494473072,\n",
       " 'scale': 0.000271927558498416,\n",
       " 'impact': 0.0006299655105213304,\n",
       " 'collabor': 0.00012689952729926082,\n",
       " 'project': 0.000657158266371172,\n",
       " 'depart': 0.0002583311805734952,\n",
       " 'energi': 0.0009653428326693769,\n",
       " 'radiat': 0.00038523070787275604,\n",
       " 'measur': 0.0009789392105942977,\n",
       " 'program': 0.00043508409359746563,\n",
       " 'csiro': 0.00023567055069862721,\n",
       " 'japan': 0.0002130099208237592,\n",
       " 'ice': 0.0006707546442960929,\n",
       " 'crystal': 8.61103935244984e-05,\n",
       " 'devast': 9.064251949947201e-05,\n",
       " 'cyclon': 0.0002991203143482576,\n",
       " 'traci': 4.532125974973601e-06,\n",
       " 'watcher': 2.7192755849841602e-05,\n",
       " 'and': 0.0012916559028674762,\n",
       " 'monsoon': 1.8128503899894402e-05,\n",
       " 'season': 0.00024020267667360082,\n",
       " 'multi': 4.5321259749736004e-05,\n",
       " 'dollar': 9.970677144941921e-05,\n",
       " 'three': 0.001123967241793453,\n",
       " 'prong': 4.532125974973601e-06,\n",
       " 'plane': 0.0001495601571741288,\n",
       " 'fli': 0.0003671022039728616,\n",
       " 'directli': 0.0001540922831491024,\n",
       " 'ground': 0.00032631307019809924,\n",
       " 'network': 0.0003671022039728616,\n",
       " 'radar': 0.00010877102339936641,\n",
       " 'laser': 0.00010877102339936641,\n",
       " 'infrar': 0.00011783527534931361,\n",
       " 'characterist': 0.00016768866107402322,\n",
       " 'moistur': 4.0789133774762405e-05,\n",
       " 'light': 0.0015635834613658922,\n",
       " 'everi': 0.0004033592117726504,\n",
       " 'day': 0.0011874170054430833,\n",
       " 'five': 0.00046680897542228085,\n",
       " 'site': 0.0006662225183211193,\n",
       " 'resourc': 0.00024473480264857443,\n",
       " 'dozen': 6.344976364963041e-05,\n",
       " 'hundr': 0.00033990944812302003,\n",
       " 'paper': 0.0006934152741709609,\n",
       " 'global': 0.0005846442507715944,\n",
       " 'pattern': 0.0004849374793221752,\n",
       " 'unlik': 0.0003308451961730728,\n",
       " 'isol': 0.0001767529130239704,\n",
       " 'event': 0.0006209012585713833,\n",
       " 'hurrican': 7.251401559957761e-05,\n",
       " 'katrina': 3.6257007799788805e-05,\n",
       " 'struck': 4.5321259749736004e-05,\n",
       " 'orlean': 3.6257007799788805e-05,\n",
       " 'last': 0.0010967744859436113,\n",
       " 'question': 0.000498533857247096,\n",
       " 'due': 0.00047134110139725443,\n",
       " 'load': 0.0001314316532742344,\n",
       " 'dice': 1.8128503899894402e-05,\n",
       " 'expect': 0.0006073048806464625,\n",
       " 'intens': 0.00019941354289883842,\n",
       " 'artist': 0.000271927558498416,\n",
       " 'pluto': 0.0004396162195724392,\n",
       " 'know': 0.0008112505495202745,\n",
       " 'horizon': 9.064251949947201e-05,\n",
       " 'spacecraft': 0.00020394566887381202,\n",
       " 'launch': 0.00033990944812302003,\n",
       " 'reach': 0.00041695558969757126,\n",
       " 'distant': 0.0002311384247236536,\n",
       " 'decad': 0.0003444415740979936,\n",
       " 'agenc': 0.00043055196762249205,\n",
       " 'shed': 0.000158624409124076,\n",
       " 'wednesday': 1.8128503899894402e-05,\n",
       " 'investig': 0.00045774472347233364,\n",
       " 'class': 0.00010877102339936641,\n",
       " 'dwarf': 0.0001495601571741288,\n",
       " 'exist': 0.0006299655105213304,\n",
       " 'beyond': 0.00018581716497391762,\n",
       " 'orbit': 0.0007432686598956705,\n",
       " 'neptun': 0.0001314316532742344,\n",
       " 'over': 0.00011783527534931361,\n",
       " 'close': 0.0005982406286965152,\n",
       " 'sister': 5.4385511699683204e-05,\n",
       " 'remain': 0.0007115437780708553,\n",
       " 'shroud': 9.064251949947201e-06,\n",
       " 'mysteri': 0.00021754204679873282,\n",
       " 'fuzzi': 9.064251949947201e-06,\n",
       " 'telescop': 0.0006707546442960929,\n",
       " 'ponder': 1.3596377924920801e-05,\n",
       " 'what': 0.0005755799988216472,\n",
       " 'postag': 9.064251949947201e-06,\n",
       " 'stamp': 1.3596377924920801e-05,\n",
       " 'colleen': 9.064251949947201e-06,\n",
       " 'hartman': 4.532125974973601e-06,\n",
       " 'deputi': 3.6257007799788805e-05,\n",
       " 'administr': 0.00010423889742439281,\n",
       " 'scienc': 0.001667822358790285,\n",
       " 'alan': 6.7981889624604e-05,\n",
       " 'stern': 2.7192755849841602e-05,\n",
       " 'southwest': 3.1724881824815205e-05,\n",
       " 'boulder': 4.9853385724709604e-05,\n",
       " 'colorado': 6.344976364963041e-05,\n",
       " 'agre': 0.00033537732214804645,\n",
       " 'learn': 0.00039429495982270325,\n",
       " 'grow': 0.0005347908650468849,\n",
       " 'planethood': 4.532125974973601e-06,\n",
       " 'stop': 0.000317248818248152,\n",
       " 'growth': 0.00022207417277370642,\n",
       " 'mid': 0.0001495601571741288,\n",
       " 'gestat': 2.2660629874868002e-05,\n",
       " 'left': 0.0004849374793221752,\n",
       " 'fossil': 0.0006707546442960929,\n",
       " 'relic': 9.064251949947201e-06,\n",
       " 'kuiper': 9.970677144941921e-05,\n",
       " 'belt': 0.00018581716497391762,\n",
       " 'region': 0.0005846442507715944,\n",
       " 'planetari': 0.000158624409124076,\n",
       " 'how': 0.0001722207870489968,\n",
       " 'u': 2.2660629874868002e-05,\n",
       " 'attempt': 0.0002311384247236536,\n",
       " 'piano': 4.532125974973601e-06,\n",
       " 'instrument': 0.0001767529130239704,\n",
       " 'map': 0.0002628633065484688,\n",
       " 'moon': 0.0005619836208967265,\n",
       " 'charon': 7.251401559957761e-05,\n",
       " 'recent': 0.0009608107066944033,\n",
       " 'smaller': 0.0004033592117726504,\n",
       " 'to': 0.00040789133774762405,\n",
       " 'shave': 9.064251949947201e-06,\n",
       " 'journey': 4.5321259749736004e-05,\n",
       " 'nine': 0.00016768866107402322,\n",
       " 'aboard': 8.61103935244984e-05,\n",
       " 'massiv': 0.0003081845662982048,\n",
       " 'rocket': 3.6257007799788805e-05,\n",
       " 'typic': 0.00024020267667360082,\n",
       " 'satellit': 0.00043508409359746563,\n",
       " 'heftier': 4.532125974973601e-06,\n",
       " 'half': 0.00043055196762249205,\n",
       " 'tonn': 0.0001495601571741288,\n",
       " 'vehicl': 0.0001495601571741288,\n",
       " 'atla': 4.532125974973601e-06,\n",
       " 'upper': 6.7981889624604e-05,\n",
       " 'stage': 0.00025379905459852164,\n",
       " 'motor': 6.344976364963041e-05,\n",
       " 'unpreced': 5.8917637674656804e-05,\n",
       " 'solid': 9.970677144941921e-05,\n",
       " 'strap': 2.7192755849841602e-05,\n",
       " 'booster': 2.2660629874868002e-05,\n",
       " 'extra': 9.51746454744456e-05,\n",
       " 'punch': 1.8128503899894402e-05,\n",
       " 'success': 0.00027645968447338965,\n",
       " 'six': 0.00039429495982270325,\n",
       " 'previou': 0.00036257007799788803,\n",
       " 'roll': 4.5321259749736004e-05,\n",
       " 'hangar': 4.532125974973601e-06,\n",
       " 'cape': 1.8128503899894402e-05,\n",
       " 'canaver': 4.532125974973601e-06,\n",
       " 'pad': 1.3596377924920801e-05,\n",
       " 'earlier': 0.0003897628338477296,\n",
       " 'even': 0.0013596377924920801,\n",
       " 'boost': 0.00011783527534931361,\n",
       " 'need': 0.0015590513353909185,\n",
       " 'bounc': 5.8917637674656804e-05,\n",
       " 'jupit': 0.000158624409124076,\n",
       " 'graviti': 0.0002991203143482576,\n",
       " 'field': 0.0006390297624712777,\n",
       " 'nearli': 0.00025379905459852164,\n",
       " 'hitch': 1.8128503899894402e-05,\n",
       " 'ride': 4.9853385724709604e-05,\n",
       " 'if': 0.000997067714494192,\n",
       " 'cannot': 0.00021754204679873282,\n",
       " 'feburari': 4.532125974973601e-06,\n",
       " 'opportun': 0.000135963779249208,\n",
       " 'slingshot': 4.532125974973601e-06,\n",
       " 'addit': 0.0002583311805734952,\n",
       " 'depend': 0.00023567055069862721,\n",
       " 'actual': 0.0006163691325964096,\n",
       " 'wait': 0.0001540922831491024,\n",
       " 'approv': 0.00011330314937434001,\n",
       " 'fund': 0.00044414834554741284,\n",
       " 'happi': 0.00012689952729926082,\n",
       " 'place': 0.0009018930690197465,\n",
       " 'oper': 0.0003806985818977824,\n",
       " 'vastli': 2.2660629874868002e-05,\n",
       " 'differ': 0.0019034929094889123,\n",
       " 'temperatur': 0.0007432686598956705,\n",
       " 'familiar': 7.70461415745512e-05,\n",
       " 'co': 0.0006843510222210137,\n",
       " 'richard': 0.0001404959052241816,\n",
       " 'binzel': 9.064251949947201e-06,\n",
       " 'massachusett': 0.0001404959052241816,\n",
       " 'pois': 1.3596377924920801e-05,\n",
       " 'explor': 0.0003535058260479408,\n",
       " 'add': 0.0006209012585713833,\n",
       " 'dale': 3.1724881824815205e-05,\n",
       " 'cruikshank': 4.532125974973601e-06,\n",
       " 'ame': 1.3596377924920801e-05,\n",
       " 'california': 0.0003897628338477296,\n",
       " 'backpack': 4.9853385724709604e-05,\n",
       " 'highli': 0.00026739543252344243,\n",
       " 'mobil': 0.0002809918104483632,\n",
       " 'especi': 0.00029005606239831044,\n",
       " 'ring': 0.0002991203143482576,\n",
       " 'home': 0.00045321259749736006,\n",
       " 'cost': 0.0003444415740979936,\n",
       " 'bu': 2.7192755849841602e-05,\n",
       " 'ticket': 2.7192755849841602e-05,\n",
       " 'bondi': 4.532125974973601e-06,\n",
       " 'beach': 9.064251949947201e-05,\n",
       " 'digit': 0.00018128503899894402,\n",
       " 'buddi': 4.532125974973601e-06,\n",
       " 'act': 0.00041695558969757126,\n",
       " 'guid': 6.7981889624604e-05,\n",
       " 'jeff': 1.8128503899894402e-05,\n",
       " 'axup': 2.7192755849841602e-05,\n",
       " 'phd': 0.00020394566887381202,\n",
       " 'student': 0.00039429495982270325,\n",
       " 'queensland': 0.0003217809442231256,\n",
       " 'former': 0.00014502803119915522,\n",
       " 'focus': 0.0001404959052241816,\n",
       " 'applic': 0.00024473480264857443,\n",
       " 'shadow': 6.344976364963041e-05,\n",
       " 'document': 0.00010877102339936641,\n",
       " 'tour': 9.064251949947201e-06,\n",
       " 'brisban': 0.00010423889742439281,\n",
       " 'took': 0.0004396162195724392,\n",
       " 'boat': 1.8128503899894402e-05,\n",
       " 'cruis': 3.6257007799788805e-05,\n",
       " 'park': 0.00010423889742439281,\n",
       " 'eprintsuq': 4.532125974973601e-06,\n",
       " 'submit': 7.251401559957761e-05,\n",
       " 'public': 0.0007704614157455121,\n",
       " 'latest': 0.00045321259749736006,\n",
       " 'prototyp': 0.00012236740132428721,\n",
       " 'australasian': 4.0789133774762405e-05,\n",
       " 'crc': 3.6257007799788805e-05,\n",
       " 'product': 0.0007070116520958816,\n",
       " 'softwar': 0.0002991203143482576,\n",
       " 'download': 0.00011330314937434001,\n",
       " 'onto': 0.00024473480264857443,\n",
       " 'phone': 0.0002809918104483632,\n",
       " 'instant': 3.1724881824815205e-05,\n",
       " 'discuss': 0.00016768866107402322,\n",
       " 'forum': 3.6257007799788805e-05,\n",
       " 'wireless': 9.064251949947201e-05,\n",
       " 'symbol': 8.61103935244984e-05,\n",
       " 'increasingli': 8.157826754952481e-05,\n",
       " 'brought': 0.00012689952729926082,\n",
       " 'transport': 0.000135963779249208,\n",
       " 'breakdown': 3.1724881824815205e-05,\n",
       " 'neighbourhood': 4.0789133774762405e-05,\n",
       " 'glocalis': 4.532125974973601e-06,\n",
       " 'gone': 8.61103935244984e-05,\n",
       " 'text': 0.00012689952729926082,\n",
       " 'ani': 6.344976364963041e-05,\n",
       " 'introduc': 0.0001903492909488912,\n",
       " 'enter': 0.000158624409124076,\n",
       " 'social': 0.0005483872429718057,\n",
       " 'environ': 0.0005937085027215417,\n",
       " 'cours': 0.00016315653509904962,\n",
       " 'never': 0.00034897370007296724,\n",
       " 'watch': 0.000249266928623548,\n",
       " 'comedi': 2.7192755849841602e-05,\n",
       " 'aerob': 9.064251949947201e-06,\n",
       " 'film': 0.0001722207870489968,\n",
       " 'flow': 0.0004396162195724392,\n",
       " 'sad': 1.8128503899894402e-05,\n",
       " 'declin': 0.00023567055069862721,\n",
       " 'maryland': 4.0789133774762405e-05,\n",
       " 'movi': 0.0001540922831491024,\n",
       " 'segment': 3.6257007799788805e-05,\n",
       " 'seriou': 0.00025379905459852164,\n",
       " 'least': 0.0005574514949217528,\n",
       " 'session': 2.2660629874868002e-05,\n",
       " 'with': 0.000158624409124076,\n",
       " 'ultrasound': 9.970677144941921e-05,\n",
       " 'brachial': 4.532125974973601e-06,\n",
       " 'arteri': 1.8128503899894402e-05,\n",
       " 'vessel': 0.000135963779249208,\n",
       " 'run': 0.0004622768494473072,\n",
       " 'shoulder': 8.61103935244984e-05,\n",
       " 'elbow': 1.8128503899894402e-05,\n",
       " 'indic': 0.000498533857247096,\n",
       " 'extract': 0.00010423889742439281,\n",
       " 'distress': 5.4385511699683204e-05,\n",
       " ...}"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf_science"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "w=rural_vocab\n",
    "w.update(science_vocab)\n",
    "words=[rural_vocab,science_vocab]\n",
    "def idf():\n",
    "    idf={}\n",
    "    count=0\n",
    "    idf=dict.fromkeys(w.keys(),0)\n",
    "    for i in range(len(words)):\n",
    "        for (word,count) in words[i].items():\n",
    "            if(count>0):\n",
    "                    idf[word]+=1\n",
    "    for i,j in idf.items():\n",
    "        idf[i]=math.log(len(words)/j)\n",
    "    return(idf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "idf_data=idf()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'pm': 0.0,\n",
       " 'deni': 0.0,\n",
       " 'knowledg': 0.0,\n",
       " 'awb': 0.6931471805599453,\n",
       " 'kickback': 0.6931471805599453,\n",
       " 'the': 0.0,\n",
       " 'prime': 0.0,\n",
       " 'minist': 0.0,\n",
       " 'knew': 0.0,\n",
       " 'pay': 0.0,\n",
       " 'iraq': 0.0,\n",
       " 'despit': 0.0,\n",
       " 'write': 0.0,\n",
       " 'wheat': 0.0,\n",
       " 'export': 0.0,\n",
       " 'ask': 0.0,\n",
       " 'kept': 0.0,\n",
       " 'fulli': 0.0,\n",
       " 'inform': 0.0,\n",
       " 'sale': 0.0,\n",
       " 'letter': 0.0,\n",
       " 'john': 0.0,\n",
       " 'howard': 0.0,\n",
       " 'deputi': 0.0,\n",
       " 'mark': 0.0,\n",
       " 'vail': 0.6931471805599453,\n",
       " 'releas': 0.0,\n",
       " 'cole': 0.0,\n",
       " 'inquiri': 0.0,\n",
       " 'oil': 0.0,\n",
       " 'food': 0.0,\n",
       " 'program': 0.0,\n",
       " 'in': 0.0,\n",
       " 'one': 0.0,\n",
       " 'mr': 0.0,\n",
       " 'manag': 0.0,\n",
       " 'director': 0.0,\n",
       " 'andrew': 0.0,\n",
       " 'lindberg': 0.6931471805599453,\n",
       " 'remain': 0.0,\n",
       " 'close': 0.0,\n",
       " 'contact': 0.0,\n",
       " 'govern': 0.0,\n",
       " 'opposit': 0.0,\n",
       " 'gavan': 0.6931471805599453,\n",
       " 'o': 0.0,\n",
       " 'connor': 0.0,\n",
       " 'say': 0.0,\n",
       " 'sent': 0.0,\n",
       " 'time': 0.0,\n",
       " 'though': 0.0,\n",
       " 'jordanian': 0.6931471805599453,\n",
       " 'truck': 0.0,\n",
       " 'compani': 0.0,\n",
       " 'he': 0.0,\n",
       " 'longer': 0.0,\n",
       " 'wipe': 0.0,\n",
       " 'hand': 0.0,\n",
       " 'illicit': 0.0,\n",
       " 'payment': 0.0,\n",
       " 'total': 0.0,\n",
       " 'million': 0.0,\n",
       " 'respons': 0.0,\n",
       " 'must': 0.0,\n",
       " 'lay': 0.0,\n",
       " 'may': 0.0,\n",
       " 'squar': 0.0,\n",
       " 'feet': 0.0,\n",
       " 'coalit': 0.0,\n",
       " 'trade': 0.0,\n",
       " 'agricultur': 0.0,\n",
       " 'said': 0.0,\n",
       " 'but': 0.0,\n",
       " 'show': 0.0,\n",
       " 'inquir': 0.0,\n",
       " 'futur': 0.0,\n",
       " 'prove': 0.0,\n",
       " 'it': 0.0,\n",
       " 'would': 0.0,\n",
       " 'astonish': 0.0,\n",
       " 'i': 0.0,\n",
       " 'done': 0.0,\n",
       " 'anyth': 0.0,\n",
       " 'possibl': 0.0,\n",
       " 'could': 0.0,\n",
       " 'preserv': 0.0,\n",
       " 'australia': 0.0,\n",
       " 'valuabl': 0.0,\n",
       " 'market': 0.0,\n",
       " 'email': 0.0,\n",
       " 'question': 0.0,\n",
       " 'today': 0.0,\n",
       " 'peter': 0.0,\n",
       " 'geari': 0.6931471805599453,\n",
       " 'receiv': 0.0,\n",
       " 'indic': 0.0,\n",
       " 'iraqi': 0.0,\n",
       " 'grain': 0.0,\n",
       " 'board': 0.0,\n",
       " 'approach': 0.0,\n",
       " 'provid': 0.0,\n",
       " 'servic': 0.0,\n",
       " 'forward': 0.0,\n",
       " 'two': 0.0,\n",
       " 'colleagu': 0.0,\n",
       " 'rememb': 0.0,\n",
       " 'read': 0.0,\n",
       " 'although': 0.0,\n",
       " 'skim': 0.6931471805599453,\n",
       " 'support': 0.0,\n",
       " 'still': 0.0,\n",
       " 'plenti': 0.0,\n",
       " 'among': 0.0,\n",
       " 'grower': 0.0,\n",
       " 'central': 0.0,\n",
       " 'western': 0.0,\n",
       " 'new': 0.0,\n",
       " 'south': 0.0,\n",
       " 'wale': 0.0,\n",
       " 'revel': 0.0,\n",
       " 'produc': 0.0,\n",
       " 'broadli': 0.0,\n",
       " 'attempt': 0.0,\n",
       " 'get': 0.0,\n",
       " 'best': 0.0,\n",
       " 'price': 0.0,\n",
       " 'product': 0.0,\n",
       " 'think': 0.0,\n",
       " 'ploy': 0.6931471805599453,\n",
       " 'oversea': 0.0,\n",
       " 'interest': 0.0,\n",
       " 'tri': 0.0,\n",
       " 'singl': 0.0,\n",
       " 'desk': 0.0,\n",
       " 'put': 0.0,\n",
       " 'asid': 0.0,\n",
       " 'stori': 0.0,\n",
       " 'go': 0.0,\n",
       " 'round': 0.0,\n",
       " 'commiss': 0.0,\n",
       " 'everyth': 0.0,\n",
       " 'way': 0.0,\n",
       " 'peopl': 0.0,\n",
       " 'got': 0.0,\n",
       " 'thing': 0.0,\n",
       " 'busi': 0.0,\n",
       " 'middl': 0.0,\n",
       " 'east': 0.0,\n",
       " 'asian': 0.0,\n",
       " 'countri': 0.0,\n",
       " 'actual': 0.0,\n",
       " 'pretti': 0.0,\n",
       " 'reason': 0.0,\n",
       " 'system': 0.0,\n",
       " 'give': 0.0,\n",
       " 'fair': 0.0,\n",
       " 'moment': 0.0,\n",
       " 'averag': 0.0,\n",
       " 'perform': 0.0,\n",
       " 'fairli': 0.0,\n",
       " 'well': 0.0,\n",
       " 'anoth': 0.0,\n",
       " 'biggest': 0.0,\n",
       " 'someon': 0.0,\n",
       " 'els': 0.0,\n",
       " 'take': 0.0,\n",
       " 'whether': 0.0,\n",
       " 'multin': 0.6931471805599453,\n",
       " 'much': 0.0,\n",
       " 'foothold': 0.0,\n",
       " 'advantag': 0.0,\n",
       " 'analyst': 0.0,\n",
       " 'predict': 0.0,\n",
       " 'drop': 0.0,\n",
       " 'tonn': 0.0,\n",
       " 'back': 0.0,\n",
       " 'malcolm': 0.0,\n",
       " 'bartholomaeu': 0.6931471805599453,\n",
       " 'pool': 0.0,\n",
       " 'return': 0.0,\n",
       " 'alreadi': 0.0,\n",
       " 'year': 0.0,\n",
       " 'past': 0.0,\n",
       " 'five': 0.0,\n",
       " 'premium': 0.6931471805599453,\n",
       " 'achiev': 0.0,\n",
       " 'monopoli': 0.6931471805599453,\n",
       " 'sever': 0.0,\n",
       " 'erod': 0.0,\n",
       " 'sa': 0.0,\n",
       " 'farmer': 0.0,\n",
       " 'help': 0.0,\n",
       " 'fire': 0.0,\n",
       " 'ravag': 0.0,\n",
       " 'neighbour': 0.0,\n",
       " 'donat': 0.0,\n",
       " 'truckload': 0.6931471805599453,\n",
       " 'hay': 0.0,\n",
       " 'across': 0.0,\n",
       " 'border': 0.0,\n",
       " 'wake': 0.0,\n",
       " 'grampian': 0.6931471805599453,\n",
       " 'bushfir': 0.0,\n",
       " 'day': 0.0,\n",
       " 'agist': 0.6931471805599453,\n",
       " 'cattl': 0.0,\n",
       " 'they': 0.0,\n",
       " 'begin': 0.0,\n",
       " 'fodder': 0.6931471805599453,\n",
       " 'drive': 0.0,\n",
       " 'coordin': 0.0,\n",
       " 'conner': 0.6931471805599453,\n",
       " 'overwhelm': 0.0,\n",
       " 'all': 0.0,\n",
       " 'week': 0.0,\n",
       " 'gone': 0.0,\n",
       " 'place': 0.0,\n",
       " 'load': 0.0,\n",
       " 'we': 0.0,\n",
       " 'man': 0.0,\n",
       " 'full': 0.0,\n",
       " 'rest': 0.0,\n",
       " 'straight': 0.0,\n",
       " 'move': 0.0,\n",
       " 'floodwat': 0.0,\n",
       " 'highway': 0.0,\n",
       " 'a': 0.0,\n",
       " 'major': 0.0,\n",
       " 'northern': 0.0,\n",
       " 'territori': 0.0,\n",
       " 'block': 0.0,\n",
       " 'victoria': 0.0,\n",
       " 'river': 0.0,\n",
       " 'cut': 0.0,\n",
       " 'also': 0.0,\n",
       " 'flood': 0.0,\n",
       " 'remot': 0.0,\n",
       " 'pigeon': 0.0,\n",
       " 'hole': 0.0,\n",
       " 'aborigin': 0.0,\n",
       " 'commun': 0.0,\n",
       " 'stockman': 0.6931471805599453,\n",
       " 'simon': 0.0,\n",
       " 'ham': 0.0,\n",
       " 'describ': 0.0,\n",
       " 'reloc': 0.0,\n",
       " 'hundr': 0.0,\n",
       " 'higher': 0.0,\n",
       " 'ground': 0.0,\n",
       " 'vehicl': 0.0,\n",
       " 'camp': 0.0,\n",
       " 'set': 0.0,\n",
       " 'belong': 0.0,\n",
       " 'coupl': 0.0,\n",
       " 'boat': 0.0,\n",
       " 'onto': 0.0,\n",
       " 'metr': 0.0,\n",
       " 'sandston': 0.0,\n",
       " 'ridg': 0.0,\n",
       " 'height': 0.0,\n",
       " 'might': 0.0,\n",
       " 'littl': 0.0,\n",
       " 'start': 0.0,\n",
       " 'sold': 0.0,\n",
       " 'm': 0.0,\n",
       " 'tasmania': 0.0,\n",
       " 'main': 0.0,\n",
       " 'chang': 0.0,\n",
       " 'second': 0.0,\n",
       " 'three': 0.0,\n",
       " 'former': 0.0,\n",
       " 'state': 0.0,\n",
       " 'own': 0.0,\n",
       " 'tasmanian': 0.0,\n",
       " 'elev': 0.0,\n",
       " 'local': 0.0,\n",
       " 'agribusi': 0.0,\n",
       " 'robert': 0.0,\n",
       " 'limit': 0.0,\n",
       " 'deal': 0.0,\n",
       " 'includ': 0.0,\n",
       " 'silo': 0.6931471805599453,\n",
       " 'launceston': 0.0,\n",
       " 'devonport': 0.6931471805599453,\n",
       " 'powranna': 0.6931471805599453,\n",
       " 'macleod': 0.6931471805599453,\n",
       " 'bid': 0.0,\n",
       " 'first': 0.0,\n",
       " 'privatis': 0.6931471805599453,\n",
       " 'disappoint': 0.0,\n",
       " 'success': 0.0,\n",
       " 'point': 0.0,\n",
       " 'wine': 0.0,\n",
       " 'worker': 0.0,\n",
       " 'stop': 0.0,\n",
       " 'work': 0.0,\n",
       " 'hardi': 0.0,\n",
       " 'stanley': 0.0,\n",
       " 'wineri': 0.6931471805599453,\n",
       " 'walk': 0.0,\n",
       " 'job': 0.0,\n",
       " 'forti': 0.0,\n",
       " 'staff': 0.0,\n",
       " 'morn': 0.0,\n",
       " 'disput': 0.0,\n",
       " 'enterpris': 0.0,\n",
       " 'bargain': 0.0,\n",
       " 'agreement': 0.0,\n",
       " 'stopwork': 0.6931471805599453,\n",
       " 'come': 0.0,\n",
       " 'region': 0.0,\n",
       " 'grape': 0.0,\n",
       " 'crush': 0.0,\n",
       " 'took': 0.0,\n",
       " 'matter': 0.0,\n",
       " 'industri': 0.0,\n",
       " 'relat': 0.0,\n",
       " 'irc': 0.6931471805599453,\n",
       " 'friday': 0.0,\n",
       " 'wool': 0.0,\n",
       " 'bodi': 0.0,\n",
       " 'eye': 0.0,\n",
       " 'sportswear': 0.0,\n",
       " 'billion': 0.0,\n",
       " 'global': 0.0,\n",
       " 'target': 0.0,\n",
       " 'promot': 0.0,\n",
       " 'australian': 0.0,\n",
       " 'innov': 0.0,\n",
       " 'awi': 0.6931471805599453,\n",
       " 'blend': 0.0,\n",
       " 'shirt': 0.0,\n",
       " 'casual': 0.6931471805599453,\n",
       " 'wear': 0.0,\n",
       " 'manufactur': 0.0,\n",
       " 'largest': 0.0,\n",
       " 'held': 0.0,\n",
       " 'germani': 0.0,\n",
       " 'len': 0.0,\n",
       " 'stephen': 0.0,\n",
       " 'pricier': 0.6931471805599453,\n",
       " 'end': 0.0,\n",
       " 'shopper': 0.0,\n",
       " 'will': 0.0,\n",
       " 'sport': 0.0,\n",
       " 'apparel': 0.6931471805599453,\n",
       " 'sector': 0.0,\n",
       " 'certainli': 0.0,\n",
       " 'fastest': 0.0,\n",
       " 'grow': 0.0,\n",
       " 'world': 0.0,\n",
       " 'secret': 0.0,\n",
       " 'big': 0.0,\n",
       " 'share': 0.0,\n",
       " 'level': 0.0,\n",
       " 'particularli': 0.0,\n",
       " 'merino': 0.0,\n",
       " 'realli': 0.0,\n",
       " 'almost': 0.0,\n",
       " 'radar': 0.0,\n",
       " 'organis': 0.0,\n",
       " 'step': 0.0,\n",
       " 'ban': 0.0,\n",
       " 'rodeo': 0.6931471805599453,\n",
       " 'campaign': 0.0,\n",
       " 'sustain': 0.0,\n",
       " 'attack': 0.0,\n",
       " 'anim': 0.0,\n",
       " 'right': 0.0,\n",
       " 'activist': 0.0,\n",
       " 'destroy': 0.0,\n",
       " 'weekend': 0.0,\n",
       " 'event': 0.0,\n",
       " 'buck': 0.0,\n",
       " 'hors': 0.0,\n",
       " 'break': 0.0,\n",
       " 'leg': 0.0,\n",
       " 'ulverston': 0.6931471805599453,\n",
       " 'saturday': 0.0,\n",
       " 'ago': 0.0,\n",
       " 'bull': 0.0,\n",
       " 'appar': 0.0,\n",
       " 'ride': 0.0,\n",
       " 'competit': 0.0,\n",
       " 'carrick': 0.6931471805599453,\n",
       " 'owner': 0.0,\n",
       " 'brian': 0.0,\n",
       " 'fish': 0.0,\n",
       " 'unfortun': 0.0,\n",
       " 'accid': 0.0,\n",
       " 'welfar': 0.0,\n",
       " 'issu': 0.0,\n",
       " 'emma': 0.0,\n",
       " 'haswel': 0.0,\n",
       " 'against': 0.6931471805599453,\n",
       " 'cruelti': 0.6931471805599453,\n",
       " 'effort': 0.0,\n",
       " 'snail': 0.0,\n",
       " 'use': 0.0,\n",
       " 'cancer': 0.0,\n",
       " 'research': 0.0,\n",
       " 'investig': 0.0,\n",
       " 'sea': 0.0,\n",
       " 'eventu': 0.0,\n",
       " 'treat': 0.0,\n",
       " 'until': 0.0,\n",
       " 'popul': 0.0,\n",
       " 'virtual': 0.0,\n",
       " 'unexplor': 0.0,\n",
       " 'flinder': 0.0,\n",
       " 'univers': 0.0,\n",
       " 'hope': 0.0,\n",
       " 'discov': 0.0,\n",
       " 'benefici': 0.0,\n",
       " 'compound': 0.0,\n",
       " 'predatori': 0.0,\n",
       " 'known': 0.0,\n",
       " 'whelk': 0.6931471805599453,\n",
       " 'marin': 0.0,\n",
       " 'biologist': 0.0,\n",
       " 'dr': 0.0,\n",
       " 'kirsten': 0.6931471805599453,\n",
       " 'benkendorff': 0.6931471805599453,\n",
       " 'follow': 0.0,\n",
       " 'clinic': 0.0,\n",
       " 'trial': 0.0,\n",
       " 'anti': 0.0,\n",
       " 'properti': 0.0,\n",
       " 'look': 0.0,\n",
       " 'current': 0.0,\n",
       " 'harvest': 0.0,\n",
       " 'consid': 0.0,\n",
       " 'resourc': 0.0,\n",
       " 'lot': 0.0,\n",
       " 'potenti': 0.0,\n",
       " 'econom': 0.0,\n",
       " 'benefit': 0.0,\n",
       " 'side': 0.0,\n",
       " 'processor': 0.0,\n",
       " 'fail': 0.0,\n",
       " 'meet': 0.0,\n",
       " 'kangaroo': 0.0,\n",
       " 'meat': 0.0,\n",
       " 'demand': 0.0,\n",
       " 'intern': 0.0,\n",
       " 'pressur': 0.0,\n",
       " 'high': 0.0,\n",
       " 'quota': 0.6931471805599453,\n",
       " 'met': 0.0,\n",
       " 'leav': 0.0,\n",
       " 'gap': 0.0,\n",
       " 'phil': 0.0,\n",
       " 'franolick': 0.6931471805599453,\n",
       " 'king': 0.0,\n",
       " 'perth': 0.0,\n",
       " 'european': 0.0,\n",
       " 'attract': 0.0,\n",
       " 'low': 0.0,\n",
       " 'fat': 0.0,\n",
       " 'content': 0.0,\n",
       " 'russian': 0.0,\n",
       " 'consum': 0.0,\n",
       " 'larg': 0.0,\n",
       " 'amount': 0.0,\n",
       " 'sausag': 0.6931471805599453,\n",
       " 'minc': 0.6931471805599453,\n",
       " 'as': 0.0,\n",
       " 'far': 0.0,\n",
       " 'know': 0.0,\n",
       " 'everi': 0.0,\n",
       " 'skerrick': 0.6931471805599453,\n",
       " 'human': 0.0,\n",
       " 'consumpt': 0.0,\n",
       " 'goe': 0.0,\n",
       " 'warn': 0.0,\n",
       " 'snake': 0.0,\n",
       " 'alert': 0.0,\n",
       " 'with': 0.0,\n",
       " 'temperatur': 0.0,\n",
       " 'soar': 0.0,\n",
       " 'summer': 0.0,\n",
       " 'southern': 0.0,\n",
       " 'keep': 0.0,\n",
       " 'open': 0.0,\n",
       " 'paddock': 0.6931471805599453,\n",
       " 'especi': 0.0,\n",
       " 'area': 0.0,\n",
       " 'hit': 0.0,\n",
       " 'forc': 0.0,\n",
       " 'usual': 0.0,\n",
       " 'habitat': 0.0,\n",
       " 'maria': 0.0,\n",
       " 'dovey': 0.6931471805599453,\n",
       " 'reptil': 0.0,\n",
       " 'rescu': 0.0,\n",
       " 'word': 0.0,\n",
       " 'advic': 0.0,\n",
       " 'ani': 0.0,\n",
       " 'regularli': 0.0,\n",
       " 'outdoor': 0.0,\n",
       " 'bush': 0.0,\n",
       " 'land': 0.0,\n",
       " 'alway': 0.0,\n",
       " 'carri': 0.0,\n",
       " 'least': 0.0,\n",
       " 'bandag': 0.6931471805599453,\n",
       " 'mobil': 0.0,\n",
       " 'phone': 0.0,\n",
       " 'aerial': 0.0,\n",
       " 'spray': 0.0,\n",
       " 'control': 0.0,\n",
       " 'locust': 0.0,\n",
       " 'threat': 0.0,\n",
       " 'begun': 0.0,\n",
       " 'plagu': 0.0,\n",
       " 'made': 0.0,\n",
       " 'continu': 0.0,\n",
       " 'rain': 0.0,\n",
       " 'warm': 0.0,\n",
       " 'weather': 0.0,\n",
       " 'assist': 0.0,\n",
       " 'breed': 0.0,\n",
       " 'condit': 0.0,\n",
       " 'pastur': 0.0,\n",
       " 'crop': 0.0,\n",
       " 'pest': 0.0,\n",
       " 'injuri': 0.0,\n",
       " 'baton': 0.0,\n",
       " 'runner': 0.0,\n",
       " 'queensland': 0.0,\n",
       " 'knee': 0.0,\n",
       " 'run': 0.0,\n",
       " 'queen': 0.0,\n",
       " 'relay': 0.0,\n",
       " 'lead': 0.0,\n",
       " 'commonwealth': 0.0,\n",
       " 'game': 0.0,\n",
       " 'stanthorp': 0.6931471805599453,\n",
       " 'bill': 0.0,\n",
       " 'bonner': 0.6931471805599453,\n",
       " 'travel': 0.0,\n",
       " 'coff': 0.6931471805599453,\n",
       " 'harbour': 0.0,\n",
       " 'and': 0.0,\n",
       " 'prevent': 0.0,\n",
       " 'make': 0.0,\n",
       " 'dash': 0.6931471805599453,\n",
       " 'about': 0.0,\n",
       " 'jump': 0.0,\n",
       " 'motorbik': 0.0,\n",
       " 'twist': 0.0,\n",
       " 'ligament': 0.6931471805599453,\n",
       " 'abl': 0.0,\n",
       " 'sinc': 0.0,\n",
       " 'promis': 0.0,\n",
       " 'adrenalin': 0.0,\n",
       " 'rush': 0.0,\n",
       " 'nz': 0.6931471805599453,\n",
       " 'appl': 0.0,\n",
       " 'debat': 0.0,\n",
       " 'zealand': 0.0,\n",
       " 'newspap': 0.0,\n",
       " 'member': 0.0,\n",
       " 'voic': 0.0,\n",
       " 'concern': 0.0,\n",
       " 'regard': 0.0,\n",
       " 'legitimaci': 0.6931471805599453,\n",
       " 'scienc': 0.0,\n",
       " 'draft': 0.0,\n",
       " 'import': 0.0,\n",
       " 'risk': 0.0,\n",
       " 'analysi': 0.0,\n",
       " 'last': 0.0,\n",
       " 'month': 0.0,\n",
       " 'biosecur': 0.0,\n",
       " 'ira': 0.6931471805599453,\n",
       " 'determin': 0.0,\n",
       " 'fireblight': 0.6931471805599453,\n",
       " 'diseas': 0.0,\n",
       " 'under': 0.0,\n",
       " 'rule': 0.0,\n",
       " 'refus': 0.0,\n",
       " 'scientif': 0.0,\n",
       " 'evid': 0.0,\n",
       " 'substanti': 0.0,\n",
       " 'accord': 0.0,\n",
       " 'profession': 0.0,\n",
       " 'matur': 0.0,\n",
       " 'pose': 0.0,\n",
       " 'howev': 0.0,\n",
       " 'spokesperson': 0.0,\n",
       " 'base': 0.0,\n",
       " 'access': 0.0,\n",
       " 'group': 0.0,\n",
       " 'allison': 0.6931471805599453,\n",
       " 'mathemat': 0.0,\n",
       " 'model': 0.0,\n",
       " 'formula': 0.0,\n",
       " 'assess': 0.0,\n",
       " 'report': 0.0,\n",
       " 'publish': 0.0,\n",
       " 'decemb': 0.0,\n",
       " 'anyon': 0.0,\n",
       " 'compet': 0.0,\n",
       " 'reproduc': 0.0,\n",
       " 'given': 0.0,\n",
       " 'chairman': 0.0,\n",
       " 'task': 0.0,\n",
       " 'corboy': 0.6931471805599453,\n",
       " 'surpris': 0.0,\n",
       " 'exercis': 0.0,\n",
       " 'present': 0.0,\n",
       " 'younger': 0.0,\n",
       " 'tree': 0.0,\n",
       " 'pear': 0.6931471805599453,\n",
       " 'plot': 0.0,\n",
       " 'young': 0.0,\n",
       " 'william': 0.0,\n",
       " 'fruit': 0.0,\n",
       " 'plant': 0.0,\n",
       " 'six': 0.0,\n",
       " 'old': 0.0,\n",
       " 'part': 0.0,\n",
       " 'experiment': 0.0,\n",
       " 'project': 0.0,\n",
       " 'fund': 0.0,\n",
       " 'spc': 0.6931471805599453,\n",
       " 'ardmona': 0.6931471805599453,\n",
       " 'can': 0.0,\n",
       " 'associ': 0.0,\n",
       " 'horticultur': 0.6931471805599453,\n",
       " 'lenn': 0.6931471805599453,\n",
       " 'famili': 0.0,\n",
       " 'orchard': 0.6931471805599453,\n",
       " 'locat': 0.0,\n",
       " 'consult': 0.0,\n",
       " 'ba': 0.0,\n",
       " 'van': 0.0,\n",
       " 'den': 0.0,\n",
       " 'aim': 0.0,\n",
       " 'increas': 0.0,\n",
       " 'number': 0.0,\n",
       " 'grown': 0.0,\n",
       " 'encourag': 0.0,\n",
       " 'earlier': 0.0,\n",
       " 'feder': 0.0,\n",
       " 'nation': 0.0,\n",
       " 'mp': 0.0,\n",
       " 'park': 0.0,\n",
       " 'cobb': 0.6931471805599453,\n",
       " 'brought': 0.0,\n",
       " 'thi': 0.0,\n",
       " 'examin': 0.0,\n",
       " 'role': 0.0,\n",
       " 'on': 0.0,\n",
       " 'top': 0.0,\n",
       " 'parti': 0.0,\n",
       " 'reinvent': 0.6931471805599453,\n",
       " 'becom': 0.0,\n",
       " 'pro': 0.0,\n",
       " 'activ': 0.0,\n",
       " 'differ': 0.0,\n",
       " 'liber': 0.0,\n",
       " 'repres': 0.0,\n",
       " 'better': 0.0,\n",
       " 'death': 0.0,\n",
       " 'believ': 0.0,\n",
       " 'due': 0.0,\n",
       " 'lack': 0.0,\n",
       " 'oxygen': 0.0,\n",
       " 'depart': 0.0,\n",
       " 'primari': 0.0,\n",
       " 'kill': 0.0,\n",
       " 'lachlan': 0.6931471805599453,\n",
       " 'upstream': 0.0,\n",
       " 'lake': 0.0,\n",
       " 'cargelligo': 0.6931471805599453,\n",
       " 'weir': 0.6931471805599453,\n",
       " 'between': 0.0,\n",
       " 'nativ': 0.0,\n",
       " 'die': 0.0,\n",
       " 'incid': 0.0,\n",
       " 'district': 0.0,\n",
       " 'fisheri': 0.0,\n",
       " 'offic': 0.0,\n",
       " 'phillip': 0.0,\n",
       " 'mccarthi': 0.6931471805599453,\n",
       " 'murray': 0.0,\n",
       " 'cod': 0.0,\n",
       " 'bream': 0.6931471805599453,\n",
       " 'golden': 0.0,\n",
       " 'silver': 0.0,\n",
       " 'perch': 0.0,\n",
       " 'introduc': 0.0,\n",
       " 'speci': 0.0,\n",
       " 'suggest': 0.0,\n",
       " 'result': 0.0,\n",
       " 'water': 0.0,\n",
       " 'flow': 0.0,\n",
       " 'chemic': 0.0,\n",
       " 'poison': 0.0,\n",
       " 'veget': 0.0,\n",
       " 'label': 0.0,\n",
       " 'ausveg': 0.6931471805599453,\n",
       " 'law': 0.0,\n",
       " 'joke': 0.0,\n",
       " 'nobodi': 0.0,\n",
       " 'enforc': 0.0,\n",
       " 'non': 0.0,\n",
       " 'exist': 0.0,\n",
       " 'fool': 0.0,\n",
       " 'buy': 0.0,\n",
       " 'fals': 0.0,\n",
       " 'poor': 0.0,\n",
       " 'red': 0.0,\n",
       " 'tape': 0.0,\n",
       " 'slow': 0.0,\n",
       " 'west': 0.0,\n",
       " 'kimberley': 0.0,\n",
       " 'wa': 0.0,\n",
       " 'bowl': 0.0,\n",
       " 'soil': 0.0,\n",
       " 'avail': 0.0,\n",
       " 'sunshin': 0.0,\n",
       " 'face': 0.0,\n",
       " 'hurdl': 0.0,\n",
       " 'expand': 0.0,\n",
       " 'sweet': 0.0,\n",
       " 'potato': 0.0,\n",
       " 'paul': 0.0,\n",
       " 'higgin': 0.0,\n",
       " 'reli': 0.0,\n",
       " 'gener': 0.0,\n",
       " 'broom': 0.0,\n",
       " 'horticulturalist': 0.6931471805599453,\n",
       " 'connect': 0.0,\n",
       " 'grid': 0.0,\n",
       " 'appli': 0.0,\n",
       " 'nine': 0.0,\n",
       " 'kilometr': 0.0,\n",
       " 'extens': 0.0,\n",
       " 'bring': 0.0,\n",
       " 'line': 0.0,\n",
       " 'power': 0.0,\n",
       " 'custom': 0.0,\n",
       " 'cost': 0.0,\n",
       " 'regardless': 0.0,\n",
       " 'distanc': 0.0,\n",
       " 'acquisit': 0.0,\n",
       " 'caus': 0.0,\n",
       " 'headach': 0.0,\n",
       " 'purchas': 0.0,\n",
       " 'adjoin': 0.6931471805599453,\n",
       " 'june': 0.0,\n",
       " 'titl': 0.0,\n",
       " 'improv': 0.0,\n",
       " 'clear': 0.0,\n",
       " 'resolv': 0.0,\n",
       " 'survey': 0.0,\n",
       " 'bed': 0.0,\n",
       " 'moni': 0.6931471805599453,\n",
       " 'paid': 0.0,\n",
       " 'plan': 0.0,\n",
       " 'infrastructur': 0.0,\n",
       " 'alannah': 0.6931471805599453,\n",
       " 'mctiernan': 0.6931471805599453,\n",
       " 'seven': 0.0,\n",
       " 'delay': 0.0,\n",
       " 'regrett': 0.6931471805599453,\n",
       " 'awar': 0.0,\n",
       " 'hold': 0.0,\n",
       " 'correctli': 0.0,\n",
       " 'zone': 0.0,\n",
       " 'ms': 0.0,\n",
       " 'favour': 0.0,\n",
       " 'wait': 0.0,\n",
       " 'offici': 0.0,\n",
       " 'plough': 0.0,\n",
       " 'pindan': 0.6931471805599453,\n",
       " 'richmond': 0.0,\n",
       " 'stay': 0.0,\n",
       " 'fisher': 0.0,\n",
       " 'decid': 0.0,\n",
       " 'commerci': 0.0,\n",
       " 'recreat': 0.0,\n",
       " 'dead': 0.0,\n",
       " 'eel': 0.0,\n",
       " 'near': 0.0,\n",
       " 'wardel': 0.6931471805599453,\n",
       " 'thursday': 0.0,\n",
       " 'alarmingli': 0.6931471805599453,\n",
       " 'act': 0.0,\n",
       " 'anthoni': 0.0,\n",
       " 'hirst': 0.6931471805599453,\n",
       " 'inspector': 0.0,\n",
       " 'sampl': 0.0,\n",
       " 'tomorrow': 0.0,\n",
       " 'rural': 0.0,\n",
       " 'woman': 0.0,\n",
       " 'finalist': 0.6931471805599453,\n",
       " 'next': 0.0,\n",
       " 'seem': 0.0,\n",
       " 'like': 0.0,\n",
       " 'etern': 0.0,\n",
       " 'rosebank': 0.6931471805599453,\n",
       " 'catherin': 0.0,\n",
       " 'ford': 0.0,\n",
       " 'quest': 0.0,\n",
       " 'find': 0.0,\n",
       " 'macadamia': 0.0,\n",
       " 'win': 0.0,\n",
       " 'guid': 0.0,\n",
       " 'admit': 0.0,\n",
       " 'amaz': 0.0,\n",
       " 'even': 0.0,\n",
       " 'dse': 0.6931471805599453,\n",
       " 'proactiv': 0.6931471805599453,\n",
       " 'burn': 0.0,\n",
       " 'yesterday': 0.0,\n",
       " 'heard': 0.0,\n",
       " 'cfa': 0.6931471805599453,\n",
       " 'feel': 0.0,\n",
       " 'intens': 0.0,\n",
       " 'lessen': 0.0,\n",
       " 'fuel': 0.0,\n",
       " 'reduct': 0.0,\n",
       " 'charley': 0.6931471805599453,\n",
       " 'defegley': 0.6931471805599453,\n",
       " 'ararat': 0.6931471805599453,\n",
       " 'vff': 0.6931471805599453,\n",
       " 'branch': 0.0,\n",
       " 'secretari': 0.0,\n",
       " 'division': 0.6931471805599453,\n",
       " 'command': 0.0,\n",
       " 'felt': 0.0,\n",
       " 'enough': 0.0,\n",
       " 'contribut': 0.0,\n",
       " 'destruct': 0.0,\n",
       " 'horsham': 0.6931471805599453,\n",
       " 'geoff': 0.0,\n",
       " 'evan': 0.0,\n",
       " 'agenc': 0.0,\n",
       " 'polici': 0.0,\n",
       " 'hamper': 0.0,\n",
       " 'landown': 0.0,\n",
       " 'keen': 0.0,\n",
       " 'besid': 0.0,\n",
       " 'allow': 0.0,\n",
       " 'small': 0.0,\n",
       " 'window': 0.0,\n",
       " 'opportun': 0.0,\n",
       " 'strong': 0.0,\n",
       " 'divis': 0.0,\n",
       " 'within': 0.0,\n",
       " 'less': 0.0,\n",
       " 'hectar': 0.0,\n",
       " 'burnt': 0.0,\n",
       " 'ag': 0.0,\n",
       " 'speedi': 0.0,\n",
       " 'aid': 0.0,\n",
       " 'announc': 0.0,\n",
       " 'commit': 0.0,\n",
       " 'dollar': 0.0,\n",
       " 'recoveri': 0.0,\n",
       " 'blaze': 0.0,\n",
       " 'ministeri': 0.0,\n",
       " 'need': 0.0,\n",
       " 'hardship': 0.6931471805599453,\n",
       " 'suppli': 0.0,\n",
       " 'firefight': 0.0,\n",
       " 'victorian': 0.0,\n",
       " 'haulag': 0.6931471805599453,\n",
       " 'contractor': 0.0,\n",
       " 'grassroot': 0.6931471805599453,\n",
       " 'affect': 0.0,\n",
       " 'semi': 0.0,\n",
       " 'collect': 0.0,\n",
       " 'moyston': 0.6931471805599453,\n",
       " 'warrnambool': 0.6931471805599453,\n",
       " 'surviv': 0.0,\n",
       " 'bruce': 0.0,\n",
       " 'mckay': 0.6931471805599453,\n",
       " 'farm': 0.0,\n",
       " 'sheep': 0.0,\n",
       " 'fix': 0.0,\n",
       " 'fenc': 0.0,\n",
       " 'willaura': 0.6931471805599453,\n",
       " 'tim': 0.0,\n",
       " 'wrench': 0.6931471805599453,\n",
       " 'coop': 0.6931471805599453,\n",
       " 'came': 0.0,\n",
       " 'around': 0.0,\n",
       " 'haulier': 0.6931471805599453,\n",
       " 'scott': 0.0,\n",
       " 'wilson': 0.0,\n",
       " 'koroit': 0.6931471805599453,\n",
       " 'went': 0.0,\n",
       " 'neil': 0.0,\n",
       " 'jame': 0.0,\n",
       " 'dpi': 0.6931471805599453,\n",
       " 'charli': 0.6931471805599453,\n",
       " 'defegeley': 0.6931471805599453,\n",
       " 'blame': 0.0,\n",
       " 'hike': 0.0,\n",
       " 'fresh': 0.0,\n",
       " 'fluctuat': 0.0,\n",
       " 'per': 0.0,\n",
       " 'cent': 0.0,\n",
       " 'latest': 0.0,\n",
       " 'rabobank': 0.6931471805599453,\n",
       " 'milk': 0.0,\n",
       " 'ben': 0.0,\n",
       " 'russel': 0.0,\n",
       " 'money': 0.0,\n",
       " 'case': 0.0,\n",
       " 'pass': 0.0,\n",
       " 'contrari': 0.0,\n",
       " 'see': 0.0,\n",
       " 'retail': 0.0,\n",
       " 'fall': 0.0,\n",
       " 'tradit': 0.0,\n",
       " 'egypt': 0.0,\n",
       " 'pakistan': 0.0,\n",
       " 'china': 0.0,\n",
       " 'india': 0.0,\n",
       " 'deliv': 0.0,\n",
       " 'expect': 0.0,\n",
       " 'contract': 0.0,\n",
       " 'record': 0.0,\n",
       " 'tender': 0.0,\n",
       " 'fierc': 0.0,\n",
       " 'trader': 0.0,\n",
       " 'lloyd': 0.0,\n",
       " 'georg': 0.0,\n",
       " 'rang': 0.0,\n",
       " 'factor': 0.0,\n",
       " 'period': 0.0,\n",
       " 'signific': 0.0,\n",
       " 'tonnag': 0.6931471805599453,\n",
       " 'recent': 0.0,\n",
       " 'quantiti': 0.0,\n",
       " 'whole': 0.0,\n",
       " 'bunch': 0.0,\n",
       " 'broader': 0.0,\n",
       " 'wherebi': 0.0,\n",
       " 'sell': 0.0,\n",
       " 'quiet': 0.0,\n",
       " 'david': 0.0,\n",
       " 'johnson': 0.0,\n",
       " 'impact': 0.0,\n",
       " 'lost': 0.0,\n",
       " 'submit': 0.0,\n",
       " 'particip': 0.0,\n",
       " 'absolut': 0.0,\n",
       " 'probabl': 0.0,\n",
       " 'rel': 0.0,\n",
       " 'compar': 0.0,\n",
       " 'structur': 0.0,\n",
       " 'previou': 0.0,\n",
       " 'fact': 0.0,\n",
       " 'extrem': 0.0,\n",
       " 'comment': 0.0,\n",
       " 'season': 0.0,\n",
       " 'estim': 0.0,\n",
       " 'lift': 0.0,\n",
       " 'hard': 0.0,\n",
       " 'noodl': 0.6931471805599453,\n",
       " 'grade': 0.0,\n",
       " 'there': 0.0,\n",
       " 'mix': 0.0,\n",
       " 'outlook': 0.6931471805599453,\n",
       " 'soy': 0.0,\n",
       " 'bean': 0.0,\n",
       " 'us': 0.0,\n",
       " 'ukrain': 0.0,\n",
       " 'brazil': 0.0,\n",
       " 'mean': 0.0,\n",
       " 'canola': 0.6931471805599453,\n",
       " 'altern': 0.0,\n",
       " 'seed': 0.0,\n",
       " 'syme': 0.6931471805599453,\n",
       " 'malt': 0.6931471805599453,\n",
       " 'barley': 0.0,\n",
       " 'chines': 0.0,\n",
       " 'soon': 0.0,\n",
       " 'pick': 0.0,\n",
       " 'at': 0.0,\n",
       " 'stock': 0.0,\n",
       " 'canadian': 0.0,\n",
       " 'offer': 0.0,\n",
       " 'marketplac': 0.6931471805599453,\n",
       " 'good': 0.0,\n",
       " 'posit': 0.0,\n",
       " 'cours': 0.0,\n",
       " 'europ': 0.0,\n",
       " 'august': 0.0,\n",
       " 'away': 0.0,\n",
       " 'yet': 0.0,\n",
       " 'onc': 0.0,\n",
       " 'emerg': 0.0,\n",
       " ...}"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idf_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "def tfidf(tf,idf_data):\n",
    "    tf_idf={}\n",
    "    for word,value in tf.items():\n",
    "        tf_idf[word]=value*idf_data[word]\n",
    "    return(tf_idf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "tfidf_rural=tfidf(tf_rural,idf_data)\n",
    "tfidf_science=tfidf(tf_science,idf_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'pm': 0.0,\n",
       " 'deni': 0.0,\n",
       " 'knowledg': 0.0,\n",
       " 'awb': 0.002055900353022154,\n",
       " 'kickback': 0.00013178848416808676,\n",
       " 'the': 0.0,\n",
       " 'prime': 0.0,\n",
       " 'minist': 0.0,\n",
       " 'knew': 0.0,\n",
       " 'pay': 0.0,\n",
       " 'iraq': 0.0,\n",
       " 'despit': 0.0,\n",
       " 'write': 0.0,\n",
       " 'wheat': 0.0,\n",
       " 'export': 0.0,\n",
       " 'ask': 0.0,\n",
       " 'kept': 0.0,\n",
       " 'fulli': 0.0,\n",
       " 'inform': 0.0,\n",
       " 'sale': 0.0,\n",
       " 'letter': 0.0,\n",
       " 'john': 0.0,\n",
       " 'howard': 0.0,\n",
       " 'deputi': 0.0,\n",
       " 'mark': 0.0,\n",
       " 'vail': 0.0002560461978122829,\n",
       " 'releas': 0.0,\n",
       " 'cole': 0.0,\n",
       " 'inquiri': 0.0,\n",
       " 'oil': 0.0,\n",
       " 'food': 0.0,\n",
       " 'program': 0.0,\n",
       " 'in': 0.0,\n",
       " 'one': 0.0,\n",
       " 'mr': 0.0,\n",
       " 'manag': 0.0,\n",
       " 'director': 0.0,\n",
       " 'andrew': 0.0,\n",
       " 'lindberg': 6.401154945307072e-05,\n",
       " 'remain': 0.0,\n",
       " 'close': 0.0,\n",
       " 'contact': 0.0,\n",
       " 'govern': 0.0,\n",
       " 'opposit': 0.0,\n",
       " 'gavan': 3.388846735750803e-05,\n",
       " 'o': 0.0,\n",
       " 'connor': 0.0,\n",
       " 'say': 0.0,\n",
       " 'sent': 0.0,\n",
       " 'time': 0.0,\n",
       " 'though': 0.0,\n",
       " 'jordanian': 1.5061541047781346e-05,\n",
       " 'truck': 0.0,\n",
       " 'compani': 0.0,\n",
       " 'he': 0.0,\n",
       " 'longer': 0.0,\n",
       " 'wipe': 0.0,\n",
       " 'hand': 0.0,\n",
       " 'illicit': 0.0,\n",
       " 'payment': 0.0,\n",
       " 'total': 0.0,\n",
       " 'million': 0.0,\n",
       " 'respons': 0.0,\n",
       " 'must': 0.0,\n",
       " 'lay': 0.0,\n",
       " 'may': 0.0,\n",
       " 'squar': 0.0,\n",
       " 'feet': 0.0,\n",
       " 'coalit': 0.0,\n",
       " 'trade': 0.0,\n",
       " 'agricultur': 0.0,\n",
       " 'said': 0.0,\n",
       " 'but': 0.0,\n",
       " 'show': 0.0,\n",
       " 'inquir': 0.0,\n",
       " 'futur': 0.0,\n",
       " 'prove': 0.0,\n",
       " 'it': 0.0,\n",
       " 'would': 0.0,\n",
       " 'astonish': 0.0,\n",
       " 'i': 0.0,\n",
       " 'done': 0.0,\n",
       " 'anyth': 0.0,\n",
       " 'possibl': 0.0,\n",
       " 'could': 0.0,\n",
       " 'preserv': 0.0,\n",
       " 'australia': 0.0,\n",
       " 'valuabl': 0.0,\n",
       " 'market': 0.0,\n",
       " 'email': 0.0,\n",
       " 'question': 0.0,\n",
       " 'today': 0.0,\n",
       " 'peter': 0.0,\n",
       " 'geari': 1.5061541047781346e-05,\n",
       " 'receiv': 0.0,\n",
       " 'indic': 0.0,\n",
       " 'iraqi': 0.0,\n",
       " 'grain': 0.0,\n",
       " 'board': 0.0,\n",
       " 'approach': 0.0,\n",
       " 'provid': 0.0,\n",
       " 'servic': 0.0,\n",
       " 'forward': 0.0,\n",
       " 'two': 0.0,\n",
       " 'colleagu': 0.0,\n",
       " 'rememb': 0.0,\n",
       " 'read': 0.0,\n",
       " 'although': 0.0,\n",
       " 'skim': 3.7653852619453364e-06,\n",
       " 'support': 0.0,\n",
       " 'still': 0.0,\n",
       " 'plenti': 0.0,\n",
       " 'among': 0.0,\n",
       " 'grower': 0.0,\n",
       " 'central': 0.0,\n",
       " 'western': 0.0,\n",
       " 'new': 0.0,\n",
       " 'south': 0.0,\n",
       " 'wale': 0.0,\n",
       " 'revel': 0.0,\n",
       " 'produc': 0.0,\n",
       " 'broadli': 0.0,\n",
       " 'attempt': 0.0,\n",
       " 'get': 0.0,\n",
       " 'best': 0.0,\n",
       " 'price': 0.0,\n",
       " 'product': 0.0,\n",
       " 'think': 0.0,\n",
       " 'ploy': 3.7653852619453364e-06,\n",
       " 'oversea': 0.0,\n",
       " 'interest': 0.0,\n",
       " 'tri': 0.0,\n",
       " 'singl': 0.0,\n",
       " 'desk': 0.0,\n",
       " 'put': 0.0,\n",
       " 'asid': 0.0,\n",
       " 'stori': 0.0,\n",
       " 'go': 0.0,\n",
       " 'round': 0.0,\n",
       " 'commiss': 0.0,\n",
       " 'everyth': 0.0,\n",
       " 'way': 0.0,\n",
       " 'peopl': 0.0,\n",
       " 'got': 0.0,\n",
       " 'thing': 0.0,\n",
       " 'busi': 0.0,\n",
       " 'middl': 0.0,\n",
       " 'east': 0.0,\n",
       " 'asian': 0.0,\n",
       " 'countri': 0.0,\n",
       " 'actual': 0.0,\n",
       " 'pretti': 0.0,\n",
       " 'reason': 0.0,\n",
       " 'system': 0.0,\n",
       " 'give': 0.0,\n",
       " 'fair': 0.0,\n",
       " 'moment': 0.0,\n",
       " 'averag': 0.0,\n",
       " 'perform': 0.0,\n",
       " 'fairli': 0.0,\n",
       " 'well': 0.0,\n",
       " 'anoth': 0.0,\n",
       " 'biggest': 0.0,\n",
       " 'someon': 0.0,\n",
       " 'els': 0.0,\n",
       " 'take': 0.0,\n",
       " 'whether': 0.0,\n",
       " 'multin': 7.530770523890673e-06,\n",
       " 'much': 0.0,\n",
       " 'foothold': 0.0,\n",
       " 'advantag': 0.0,\n",
       " 'analyst': 0.0,\n",
       " 'predict': 0.0,\n",
       " 'drop': 0.0,\n",
       " 'tonn': 0.0,\n",
       " 'back': 0.0,\n",
       " 'malcolm': 0.0,\n",
       " 'bartholomaeu': 3.012308209556269e-05,\n",
       " 'pool': 0.0,\n",
       " 'return': 0.0,\n",
       " 'alreadi': 0.0,\n",
       " 'year': 0.0,\n",
       " 'past': 0.0,\n",
       " 'five': 0.0,\n",
       " 'premium': 0.00015438079573975878,\n",
       " 'achiev': 0.0,\n",
       " 'monopoli': 0.00013555386943003213,\n",
       " 'sever': 0.0,\n",
       " 'erod': 0.0,\n",
       " 'sa': 0.0,\n",
       " 'farmer': 0.0,\n",
       " 'help': 0.0,\n",
       " 'fire': 0.0,\n",
       " 'ravag': 0.0,\n",
       " 'neighbour': 0.0,\n",
       " 'donat': 0.0,\n",
       " 'truckload': 7.530770523890673e-06,\n",
       " 'hay': 0.0,\n",
       " 'across': 0.0,\n",
       " 'border': 0.0,\n",
       " 'wake': 0.0,\n",
       " 'grampian': 3.7653852619453366e-05,\n",
       " 'bushfir': 0.0,\n",
       " 'day': 0.0,\n",
       " 'agist': 4.5184623143344033e-05,\n",
       " 'cattl': 0.0,\n",
       " 'they': 0.0,\n",
       " 'begin': 0.0,\n",
       " 'fodder': 7.907309050085206e-05,\n",
       " 'drive': 0.0,\n",
       " 'coordin': 0.0,\n",
       " 'conner': 3.7653852619453364e-06,\n",
       " 'overwhelm': 0.0,\n",
       " 'all': 0.0,\n",
       " 'week': 0.0,\n",
       " 'gone': 0.0,\n",
       " 'place': 0.0,\n",
       " 'load': 0.0,\n",
       " 'we': 0.0,\n",
       " 'man': 0.0,\n",
       " 'full': 0.0,\n",
       " 'rest': 0.0,\n",
       " 'straight': 0.0,\n",
       " 'move': 0.0,\n",
       " 'floodwat': 0.0,\n",
       " 'highway': 0.0,\n",
       " 'a': 0.0,\n",
       " 'major': 0.0,\n",
       " 'northern': 0.0,\n",
       " 'territori': 0.0,\n",
       " 'block': 0.0,\n",
       " 'victoria': 0.0,\n",
       " 'river': 0.0,\n",
       " 'cut': 0.0,\n",
       " 'also': 0.0,\n",
       " 'flood': 0.0,\n",
       " 'remot': 0.0,\n",
       " 'pigeon': 0.0,\n",
       " 'hole': 0.0,\n",
       " 'aborigin': 0.0,\n",
       " 'commun': 0.0,\n",
       " 'stockman': 2.2592311571672017e-05,\n",
       " 'simon': 0.0,\n",
       " 'ham': 0.0,\n",
       " 'describ': 0.0,\n",
       " 'reloc': 0.0,\n",
       " 'hundr': 0.0,\n",
       " 'higher': 0.0,\n",
       " 'ground': 0.0,\n",
       " 'vehicl': 0.0,\n",
       " 'camp': 0.0,\n",
       " 'set': 0.0,\n",
       " 'belong': 0.0,\n",
       " 'coupl': 0.0,\n",
       " 'boat': 0.0,\n",
       " 'onto': 0.0,\n",
       " 'metr': 0.0,\n",
       " 'sandston': 0.0,\n",
       " 'ridg': 0.0,\n",
       " 'height': 0.0,\n",
       " 'might': 0.0,\n",
       " 'littl': 0.0,\n",
       " 'start': 0.0,\n",
       " 'sold': 0.0,\n",
       " 'm': 0.0,\n",
       " 'tasmania': 0.0,\n",
       " 'main': 0.0,\n",
       " 'chang': 0.0,\n",
       " 'second': 0.0,\n",
       " 'three': 0.0,\n",
       " 'former': 0.0,\n",
       " 'state': 0.0,\n",
       " 'own': 0.0,\n",
       " 'tasmanian': 0.0,\n",
       " 'elev': 0.0,\n",
       " 'local': 0.0,\n",
       " 'agribusi': 0.0,\n",
       " 'robert': 0.0,\n",
       " 'limit': 0.0,\n",
       " 'deal': 0.0,\n",
       " 'includ': 0.0,\n",
       " 'silo': 7.154231997696139e-05,\n",
       " 'launceston': 0.0,\n",
       " 'devonport': 3.7653852619453364e-06,\n",
       " 'powranna': 7.530770523890673e-06,\n",
       " 'macleod': 3.7653852619453364e-06,\n",
       " 'bid': 0.0,\n",
       " 'first': 0.0,\n",
       " 'privatis': 6.401154945307072e-05,\n",
       " 'disappoint': 0.0,\n",
       " 'success': 0.0,\n",
       " 'point': 0.0,\n",
       " 'wine': 0.0,\n",
       " 'worker': 0.0,\n",
       " 'stop': 0.0,\n",
       " 'work': 0.0,\n",
       " 'hardi': 0.0,\n",
       " 'stanley': 0.0,\n",
       " 'wineri': 0.00010166540207252408,\n",
       " 'walk': 0.0,\n",
       " 'job': 0.0,\n",
       " 'forti': 0.0,\n",
       " 'staff': 0.0,\n",
       " 'morn': 0.0,\n",
       " 'disput': 0.0,\n",
       " 'enterpris': 0.0,\n",
       " 'bargain': 0.0,\n",
       " 'agreement': 0.0,\n",
       " 'stopwork': 3.7653852619453364e-06,\n",
       " 'come': 0.0,\n",
       " 'region': 0.0,\n",
       " 'grape': 0.0,\n",
       " 'crush': 0.0,\n",
       " 'took': 0.0,\n",
       " 'matter': 0.0,\n",
       " 'industri': 0.0,\n",
       " 'relat': 0.0,\n",
       " 'irc': 3.7653852619453364e-06,\n",
       " 'friday': 0.0,\n",
       " 'wool': 0.0,\n",
       " 'bodi': 0.0,\n",
       " 'eye': 0.0,\n",
       " 'sportswear': 0.0,\n",
       " 'billion': 0.0,\n",
       " 'global': 0.0,\n",
       " 'target': 0.0,\n",
       " 'promot': 0.0,\n",
       " 'australian': 0.0,\n",
       " 'innov': 0.0,\n",
       " 'awi': 0.00024098465676450153,\n",
       " 'blend': 0.0,\n",
       " 'shirt': 0.0,\n",
       " 'casual': 1.8826926309726683e-05,\n",
       " 'wear': 0.0,\n",
       " 'manufactur': 0.0,\n",
       " 'largest': 0.0,\n",
       " 'held': 0.0,\n",
       " 'germani': 0.0,\n",
       " 'len': 0.0,\n",
       " 'stephen': 0.0,\n",
       " 'pricier': 3.7653852619453364e-06,\n",
       " 'end': 0.0,\n",
       " 'shopper': 0.0,\n",
       " 'will': 0.0,\n",
       " 'sport': 0.0,\n",
       " 'apparel': 7.530770523890673e-06,\n",
       " 'sector': 0.0,\n",
       " 'certainli': 0.0,\n",
       " 'fastest': 0.0,\n",
       " 'grow': 0.0,\n",
       " 'world': 0.0,\n",
       " 'secret': 0.0,\n",
       " 'big': 0.0,\n",
       " 'share': 0.0,\n",
       " 'level': 0.0,\n",
       " 'particularli': 0.0,\n",
       " 'merino': 0.0,\n",
       " 'realli': 0.0,\n",
       " 'almost': 0.0,\n",
       " 'radar': 0.0,\n",
       " 'organis': 0.0,\n",
       " 'step': 0.0,\n",
       " 'ban': 0.0,\n",
       " 'rodeo': 0.00015438079573975878,\n",
       " 'campaign': 0.0,\n",
       " 'sustain': 0.0,\n",
       " 'attack': 0.0,\n",
       " 'anim': 0.0,\n",
       " 'right': 0.0,\n",
       " 'activist': 0.0,\n",
       " 'destroy': 0.0,\n",
       " 'weekend': 0.0,\n",
       " 'event': 0.0,\n",
       " 'buck': 0.0,\n",
       " 'hors': 0.0,\n",
       " 'break': 0.0,\n",
       " 'leg': 0.0,\n",
       " 'ulverston': 1.1296155785836008e-05,\n",
       " 'saturday': 0.0,\n",
       " 'ago': 0.0,\n",
       " 'bull': 0.0,\n",
       " 'appar': 0.0,\n",
       " 'ride': 0.0,\n",
       " 'competit': 0.0,\n",
       " 'carrick': 3.7653852619453364e-06,\n",
       " 'owner': 0.0,\n",
       " 'brian': 0.0,\n",
       " 'fish': 0.0,\n",
       " 'unfortun': 0.0,\n",
       " 'accid': 0.0,\n",
       " 'welfar': 0.0,\n",
       " 'issu': 0.0,\n",
       " 'emma': 0.0,\n",
       " 'haswel': 0.0,\n",
       " 'against': 1.1296155785836008e-05,\n",
       " 'cruelti': 4.8950008405289374e-05,\n",
       " 'effort': 0.0,\n",
       " 'snail': 0.0,\n",
       " 'use': 0.0,\n",
       " 'cancer': 0.0,\n",
       " 'research': 0.0,\n",
       " 'investig': 0.0,\n",
       " 'sea': 0.0,\n",
       " 'eventu': 0.0,\n",
       " 'treat': 0.0,\n",
       " 'until': 0.0,\n",
       " 'popul': 0.0,\n",
       " 'virtual': 0.0,\n",
       " 'unexplor': 0.0,\n",
       " 'flinder': 0.0,\n",
       " 'univers': 0.0,\n",
       " 'hope': 0.0,\n",
       " 'discov': 0.0,\n",
       " 'benefici': 0.0,\n",
       " 'compound': 0.0,\n",
       " 'predatori': 0.0,\n",
       " 'known': 0.0,\n",
       " 'whelk': 7.530770523890673e-06,\n",
       " 'marin': 0.0,\n",
       " 'biologist': 0.0,\n",
       " 'dr': 0.0,\n",
       " 'kirsten': 3.7653852619453364e-06,\n",
       " 'benkendorff': 3.7653852619453364e-06,\n",
       " 'follow': 0.0,\n",
       " 'clinic': 0.0,\n",
       " 'trial': 0.0,\n",
       " 'anti': 0.0,\n",
       " 'properti': 0.0,\n",
       " 'look': 0.0,\n",
       " 'current': 0.0,\n",
       " 'harvest': 0.0,\n",
       " 'consid': 0.0,\n",
       " 'resourc': 0.0,\n",
       " 'lot': 0.0,\n",
       " 'potenti': 0.0,\n",
       " 'econom': 0.0,\n",
       " 'benefit': 0.0,\n",
       " 'side': 0.0,\n",
       " 'processor': 0.0,\n",
       " 'fail': 0.0,\n",
       " 'meet': 0.0,\n",
       " 'kangaroo': 0.0,\n",
       " 'meat': 0.0,\n",
       " 'demand': 0.0,\n",
       " 'intern': 0.0,\n",
       " 'pressur': 0.0,\n",
       " 'high': 0.0,\n",
       " 'quota': 4.8950008405289374e-05,\n",
       " 'met': 0.0,\n",
       " 'leav': 0.0,\n",
       " 'gap': 0.0,\n",
       " 'phil': 0.0,\n",
       " 'franolick': 3.7653852619453364e-06,\n",
       " 'king': 0.0,\n",
       " 'perth': 0.0,\n",
       " 'european': 0.0,\n",
       " 'attract': 0.0,\n",
       " 'low': 0.0,\n",
       " 'fat': 0.0,\n",
       " 'content': 0.0,\n",
       " 'russian': 0.0,\n",
       " 'consum': 0.0,\n",
       " 'larg': 0.0,\n",
       " 'amount': 0.0,\n",
       " 'sausag': 3.7653852619453364e-06,\n",
       " 'minc': 3.7653852619453364e-06,\n",
       " 'as': 0.0,\n",
       " 'far': 0.0,\n",
       " 'know': 0.0,\n",
       " 'everi': 0.0,\n",
       " 'skerrick': 3.7653852619453364e-06,\n",
       " 'human': 0.0,\n",
       " 'consumpt': 0.0,\n",
       " 'goe': 0.0,\n",
       " 'warn': 0.0,\n",
       " 'snake': 0.0,\n",
       " 'alert': 0.0,\n",
       " 'with': 0.0,\n",
       " 'temperatur': 0.0,\n",
       " 'soar': 0.0,\n",
       " 'summer': 0.0,\n",
       " 'southern': 0.0,\n",
       " 'keep': 0.0,\n",
       " 'open': 0.0,\n",
       " 'paddock': 0.00015061541047781346,\n",
       " 'especi': 0.0,\n",
       " 'area': 0.0,\n",
       " 'hit': 0.0,\n",
       " 'forc': 0.0,\n",
       " 'usual': 0.0,\n",
       " 'habitat': 0.0,\n",
       " 'maria': 0.0,\n",
       " 'dovey': 3.7653852619453364e-06,\n",
       " 'reptil': 0.0,\n",
       " 'rescu': 0.0,\n",
       " 'word': 0.0,\n",
       " 'advic': 0.0,\n",
       " 'ani': 0.0,\n",
       " 'regularli': 0.0,\n",
       " 'outdoor': 0.0,\n",
       " 'bush': 0.0,\n",
       " 'land': 0.0,\n",
       " 'alway': 0.0,\n",
       " 'carri': 0.0,\n",
       " 'least': 0.0,\n",
       " 'bandag': 3.7653852619453364e-06,\n",
       " 'mobil': 0.0,\n",
       " 'phone': 0.0,\n",
       " 'aerial': 0.0,\n",
       " 'spray': 0.0,\n",
       " 'control': 0.0,\n",
       " 'locust': 0.0,\n",
       " 'threat': 0.0,\n",
       " 'begun': 0.0,\n",
       " 'plagu': 0.0,\n",
       " 'made': 0.0,\n",
       " 'continu': 0.0,\n",
       " 'rain': 0.0,\n",
       " 'warm': 0.0,\n",
       " 'weather': 0.0,\n",
       " 'assist': 0.0,\n",
       " 'breed': 0.0,\n",
       " 'condit': 0.0,\n",
       " 'pastur': 0.0,\n",
       " 'crop': 0.0,\n",
       " 'pest': 0.0,\n",
       " 'injuri': 0.0,\n",
       " 'baton': 0.0,\n",
       " 'runner': 0.0,\n",
       " 'queensland': 0.0,\n",
       " 'knee': 0.0,\n",
       " 'run': 0.0,\n",
       " 'queen': 0.0,\n",
       " 'relay': 0.0,\n",
       " 'lead': 0.0,\n",
       " 'commonwealth': 0.0,\n",
       " 'game': 0.0,\n",
       " 'stanthorp': 1.8826926309726683e-05,\n",
       " 'bill': 0.0,\n",
       " 'bonner': 3.7653852619453364e-06,\n",
       " 'travel': 0.0,\n",
       " 'coff': 1.1296155785836008e-05,\n",
       " 'harbour': 0.0,\n",
       " 'and': 0.0,\n",
       " 'prevent': 0.0,\n",
       " 'make': 0.0,\n",
       " 'dash': 1.5061541047781346e-05,\n",
       " 'about': 0.0,\n",
       " 'jump': 0.0,\n",
       " 'motorbik': 0.0,\n",
       " 'twist': 0.0,\n",
       " 'ligament': 3.7653852619453364e-06,\n",
       " 'abl': 0.0,\n",
       " 'sinc': 0.0,\n",
       " 'promis': 0.0,\n",
       " 'adrenalin': 0.0,\n",
       " 'rush': 0.0,\n",
       " 'nz': 6.777693471501606e-05,\n",
       " 'appl': 0.0,\n",
       " 'debat': 0.0,\n",
       " 'zealand': 0.0,\n",
       " 'newspap': 0.0,\n",
       " 'member': 0.0,\n",
       " 'voic': 0.0,\n",
       " 'concern': 0.0,\n",
       " 'regard': 0.0,\n",
       " 'legitimaci': 7.530770523890673e-06,\n",
       " 'scienc': 0.0,\n",
       " 'draft': 0.0,\n",
       " 'import': 0.0,\n",
       " 'risk': 0.0,\n",
       " 'analysi': 0.0,\n",
       " 'last': 0.0,\n",
       " 'month': 0.0,\n",
       " 'biosecur': 0.0,\n",
       " 'ira': 1.5061541047781346e-05,\n",
       " 'determin': 0.0,\n",
       " 'fireblight': 1.8826926309726683e-05,\n",
       " 'diseas': 0.0,\n",
       " 'under': 0.0,\n",
       " 'rule': 0.0,\n",
       " 'refus': 0.0,\n",
       " 'scientif': 0.0,\n",
       " 'evid': 0.0,\n",
       " 'substanti': 0.0,\n",
       " 'accord': 0.0,\n",
       " 'profession': 0.0,\n",
       " 'matur': 0.0,\n",
       " 'pose': 0.0,\n",
       " 'howev': 0.0,\n",
       " 'spokesperson': 0.0,\n",
       " 'base': 0.0,\n",
       " 'access': 0.0,\n",
       " 'group': 0.0,\n",
       " 'allison': 3.7653852619453364e-06,\n",
       " 'mathemat': 0.0,\n",
       " 'model': 0.0,\n",
       " 'formula': 0.0,\n",
       " 'assess': 0.0,\n",
       " 'report': 0.0,\n",
       " 'publish': 0.0,\n",
       " 'decemb': 0.0,\n",
       " 'anyon': 0.0,\n",
       " 'compet': 0.0,\n",
       " 'reproduc': 0.0,\n",
       " 'given': 0.0,\n",
       " 'chairman': 0.0,\n",
       " 'task': 0.0,\n",
       " 'corboy': 3.7653852619453364e-06,\n",
       " 'surpris': 0.0,\n",
       " 'exercis': 0.0,\n",
       " 'present': 0.0,\n",
       " 'younger': 0.0,\n",
       " 'tree': 0.0,\n",
       " 'pear': 8.28384757627974e-05,\n",
       " 'plot': 0.0,\n",
       " 'young': 0.0,\n",
       " 'william': 0.0,\n",
       " 'fruit': 0.0,\n",
       " 'plant': 0.0,\n",
       " 'six': 0.0,\n",
       " 'old': 0.0,\n",
       " 'part': 0.0,\n",
       " 'experiment': 0.0,\n",
       " 'project': 0.0,\n",
       " 'fund': 0.0,\n",
       " 'spc': 1.1296155785836008e-05,\n",
       " 'ardmona': 1.1296155785836008e-05,\n",
       " 'can': 0.0,\n",
       " 'associ': 0.0,\n",
       " 'horticultur': 0.00028616927990784556,\n",
       " 'lenn': 3.7653852619453364e-06,\n",
       " 'famili': 0.0,\n",
       " 'orchard': 0.00013555386943003213,\n",
       " 'locat': 0.0,\n",
       " 'consult': 0.0,\n",
       " 'ba': 0.0,\n",
       " 'van': 0.0,\n",
       " 'den': 0.0,\n",
       " 'aim': 0.0,\n",
       " 'increas': 0.0,\n",
       " 'number': 0.0,\n",
       " 'grown': 0.0,\n",
       " 'encourag': 0.0,\n",
       " 'earlier': 0.0,\n",
       " 'feder': 0.0,\n",
       " 'nation': 0.0,\n",
       " 'mp': 0.0,\n",
       " 'park': 0.0,\n",
       " 'cobb': 1.5061541047781346e-05,\n",
       " 'brought': 0.0,\n",
       " 'thi': 0.0,\n",
       " 'examin': 0.0,\n",
       " 'role': 0.0,\n",
       " 'on': 0.0,\n",
       " 'top': 0.0,\n",
       " 'parti': 0.0,\n",
       " 'reinvent': 7.530770523890673e-06,\n",
       " 'becom': 0.0,\n",
       " 'pro': 0.0,\n",
       " 'activ': 0.0,\n",
       " 'differ': 0.0,\n",
       " 'liber': 0.0,\n",
       " 'repres': 0.0,\n",
       " 'better': 0.0,\n",
       " 'death': 0.0,\n",
       " 'believ': 0.0,\n",
       " 'due': 0.0,\n",
       " 'lack': 0.0,\n",
       " 'oxygen': 0.0,\n",
       " 'depart': 0.0,\n",
       " 'primari': 0.0,\n",
       " 'kill': 0.0,\n",
       " 'lachlan': 3.7653852619453366e-05,\n",
       " 'upstream': 0.0,\n",
       " 'lake': 0.0,\n",
       " 'cargelligo': 3.7653852619453364e-06,\n",
       " 'weir': 2.6357696833617357e-05,\n",
       " 'between': 0.0,\n",
       " 'nativ': 0.0,\n",
       " 'die': 0.0,\n",
       " 'incid': 0.0,\n",
       " 'district': 0.0,\n",
       " 'fisheri': 0.0,\n",
       " 'offic': 0.0,\n",
       " 'phillip': 0.0,\n",
       " 'mccarthi': 3.7653852619453364e-06,\n",
       " 'murray': 0.0,\n",
       " 'cod': 0.0,\n",
       " 'bream': 3.7653852619453364e-06,\n",
       " 'golden': 0.0,\n",
       " 'silver': 0.0,\n",
       " 'perch': 0.0,\n",
       " 'introduc': 0.0,\n",
       " 'speci': 0.0,\n",
       " 'suggest': 0.0,\n",
       " 'result': 0.0,\n",
       " 'water': 0.0,\n",
       " 'flow': 0.0,\n",
       " 'chemic': 0.0,\n",
       " 'poison': 0.0,\n",
       " 'veget': 0.0,\n",
       " 'label': 0.0,\n",
       " 'ausveg': 7.907309050085206e-05,\n",
       " 'law': 0.0,\n",
       " 'joke': 0.0,\n",
       " 'nobodi': 0.0,\n",
       " 'enforc': 0.0,\n",
       " 'non': 0.0,\n",
       " 'exist': 0.0,\n",
       " 'fool': 0.0,\n",
       " 'buy': 0.0,\n",
       " 'fals': 0.0,\n",
       " 'poor': 0.0,\n",
       " 'red': 0.0,\n",
       " 'tape': 0.0,\n",
       " 'slow': 0.0,\n",
       " 'west': 0.0,\n",
       " 'kimberley': 0.0,\n",
       " 'wa': 0.0,\n",
       " 'bowl': 0.0,\n",
       " 'soil': 0.0,\n",
       " 'avail': 0.0,\n",
       " 'sunshin': 0.0,\n",
       " 'face': 0.0,\n",
       " 'hurdl': 0.0,\n",
       " 'expand': 0.0,\n",
       " 'sweet': 0.0,\n",
       " 'potato': 0.0,\n",
       " 'paul': 0.0,\n",
       " 'higgin': 0.0,\n",
       " 'reli': 0.0,\n",
       " 'gener': 0.0,\n",
       " 'broom': 0.0,\n",
       " 'horticulturalist': 3.388846735750803e-05,\n",
       " 'connect': 0.0,\n",
       " 'grid': 0.0,\n",
       " 'appli': 0.0,\n",
       " 'nine': 0.0,\n",
       " 'kilometr': 0.0,\n",
       " 'extens': 0.0,\n",
       " 'bring': 0.0,\n",
       " 'line': 0.0,\n",
       " 'power': 0.0,\n",
       " 'custom': 0.0,\n",
       " 'cost': 0.0,\n",
       " 'regardless': 0.0,\n",
       " 'distanc': 0.0,\n",
       " 'acquisit': 0.0,\n",
       " 'caus': 0.0,\n",
       " 'headach': 0.0,\n",
       " 'purchas': 0.0,\n",
       " 'adjoin': 7.530770523890673e-06,\n",
       " 'june': 0.0,\n",
       " 'titl': 0.0,\n",
       " 'improv': 0.0,\n",
       " 'clear': 0.0,\n",
       " 'resolv': 0.0,\n",
       " 'survey': 0.0,\n",
       " 'bed': 0.0,\n",
       " 'moni': 3.7653852619453364e-06,\n",
       " 'paid': 0.0,\n",
       " 'plan': 0.0,\n",
       " 'infrastructur': 0.0,\n",
       " 'alannah': 3.7653852619453364e-06,\n",
       " 'mctiernan': 7.530770523890673e-06,\n",
       " 'seven': 0.0,\n",
       " 'delay': 0.0,\n",
       " 'regrett': 7.530770523890673e-06,\n",
       " 'awar': 0.0,\n",
       " 'hold': 0.0,\n",
       " 'correctli': 0.0,\n",
       " 'zone': 0.0,\n",
       " 'ms': 0.0,\n",
       " 'favour': 0.0,\n",
       " 'wait': 0.0,\n",
       " 'offici': 0.0,\n",
       " 'plough': 0.0,\n",
       " 'pindan': 3.7653852619453364e-06,\n",
       " 'richmond': 0.0,\n",
       " 'stay': 0.0,\n",
       " 'fisher': 0.0,\n",
       " 'decid': 0.0,\n",
       " 'commerci': 0.0,\n",
       " 'recreat': 0.0,\n",
       " 'dead': 0.0,\n",
       " 'eel': 0.0,\n",
       " 'near': 0.0,\n",
       " 'wardel': 3.7653852619453364e-06,\n",
       " 'thursday': 0.0,\n",
       " 'alarmingli': 3.7653852619453364e-06,\n",
       " 'act': 0.0,\n",
       " 'anthoni': 0.0,\n",
       " 'hirst': 3.7653852619453364e-06,\n",
       " 'inspector': 0.0,\n",
       " 'sampl': 0.0,\n",
       " 'tomorrow': 0.0,\n",
       " 'rural': 0.0,\n",
       " 'woman': 0.0,\n",
       " 'finalist': 1.5061541047781346e-05,\n",
       " 'next': 0.0,\n",
       " 'seem': 0.0,\n",
       " 'like': 0.0,\n",
       " 'etern': 0.0,\n",
       " 'rosebank': 3.7653852619453364e-06,\n",
       " 'catherin': 0.0,\n",
       " 'ford': 0.0,\n",
       " 'quest': 0.0,\n",
       " 'find': 0.0,\n",
       " 'macadamia': 0.0,\n",
       " 'win': 0.0,\n",
       " 'guid': 0.0,\n",
       " 'admit': 0.0,\n",
       " 'amaz': 0.0,\n",
       " 'even': 0.0,\n",
       " 'dse': 3.7653852619453364e-06,\n",
       " 'proactiv': 1.8826926309726683e-05,\n",
       " 'burn': 0.0,\n",
       " 'yesterday': 0.0,\n",
       " 'heard': 0.0,\n",
       " 'cfa': 1.5061541047781346e-05,\n",
       " 'feel': 0.0,\n",
       " 'intens': 0.0,\n",
       " 'lessen': 0.0,\n",
       " 'fuel': 0.0,\n",
       " 'reduct': 0.0,\n",
       " 'charley': 3.7653852619453364e-06,\n",
       " 'defegley': 3.7653852619453364e-06,\n",
       " 'ararat': 1.1296155785836008e-05,\n",
       " 'vff': 7.907309050085206e-05,\n",
       " 'branch': 0.0,\n",
       " 'secretari': 0.0,\n",
       " 'division': 3.7653852619453364e-06,\n",
       " 'command': 0.0,\n",
       " 'felt': 0.0,\n",
       " 'enough': 0.0,\n",
       " 'contribut': 0.0,\n",
       " 'destruct': 0.0,\n",
       " 'horsham': 1.1296155785836008e-05,\n",
       " 'geoff': 0.0,\n",
       " 'evan': 0.0,\n",
       " 'agenc': 0.0,\n",
       " 'polici': 0.0,\n",
       " 'hamper': 0.0,\n",
       " 'landown': 0.0,\n",
       " 'keen': 0.0,\n",
       " 'besid': 0.0,\n",
       " 'allow': 0.0,\n",
       " 'small': 0.0,\n",
       " 'window': 0.0,\n",
       " 'opportun': 0.0,\n",
       " 'strong': 0.0,\n",
       " 'divis': 0.0,\n",
       " 'within': 0.0,\n",
       " 'less': 0.0,\n",
       " 'hectar': 0.0,\n",
       " 'burnt': 0.0,\n",
       " 'ag': 0.0,\n",
       " 'speedi': 0.0,\n",
       " 'aid': 0.0,\n",
       " 'announc': 0.0,\n",
       " 'commit': 0.0,\n",
       " 'dollar': 0.0,\n",
       " 'recoveri': 0.0,\n",
       " 'blaze': 0.0,\n",
       " 'ministeri': 0.0,\n",
       " 'need': 0.0,\n",
       " 'hardship': 1.5061541047781346e-05,\n",
       " 'suppli': 0.0,\n",
       " 'firefight': 0.0,\n",
       " 'victorian': 0.0,\n",
       " 'haulag': 1.8826926309726683e-05,\n",
       " 'contractor': 0.0,\n",
       " 'grassroot': 3.7653852619453364e-06,\n",
       " 'affect': 0.0,\n",
       " 'semi': 0.0,\n",
       " 'collect': 0.0,\n",
       " 'moyston': 3.7653852619453364e-06,\n",
       " 'warrnambool': 1.8826926309726683e-05,\n",
       " 'surviv': 0.0,\n",
       " 'bruce': 0.0,\n",
       " 'mckay': 1.8826926309726683e-05,\n",
       " 'farm': 0.0,\n",
       " 'sheep': 0.0,\n",
       " 'fix': 0.0,\n",
       " 'fenc': 0.0,\n",
       " 'willaura': 3.7653852619453364e-06,\n",
       " 'tim': 0.0,\n",
       " 'wrench': 7.530770523890673e-06,\n",
       " 'coop': 3.7653852619453364e-06,\n",
       " 'came': 0.0,\n",
       " 'around': 0.0,\n",
       " 'haulier': 3.7653852619453364e-06,\n",
       " 'scott': 0.0,\n",
       " 'wilson': 0.0,\n",
       " 'koroit': 3.7653852619453364e-06,\n",
       " 'went': 0.0,\n",
       " 'neil': 0.0,\n",
       " 'jame': 0.0,\n",
       " 'dpi': 3.388846735750803e-05,\n",
       " 'charli': 2.2592311571672017e-05,\n",
       " 'defegeley': 3.7653852619453364e-06,\n",
       " 'blame': 0.0,\n",
       " 'hike': 0.0,\n",
       " 'fresh': 0.0,\n",
       " 'fluctuat': 0.0,\n",
       " 'per': 0.0,\n",
       " 'cent': 0.0,\n",
       " 'latest': 0.0,\n",
       " 'rabobank': 5.2715393667234715e-05,\n",
       " 'milk': 0.0,\n",
       " 'ben': 0.0,\n",
       " 'russel': 0.0,\n",
       " 'money': 0.0,\n",
       " 'case': 0.0,\n",
       " 'pass': 0.0,\n",
       " 'contrari': 0.0,\n",
       " 'see': 0.0,\n",
       " 'retail': 0.0,\n",
       " 'fall': 0.0,\n",
       " 'tradit': 0.0,\n",
       " 'egypt': 0.0,\n",
       " 'pakistan': 0.0,\n",
       " 'china': 0.0,\n",
       " 'india': 0.0,\n",
       " 'deliv': 0.0,\n",
       " 'expect': 0.0,\n",
       " 'contract': 0.0,\n",
       " 'record': 0.0,\n",
       " 'tender': 0.0,\n",
       " 'fierc': 0.0,\n",
       " 'trader': 0.0,\n",
       " 'lloyd': 0.0,\n",
       " 'georg': 0.0,\n",
       " 'rang': 0.0,\n",
       " 'factor': 0.0,\n",
       " 'period': 0.0,\n",
       " 'signific': 0.0,\n",
       " 'tonnag': 3.012308209556269e-05,\n",
       " 'recent': 0.0,\n",
       " 'quantiti': 0.0,\n",
       " 'whole': 0.0,\n",
       " 'bunch': 0.0,\n",
       " 'broader': 0.0,\n",
       " 'wherebi': 0.0,\n",
       " 'sell': 0.0,\n",
       " 'quiet': 0.0,\n",
       " 'david': 0.0,\n",
       " 'johnson': 0.0,\n",
       " 'impact': 0.0,\n",
       " 'lost': 0.0,\n",
       " 'submit': 0.0,\n",
       " 'particip': 0.0,\n",
       " 'absolut': 0.0,\n",
       " 'probabl': 0.0,\n",
       " 'rel': 0.0,\n",
       " 'compar': 0.0,\n",
       " 'structur': 0.0,\n",
       " 'previou': 0.0,\n",
       " 'fact': 0.0,\n",
       " 'extrem': 0.0,\n",
       " 'comment': 0.0,\n",
       " 'season': 0.0,\n",
       " 'estim': 0.0,\n",
       " 'lift': 0.0,\n",
       " 'hard': 0.0,\n",
       " 'noodl': 3.7653852619453364e-06,\n",
       " 'grade': 0.0,\n",
       " 'there': 0.0,\n",
       " 'mix': 0.0,\n",
       " 'outlook': 0.0002070961894069935,\n",
       " 'soy': 0.0,\n",
       " 'bean': 0.0,\n",
       " 'us': 0.0,\n",
       " 'ukrain': 0.0,\n",
       " 'brazil': 0.0,\n",
       " 'mean': 0.0,\n",
       " 'canola': 0.00034641544409897096,\n",
       " 'altern': 0.0,\n",
       " 'seed': 0.0,\n",
       " 'syme': 1.1296155785836008e-05,\n",
       " 'malt': 3.012308209556269e-05,\n",
       " 'barley': 0.0,\n",
       " 'chines': 0.0,\n",
       " 'soon': 0.0,\n",
       " 'pick': 0.0,\n",
       " 'at': 0.0,\n",
       " 'stock': 0.0,\n",
       " 'canadian': 0.0,\n",
       " 'offer': 0.0,\n",
       " 'marketplac': 2.6357696833617357e-05,\n",
       " 'good': 0.0,\n",
       " 'posit': 0.0,\n",
       " 'cours': 0.0,\n",
       " 'europ': 0.0,\n",
       " 'august': 0.0,\n",
       " 'away': 0.0,\n",
       " 'yet': 0.0,\n",
       " 'onc': 0.0,\n",
       " 'emerg': 0.0,\n",
       " ...}"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_rural"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'cystic': 0.0,\n",
       " 'fibrosi': 0.0,\n",
       " 'affect': 0.0,\n",
       " 'children': 0.0,\n",
       " 'young': 0.0,\n",
       " 'adult': 0.0,\n",
       " 'us': 0.0,\n",
       " 'alon': 0.0,\n",
       " 'inhal': 0.0,\n",
       " 'mist': 0.0,\n",
       " 'salt': 0.0,\n",
       " 'water': 0.0,\n",
       " 'reduc': 0.0,\n",
       " 'pu': 0.0,\n",
       " 'infect': 0.0,\n",
       " 'fill': 0.0,\n",
       " 'airway': 0.0,\n",
       " 'suffer': 0.0,\n",
       " 'although': 0.0,\n",
       " 'side': 0.0,\n",
       " 'effect': 0.0,\n",
       " 'includ': 0.0,\n",
       " 'nasti': 0.0,\n",
       " 'cough': 0.0,\n",
       " 'fit': 0.0,\n",
       " 'harsh': 0.0,\n",
       " 'tast': 0.0,\n",
       " 'that': 0.0,\n",
       " 'conclus': 0.0,\n",
       " 'two': 0.0,\n",
       " 'studi': 0.0,\n",
       " 'publish': 0.0,\n",
       " 'week': 0.0,\n",
       " 'issu': 0.0,\n",
       " 'the': 0.0,\n",
       " 'new': 0.0,\n",
       " 'england': 0.0,\n",
       " 'journal': 0.0,\n",
       " 'medicin': 0.0,\n",
       " 'they': 0.0,\n",
       " 'found': 0.0,\n",
       " 'content': 0.0,\n",
       " 'improv': 0.0,\n",
       " 'lung': 0.0,\n",
       " 'function': 0.0,\n",
       " 'case': 0.0,\n",
       " 'produc': 0.0,\n",
       " 'less': 0.0,\n",
       " 'absente': 0.0,\n",
       " 'school': 0.0,\n",
       " 'work': 0.0,\n",
       " 'progress': 0.0,\n",
       " 'frequent': 0.0,\n",
       " 'fatal': 0.0,\n",
       " 'genet': 0.0,\n",
       " 'diseas': 0.0,\n",
       " 'mark': 0.0,\n",
       " 'thicken': 0.0,\n",
       " 'mucu': 0.0,\n",
       " 'make': 0.0,\n",
       " 'harder': 0.0,\n",
       " 'clear': 0.0,\n",
       " 'debri': 0.0,\n",
       " 'bacteria': 0.0,\n",
       " 'solut': 0.0,\n",
       " 'realli': 0.0,\n",
       " 'open': 0.0,\n",
       " 'avenu': 0.0,\n",
       " 'approach': 0.0,\n",
       " 'patient': 0.0,\n",
       " 'treat': 0.0,\n",
       " 'say': 0.0,\n",
       " 'dr': 0.0,\n",
       " 'gail': 0.0,\n",
       " 'weinmann': 0.0,\n",
       " 'nation': 0.0,\n",
       " 'heart': 0.0,\n",
       " 'blood': 0.0,\n",
       " 'institut': 0.0,\n",
       " 'sponsor': 0.0,\n",
       " 'one': 0.0,\n",
       " 'elkin': 0.0,\n",
       " 'royal': 0.0,\n",
       " 'princ': 0.0,\n",
       " 'alfr': 0.0,\n",
       " 'hospit': 0.0,\n",
       " 'sydney': 0.0,\n",
       " 'australia': 0.0,\n",
       " 'colleagu': 0.0,\n",
       " 'author': 0.0,\n",
       " 'team': 0.0,\n",
       " 'volunt': 0.0,\n",
       " 'regularli': 0.0,\n",
       " 'salti': 0.0,\n",
       " 'fewer': 0.0,\n",
       " 'breath': 0.0,\n",
       " 'problem': 0.0,\n",
       " 'ad': 0.0,\n",
       " 'surfac': 0.0,\n",
       " 'benefici': 0.0,\n",
       " 'conclud': 0.0,\n",
       " 'all': 0.0,\n",
       " 'first': 0.0,\n",
       " 'chemic': 0.0,\n",
       " 'tri': 0.0,\n",
       " 'passag': 0.0,\n",
       " 'much': 0.0,\n",
       " 'possibl': 0.0,\n",
       " 'in': 0.0,\n",
       " 'second': 0.0,\n",
       " 'base': 0.0,\n",
       " 'assist': 0.0,\n",
       " 'profsesor': 0.0,\n",
       " 'scott': 0.0,\n",
       " 'donaldson': 0.0,\n",
       " 'univers': 0.0,\n",
       " 'north': 0.0,\n",
       " 'carolina': 0.0,\n",
       " 'chapel': 0.0,\n",
       " 'hill': 0.0,\n",
       " 'sustain': 0.0,\n",
       " 'acceler': 0.0,\n",
       " 'clearanc': 0.0,\n",
       " 'help': 0.0,\n",
       " 'hydrat': 0.0,\n",
       " 'accompani': 0.0,\n",
       " 'editori': 0.0,\n",
       " 'felix': 0.0,\n",
       " 'ratjen': 0.0,\n",
       " 'sick': 0.0,\n",
       " 'toronto': 0.0,\n",
       " 'canada': 0.0,\n",
       " 'cite': 0.0,\n",
       " 'sever': 0.0,\n",
       " 'unpleas': 0.0,\n",
       " 'treatment': 0.0,\n",
       " 'bad': 0.0,\n",
       " 'lengthi': 0.0,\n",
       " 'minut': 0.0,\n",
       " 'take': 0.0,\n",
       " 'administ': 0.0,\n",
       " 'he': 0.0,\n",
       " 'may': 0.0,\n",
       " 'receiv': 0.0,\n",
       " 'best': 0.0,\n",
       " 'long': 0.0,\n",
       " 'term': 0.0,\n",
       " 'antibiot': 0.0,\n",
       " 'would': 0.0,\n",
       " 'appear': 0.0,\n",
       " 'peopl': 0.0,\n",
       " 'get': 0.0,\n",
       " 'better': 0.0,\n",
       " 'drug': 0.0,\n",
       " 'limit': 0.0,\n",
       " 'inher': 0.0,\n",
       " 'mean': 0.0,\n",
       " 'step': 0.0,\n",
       " 'scientist': 0.0,\n",
       " 'rescu': 0.0,\n",
       " 'mous': 0.0,\n",
       " 'immun': 0.0,\n",
       " 'system': 0.0,\n",
       " 'overwhelm': 0.0,\n",
       " 'but': 0.0,\n",
       " 'human': 0.0,\n",
       " 'discov': 0.0,\n",
       " 'invad': 0.0,\n",
       " 'whole': 0.0,\n",
       " 'bodi': 0.0,\n",
       " 'like': 0.0,\n",
       " 'malaria': 0.0,\n",
       " 'disabl': 0.0,\n",
       " 'prevent': 0.0,\n",
       " 'detect': 0.0,\n",
       " 'fight': 0.0,\n",
       " 'microorgan': 0.0,\n",
       " 'australian': 0.0,\n",
       " 'german': 0.0,\n",
       " 'research': 0.0,\n",
       " 'discoveri': 0.0,\n",
       " 'develop': 0.0,\n",
       " 'vaccin': 0.0,\n",
       " 'restor': 0.0,\n",
       " 'today': 0.0,\n",
       " 'onlin': 0.0,\n",
       " 'natur': 0.0,\n",
       " 'immunolog': 0.0,\n",
       " 'respons': 0.0,\n",
       " 'alert': 0.0,\n",
       " 'specialis': 0.0,\n",
       " 'sentri': 0.0,\n",
       " 'dendrit': 0.0,\n",
       " 'cell': 0.0,\n",
       " 'viru': 0.0,\n",
       " 'captur': 0.0,\n",
       " 'display': 0.0,\n",
       " 'fragment': 0.0,\n",
       " 'antigen': 0.0,\n",
       " 'pathogen': 0.0,\n",
       " 'process': 0.0,\n",
       " 'call': 0.0,\n",
       " 'cross': 0.0,\n",
       " 'present': 0.0,\n",
       " 'lead': 0.0,\n",
       " 'jose': 0.0,\n",
       " 'villadango': 0.0,\n",
       " 'immunologist': 0.0,\n",
       " 'walter': 0.0,\n",
       " 'eliza': 0.0,\n",
       " 'medic': 0.0,\n",
       " 'melbourn': 0.0,\n",
       " 'sepsi': 0.0,\n",
       " 'bacteri': 0.0,\n",
       " 'overstimul': 0.0,\n",
       " 'thi': 0.0,\n",
       " 'result': 0.0,\n",
       " 'critic': 0.0,\n",
       " 'alarm': 0.0,\n",
       " 'shut': 0.0,\n",
       " 'occur': 0.0,\n",
       " 'local': 0.0,\n",
       " 'involv': 0.0,\n",
       " 'throughout': 0.0,\n",
       " 'concentr': 0.0,\n",
       " 'respond': 0.0,\n",
       " 'a': 0.0,\n",
       " 'type': 0.0,\n",
       " 'manag': 0.0,\n",
       " 'mice': 0.0,\n",
       " 'compromis': 0.0,\n",
       " 'inject': 0.0,\n",
       " 'live': 0.0,\n",
       " 'made': 0.0,\n",
       " 'expos': 0.0,\n",
       " 'secondari': 0.0,\n",
       " 'laboratori': 0.0,\n",
       " 'subsequ': 0.0,\n",
       " 'find': 0.0,\n",
       " 'show': 0.0,\n",
       " 'miss': 0.0,\n",
       " 'link': 0.0,\n",
       " 'immunosupress': 0.0,\n",
       " 'anim': 0.0,\n",
       " 'capac': 0.0,\n",
       " 'virus': 0.0,\n",
       " 'we': 0.0,\n",
       " 'abl': 0.0,\n",
       " 'back': 0.0,\n",
       " 'strategi': 0.0,\n",
       " 'alreadi': 0.0,\n",
       " 'use': 0.0,\n",
       " 'anti': 0.0,\n",
       " 'cancer': 0.0,\n",
       " 'yet': 0.0,\n",
       " 'analys': 0.0,\n",
       " 'capsul': 0.0,\n",
       " 'clue': 0.0,\n",
       " 'solar': 0.0,\n",
       " 'began': 0.0,\n",
       " 'seven': 0.0,\n",
       " 'year': 0.0,\n",
       " 'quest': 0.0,\n",
       " 'return': 0.0,\n",
       " 'earth': 0.0,\n",
       " 'pristin': 0.0,\n",
       " 'sampl': 0.0,\n",
       " 'format': 0.0,\n",
       " 'end': 0.0,\n",
       " 'triumph': 0.0,\n",
       " 'dark': 0.0,\n",
       " 'wet': 0.0,\n",
       " 'utah': 0.0,\n",
       " 'desert': 0.0,\n",
       " 'weekend': 0.0,\n",
       " 'it': 0.0,\n",
       " 'hard': 0.0,\n",
       " 'describ': 0.0,\n",
       " 'feel': 0.0,\n",
       " 'point': 0.0,\n",
       " 'mission': 0.0,\n",
       " 'washington': 0.0,\n",
       " 'stardust': 0.0,\n",
       " 'professor': 0.0,\n",
       " 'don': 0.0,\n",
       " 'brownle': 0.0,\n",
       " 'said': 0.0,\n",
       " 'shortli': 0.0,\n",
       " 'probe': 0.0,\n",
       " 'touchdown': 0.0,\n",
       " 'utc': 0.0,\n",
       " 'aedt': 0.0,\n",
       " 'sunday': 0.0,\n",
       " 'travel': 0.0,\n",
       " 'almost': 0.0,\n",
       " 'billion': 0.0,\n",
       " 'mile': 0.0,\n",
       " 'space': 0.0,\n",
       " 'visit': 0.0,\n",
       " 'comet': 0.0,\n",
       " 'grab': 0.0,\n",
       " 'piec': 0.0,\n",
       " 'land': 0.0,\n",
       " 'morn': 0.0,\n",
       " 'incred': 0.0,\n",
       " 'thrill': 0.0,\n",
       " 'releas': 0.0,\n",
       " 'preciou': 0.0,\n",
       " 'cargo': 0.0,\n",
       " 'high': 0.0,\n",
       " 'speed': 0.0,\n",
       " 'flybi': 0.0,\n",
       " 'send': 0.0,\n",
       " 'tyre': 0.0,\n",
       " 'size': 0.0,\n",
       " 'canist': 0.0,\n",
       " 'soar': 0.0,\n",
       " 'blaze': 0.0,\n",
       " 'kilometr': 0.0,\n",
       " 'per': 0.0,\n",
       " 'hour': 0.0,\n",
       " 'smash': 0.0,\n",
       " 'planet': 0.0,\n",
       " 'atmospher': 0.0,\n",
       " 'fastest': 0.0,\n",
       " 'object': 0.0,\n",
       " 'top': 0.0,\n",
       " 'record': 0.0,\n",
       " 'set': 0.0,\n",
       " 'apollo': 0.0,\n",
       " 'command': 0.0,\n",
       " 'modul': 0.0,\n",
       " 'as': 0.0,\n",
       " 'heat': 0.0,\n",
       " 'shield': 0.0,\n",
       " 'fri': 0.0,\n",
       " 'away': 0.0,\n",
       " 'plume': 0.0,\n",
       " 'form': 0.0,\n",
       " 'wake': 0.0,\n",
       " 'look': 0.0,\n",
       " 'torch': 0.0,\n",
       " 'visibl': 0.0,\n",
       " 'part': 0.0,\n",
       " 'western': 0.0,\n",
       " 'quit': 0.0,\n",
       " 'sure': 0.0,\n",
       " 'bright': 0.0,\n",
       " 'go': 0.0,\n",
       " 'think': 0.0,\n",
       " 'see': 0.0,\n",
       " 'anyth': 0.0,\n",
       " 'slip': 0.0,\n",
       " 'outsid': 0.0,\n",
       " 'member': 0.0,\n",
       " 'spot': 0.0,\n",
       " 'streak': 0.0,\n",
       " 'across': 0.0,\n",
       " 'sky': 0.0,\n",
       " 'after': 0.0,\n",
       " 'scan': 0.0,\n",
       " 'final': 0.0,\n",
       " 'mar': 0.0,\n",
       " 'clearli': 0.0,\n",
       " 'time': 0.0,\n",
       " 'twinkl': 0.0,\n",
       " 'littl': 0.0,\n",
       " 'bit': 0.0,\n",
       " 'brighter': 0.0,\n",
       " 'move': 0.0,\n",
       " 'i': 0.0,\n",
       " 'thought': 0.0,\n",
       " 'mayb': 0.0,\n",
       " 'helicopt': 0.0,\n",
       " 'kept': 0.0,\n",
       " 'reddish': 0.0,\n",
       " 'colour': 0.0,\n",
       " 'trail': 0.0,\n",
       " 'glow': 0.0,\n",
       " 'behind': 0.0,\n",
       " 'iron': 0.0,\n",
       " 'collect': 0.0,\n",
       " 'spacedust': 0.0,\n",
       " 'dure': 0.0,\n",
       " 'pass': 0.0,\n",
       " 'wild': 0.0,\n",
       " 'vilt': 0.0,\n",
       " 'extend': 0.0,\n",
       " 'gel': 0.0,\n",
       " 'devic': 0.0,\n",
       " 'trap': 0.0,\n",
       " 'particl': 0.0,\n",
       " 'also': 0.0,\n",
       " 'interstellar': 0.0,\n",
       " 'dust': 0.0,\n",
       " 'believ': 0.0,\n",
       " 'contain': 0.0,\n",
       " 'unalt': 0.0,\n",
       " 'materi': 0.0,\n",
       " 'primit': 0.0,\n",
       " 'could': 0.0,\n",
       " 'went': 0.0,\n",
       " 'edg': 0.0,\n",
       " 'far': 0.0,\n",
       " 'sun': 0.0,\n",
       " 'cold': 0.0,\n",
       " 'condit': 0.0,\n",
       " 'confid': 0.0,\n",
       " 'initi': 0.0,\n",
       " 'build': 0.0,\n",
       " 'block': 0.0,\n",
       " 'retriev': 0.0,\n",
       " 'crew': 0.0,\n",
       " 'taken': 0.0,\n",
       " 'clean': 0.0,\n",
       " 'room': 0.0,\n",
       " 'test': 0.0,\n",
       " 'train': 0.0,\n",
       " 'rang': 0.0,\n",
       " 'inspect': 0.0,\n",
       " 'schedul': 0.0,\n",
       " 'remov': 0.0,\n",
       " 'flown': 0.0,\n",
       " 'nasa': 0.0,\n",
       " 'center': 0.0,\n",
       " 'houston': 0.0,\n",
       " 'tuesday': 0.0,\n",
       " 'there': 0.0,\n",
       " 'catalogu': 0.0,\n",
       " 'begin': 0.0,\n",
       " 'distribut': 0.0,\n",
       " 'analysi': 0.0,\n",
       " 'gotcha': 0.0,\n",
       " 'strike': 0.0,\n",
       " 'imag': 0.0,\n",
       " 'spark': 0.0,\n",
       " 'interest': 0.0,\n",
       " 'internet': 0.0,\n",
       " 'user': 0.0,\n",
       " 'twentieth': 0.0,\n",
       " 'decid': 0.0,\n",
       " 'whether': 0.0,\n",
       " 'websit': 0.0,\n",
       " 'visual': 0.0,\n",
       " 'appeal': 0.0,\n",
       " 'assess': 0.0,\n",
       " 'within': 0.0,\n",
       " 'millisecond': 0.0,\n",
       " 'suggest': 0.0,\n",
       " 'web': 0.0,\n",
       " 'design': 0.0,\n",
       " 'good': 0.0,\n",
       " 'impress': 0.0,\n",
       " 'canadian': 0.0,\n",
       " 'report': 0.0,\n",
       " 'march': 0.0,\n",
       " 'april': 0.0,\n",
       " 'behaviour': 0.0,\n",
       " 'inform': 0.0,\n",
       " 'technolog': 0.0,\n",
       " 'gitt': 0.0,\n",
       " 'lindgaard': 0.0,\n",
       " 'carleton': 0.0,\n",
       " 'ottawa': 0.0,\n",
       " 'confirm': 0.0,\n",
       " 'fickl': 0.0,\n",
       " 'lot': 0.0,\n",
       " 'flash': 0.0,\n",
       " 'ask': 0.0,\n",
       " 'particip': 0.0,\n",
       " 'rate': 0.0,\n",
       " 'when': 0.0,\n",
       " 'repeat': 0.0,\n",
       " 'exercis': 0.0,\n",
       " 'longer': 0.0,\n",
       " 'view': 0.0,\n",
       " 'period': 0.0,\n",
       " 'consist': 0.0,\n",
       " 'medium': 0.0,\n",
       " 'messag': 0.0,\n",
       " 'come': 0.0,\n",
       " 'news': 0.0,\n",
       " 'anyon': 0.0,\n",
       " 'hope': 0.0,\n",
       " 'convey': 0.0,\n",
       " 'sue': 0.0,\n",
       " 'burgess': 0.0,\n",
       " 'evalu': 0.0,\n",
       " 'useabl': 0.0,\n",
       " 'senior': 0.0,\n",
       " 'lectur': 0.0,\n",
       " 'doubt': 0.0,\n",
       " 'quickli': 0.0,\n",
       " 'stay': 0.0,\n",
       " 'usual': 0.0,\n",
       " 'tie': 0.0,\n",
       " 'movement': 0.0,\n",
       " 'interact': 0.0,\n",
       " 'way': 0.0,\n",
       " 'structur': 0.0,\n",
       " 'unclear': 0.0,\n",
       " 'chang': 0.0,\n",
       " 'abil': 0.0,\n",
       " 'adapt': 0.0,\n",
       " 'alway': 0.0,\n",
       " 'click': 0.0,\n",
       " 'around': 0.0,\n",
       " 'halo': 0.0,\n",
       " 'associ': 0.0,\n",
       " 'psycholog': 0.0,\n",
       " 'bill': 0.0,\n",
       " 'von': 0.0,\n",
       " 'hippel': 0.0,\n",
       " 'south': 0.0,\n",
       " 'wale': 0.0,\n",
       " 'read': 0.0,\n",
       " 'word': 0.0,\n",
       " 'stunningli': 0.0,\n",
       " 'remark': 0.0,\n",
       " 'timefram': 0.0,\n",
       " 'complex': 0.0,\n",
       " 'stimuli': 0.0,\n",
       " 'fast': 0.0,\n",
       " 'hold': 0.0,\n",
       " 'later': 0.0,\n",
       " 'judgement': 0.0,\n",
       " 'emot': 0.0,\n",
       " 'independ': 0.0,\n",
       " 'cognit': 0.0,\n",
       " 'evolutionari': 0.0,\n",
       " 'rapidli': 0.0,\n",
       " 'danger': 0.0,\n",
       " 'situat': 0.0,\n",
       " 'reflect': 0.0,\n",
       " 'bia': 0.0,\n",
       " 'toward': 0.0,\n",
       " 'someth': 0.0,\n",
       " 'drive': 0.0,\n",
       " 'quick': 0.0,\n",
       " 'sort': 0.0,\n",
       " 'reaction': 0.0,\n",
       " 'consid': 0.0,\n",
       " 'still': 0.0,\n",
       " 'jewish': 0.0,\n",
       " 'histori': 0.0,\n",
       " 'migrat': 0.0,\n",
       " 'dna': 0.0,\n",
       " 'four': 0.0,\n",
       " 'mother': 0.0,\n",
       " 'ago': 0.0,\n",
       " 'europ': 0.0,\n",
       " 'ancestor': 0.0,\n",
       " 'ashkenazi': 0.0,\n",
       " 'jew': 0.0,\n",
       " 'aliv': 0.0,\n",
       " 'intern': 0.0,\n",
       " 'paint': 0.0,\n",
       " 'vivid': 0.0,\n",
       " 'pictur': 0.0,\n",
       " 'evolut': 0.0,\n",
       " 'surviv': 0.0,\n",
       " 'correl': 0.0,\n",
       " 'well': 0.0,\n",
       " 'establish': 0.0,\n",
       " 'written': 0.0,\n",
       " 'oral': 0.0,\n",
       " 'doron': 0.0,\n",
       " 'behar': 0.0,\n",
       " 'technion': 0.0,\n",
       " 'israel': 0.0,\n",
       " 'american': 0.0,\n",
       " 'million': 0.0,\n",
       " 'descend': 0.0,\n",
       " 'women': 0.0,\n",
       " 'for': 0.0,\n",
       " 'popul': 0.0,\n",
       " 'subject': 0.0,\n",
       " 'birthplac': 0.0,\n",
       " 'grandmoth': 0.0,\n",
       " 'great': 0.0,\n",
       " 'write': 0.0,\n",
       " 'mitochondri': 0.0,\n",
       " 'nucleu': 0.0,\n",
       " 'carri': 0.0,\n",
       " 'instruct': 0.0,\n",
       " 'virtual': 0.0,\n",
       " 'unchang': 0.0,\n",
       " 'daughter': 0.0,\n",
       " 'occasion': 0.0,\n",
       " 'mutat': 0.0,\n",
       " 'known': 0.0,\n",
       " 'molecular': 0.0,\n",
       " 'clock': 0.0,\n",
       " 'track': 0.0,\n",
       " 'instanc': 0.0,\n",
       " 'comput': 0.0,\n",
       " 'ancestr': 0.0,\n",
       " 'eve': 0.0,\n",
       " 'africa': 0.0,\n",
       " 'now': 0.0,\n",
       " 'kind': 0.0,\n",
       " 'pool': 0.0,\n",
       " 'near': 0.0,\n",
       " 'east': 0.0,\n",
       " 'among': 0.0,\n",
       " 'matern': 0.0,\n",
       " 'lineag': 0.0,\n",
       " 'real': 0.0,\n",
       " 'exact': 0.0,\n",
       " 'specif': 0.0,\n",
       " 'marker': 0.0,\n",
       " 'settl': 0.0,\n",
       " 'direct': 0.0,\n",
       " 'then': 0.0,\n",
       " 'certain': 0.0,\n",
       " 'probabl': 0.0,\n",
       " 'th': 0.0,\n",
       " 'centuri': 0.0,\n",
       " 'simpli': 0.0,\n",
       " 'demograph': 0.0,\n",
       " 'matter': 0.0,\n",
       " 'start': 0.0,\n",
       " 'expand': 0.0,\n",
       " 'dramat': 0.0,\n",
       " 'tradit': 0.0,\n",
       " 'famili': 0.0,\n",
       " 'might': 0.0,\n",
       " 'characteris': 0.0,\n",
       " 'number': 0.0,\n",
       " 'gave': 0.0,\n",
       " 'rise': 0.0,\n",
       " 'european': 0.0,\n",
       " 'explod': 0.0,\n",
       " 'prior': 0.0,\n",
       " 'world': 0.0,\n",
       " 'war': 0.0,\n",
       " 'ii': 0.0,\n",
       " 'nazi': 0.0,\n",
       " 'alli': 0.0,\n",
       " 'kill': 0.0,\n",
       " 'estim': 0.0,\n",
       " 'defin': 0.0,\n",
       " 'common': 0.0,\n",
       " 'northern': 0.0,\n",
       " 'central': 0.0,\n",
       " 'ancestri': 0.0,\n",
       " 'cultur': 0.0,\n",
       " 'yiddish': 0.0,\n",
       " 'languag': 0.0,\n",
       " 'commun': 0.0,\n",
       " 'pop': 0.0,\n",
       " 'non': 0.0,\n",
       " 'share': 0.0,\n",
       " 'origin': 0.0,\n",
       " 'particular': 0.0,\n",
       " 'provid': 0.0,\n",
       " 'explan': 0.0,\n",
       " 'inherit': 0.0,\n",
       " 'disproportion': 0.0,\n",
       " 'descent': 0.0,\n",
       " 'breast': 0.0,\n",
       " 'colon': 0.0,\n",
       " 'caus': 0.0,\n",
       " 'nuclear': 0.0,\n",
       " 'group': 0.0,\n",
       " 'these': 0.0,\n",
       " 'gene': 0.0,\n",
       " 'date': 0.0,\n",
       " 'bottleneck': 0.0,\n",
       " 'phenomenon': 0.0,\n",
       " 'squeez': 0.0,\n",
       " 'larg': 0.0,\n",
       " 'small': 0.0,\n",
       " 'skorecki': 0.0,\n",
       " 'founder': 0.0,\n",
       " 'huge': 0.0,\n",
       " 'gain': 0.0,\n",
       " 'insight': 0.0,\n",
       " 'secret': 0.0,\n",
       " 'life': 0.0,\n",
       " 'cloud': 0.0,\n",
       " 'biggest': 0.0,\n",
       " 'climat': 0.0,\n",
       " 'experi': 0.0,\n",
       " 'storm': 0.0,\n",
       " 'chaser': 0.0,\n",
       " 'gather': 0.0,\n",
       " 'weather': 0.0,\n",
       " 'balloon': 0.0,\n",
       " 'next': 0.0,\n",
       " 'month': 0.0,\n",
       " 'largest': 0.0,\n",
       " 'meteorolog': 0.0,\n",
       " 'tropic': 0.0,\n",
       " 'warm': 0.0,\n",
       " 'darwin': 0.0,\n",
       " 'comprehens': 0.0,\n",
       " 'data': 0.0,\n",
       " 'thundercloud': 0.0,\n",
       " 'peter': 0.0,\n",
       " 'princip': 0.0,\n",
       " 'bureau': 0.0,\n",
       " 'give': 0.0,\n",
       " 'model': 0.0,\n",
       " 'accur': 0.0,\n",
       " 'short': 0.0,\n",
       " 'forecast': 0.0,\n",
       " 'predict': 0.0,\n",
       " 'uncertainti': 0.0,\n",
       " 'greenhous': 0.0,\n",
       " 'calcul': 0.0,\n",
       " 'understand': 0.0,\n",
       " 'thunderstorm': 0.0,\n",
       " 'underli': 0.0,\n",
       " 'aim': 0.0,\n",
       " 'increas': 0.0,\n",
       " 'evolv': 0.0,\n",
       " 'scale': 0.0,\n",
       " 'impact': 0.0,\n",
       " 'collabor': 0.0,\n",
       " 'project': 0.0,\n",
       " 'depart': 0.0,\n",
       " 'energi': 0.0,\n",
       " 'radiat': 0.0,\n",
       " 'measur': 0.0,\n",
       " 'program': 0.0,\n",
       " 'csiro': 0.0,\n",
       " 'japan': 0.0,\n",
       " 'ice': 0.0,\n",
       " 'crystal': 0.0,\n",
       " 'devast': 0.0,\n",
       " 'cyclon': 0.0,\n",
       " 'traci': 0.0,\n",
       " 'watcher': 0.0,\n",
       " 'and': 0.0,\n",
       " 'monsoon': 0.0,\n",
       " 'season': 0.0,\n",
       " 'multi': 0.0,\n",
       " 'dollar': 0.0,\n",
       " 'three': 0.0,\n",
       " 'prong': 0.0,\n",
       " 'plane': 0.0,\n",
       " 'fli': 0.0,\n",
       " 'directli': 0.0,\n",
       " 'ground': 0.0,\n",
       " 'network': 0.0,\n",
       " 'radar': 0.0,\n",
       " 'laser': 0.0,\n",
       " 'infrar': 0.0,\n",
       " 'characterist': 0.0,\n",
       " 'moistur': 0.0,\n",
       " 'light': 0.0,\n",
       " 'everi': 0.0,\n",
       " 'day': 0.0,\n",
       " 'five': 0.0,\n",
       " 'site': 0.0,\n",
       " 'resourc': 0.0,\n",
       " 'dozen': 0.0,\n",
       " 'hundr': 0.0,\n",
       " 'paper': 0.0,\n",
       " 'global': 0.0,\n",
       " 'pattern': 0.0,\n",
       " 'unlik': 0.0,\n",
       " 'isol': 0.0,\n",
       " 'event': 0.0,\n",
       " 'hurrican': 0.0,\n",
       " 'katrina': 0.0,\n",
       " 'struck': 0.0,\n",
       " 'orlean': 0.0,\n",
       " 'last': 0.0,\n",
       " 'question': 0.0,\n",
       " 'due': 0.0,\n",
       " 'load': 0.0,\n",
       " 'dice': 0.0,\n",
       " 'expect': 0.0,\n",
       " 'intens': 0.0,\n",
       " 'artist': 0.0,\n",
       " 'pluto': 0.0,\n",
       " 'know': 0.0,\n",
       " 'horizon': 0.0,\n",
       " 'spacecraft': 0.0,\n",
       " 'launch': 0.0,\n",
       " 'reach': 0.0,\n",
       " 'distant': 0.0,\n",
       " 'decad': 0.0,\n",
       " 'agenc': 0.0,\n",
       " 'shed': 0.0,\n",
       " 'wednesday': 0.0,\n",
       " 'investig': 0.0,\n",
       " 'class': 0.0,\n",
       " 'dwarf': 0.0,\n",
       " 'exist': 0.0,\n",
       " 'beyond': 0.0,\n",
       " 'orbit': 0.0,\n",
       " 'neptun': 0.0,\n",
       " 'over': 0.0,\n",
       " 'close': 0.0,\n",
       " 'sister': 0.0,\n",
       " 'remain': 0.0,\n",
       " 'shroud': 0.0,\n",
       " 'mysteri': 0.0,\n",
       " 'fuzzi': 0.0,\n",
       " 'telescop': 0.0,\n",
       " 'ponder': 0.0,\n",
       " 'what': 0.0,\n",
       " 'postag': 0.0,\n",
       " 'stamp': 0.0,\n",
       " 'colleen': 0.0,\n",
       " 'hartman': 0.0,\n",
       " 'deputi': 0.0,\n",
       " 'administr': 0.0,\n",
       " 'scienc': 0.0,\n",
       " 'alan': 0.0,\n",
       " 'stern': 0.0,\n",
       " 'southwest': 0.0,\n",
       " 'boulder': 0.0,\n",
       " 'colorado': 0.0,\n",
       " 'agre': 0.0,\n",
       " 'learn': 0.0,\n",
       " 'grow': 0.0,\n",
       " 'planethood': 0.0,\n",
       " 'stop': 0.0,\n",
       " 'growth': 0.0,\n",
       " 'mid': 0.0,\n",
       " 'gestat': 0.0,\n",
       " 'left': 0.0,\n",
       " 'fossil': 0.0,\n",
       " 'relic': 0.0,\n",
       " 'kuiper': 0.0,\n",
       " 'belt': 0.0,\n",
       " 'region': 0.0,\n",
       " 'planetari': 0.0,\n",
       " 'how': 0.0,\n",
       " 'u': 0.0,\n",
       " 'attempt': 0.0,\n",
       " 'piano': 0.0,\n",
       " 'instrument': 0.0,\n",
       " 'map': 0.0,\n",
       " 'moon': 0.0,\n",
       " 'charon': 0.0,\n",
       " 'recent': 0.0,\n",
       " 'smaller': 0.0,\n",
       " 'to': 0.0,\n",
       " 'shave': 0.0,\n",
       " 'journey': 0.0,\n",
       " 'nine': 0.0,\n",
       " 'aboard': 0.0,\n",
       " 'massiv': 0.0,\n",
       " 'rocket': 0.0,\n",
       " 'typic': 0.0,\n",
       " 'satellit': 0.0,\n",
       " 'heftier': 0.0,\n",
       " 'half': 0.0,\n",
       " 'tonn': 0.0,\n",
       " 'vehicl': 0.0,\n",
       " 'atla': 0.0,\n",
       " 'upper': 0.0,\n",
       " 'stage': 0.0,\n",
       " 'motor': 0.0,\n",
       " 'unpreced': 0.0,\n",
       " 'solid': 0.0,\n",
       " 'strap': 0.0,\n",
       " 'booster': 0.0,\n",
       " 'extra': 0.0,\n",
       " 'punch': 0.0,\n",
       " 'success': 0.0,\n",
       " 'six': 0.0,\n",
       " 'previou': 0.0,\n",
       " 'roll': 0.0,\n",
       " 'hangar': 0.0,\n",
       " 'cape': 0.0,\n",
       " 'canaver': 0.0,\n",
       " 'pad': 0.0,\n",
       " 'earlier': 0.0,\n",
       " 'even': 0.0,\n",
       " 'boost': 0.0,\n",
       " 'need': 0.0,\n",
       " 'bounc': 0.0,\n",
       " 'jupit': 0.0,\n",
       " 'graviti': 0.0,\n",
       " 'field': 0.0,\n",
       " 'nearli': 0.0,\n",
       " 'hitch': 0.0,\n",
       " 'ride': 0.0,\n",
       " 'if': 0.0,\n",
       " 'cannot': 0.0,\n",
       " 'feburari': 0.0,\n",
       " 'opportun': 0.0,\n",
       " 'slingshot': 0.0,\n",
       " 'addit': 0.0,\n",
       " 'depend': 0.0,\n",
       " 'actual': 0.0,\n",
       " 'wait': 0.0,\n",
       " 'approv': 0.0,\n",
       " 'fund': 0.0,\n",
       " 'happi': 0.0,\n",
       " 'place': 0.0,\n",
       " 'oper': 0.0,\n",
       " 'vastli': 0.0,\n",
       " 'differ': 0.0,\n",
       " 'temperatur': 0.0,\n",
       " 'familiar': 0.0,\n",
       " 'co': 0.0,\n",
       " 'richard': 0.0,\n",
       " 'binzel': 0.0,\n",
       " 'massachusett': 0.0,\n",
       " 'pois': 0.0,\n",
       " 'explor': 0.0,\n",
       " 'add': 0.0,\n",
       " 'dale': 0.0,\n",
       " 'cruikshank': 0.0,\n",
       " 'ame': 0.0,\n",
       " 'california': 0.0,\n",
       " 'backpack': 0.0,\n",
       " 'highli': 0.0,\n",
       " 'mobil': 0.0,\n",
       " 'especi': 0.0,\n",
       " 'ring': 0.0,\n",
       " 'home': 0.0,\n",
       " 'cost': 0.0,\n",
       " 'bu': 0.0,\n",
       " 'ticket': 0.0,\n",
       " 'bondi': 0.0,\n",
       " 'beach': 0.0,\n",
       " 'digit': 0.0,\n",
       " 'buddi': 0.0,\n",
       " 'act': 0.0,\n",
       " 'guid': 0.0,\n",
       " 'jeff': 0.0,\n",
       " 'axup': 0.0,\n",
       " 'phd': 0.0,\n",
       " 'student': 0.0,\n",
       " 'queensland': 0.0,\n",
       " 'former': 0.0,\n",
       " 'focus': 0.0,\n",
       " 'applic': 0.0,\n",
       " 'shadow': 0.0,\n",
       " 'document': 0.0,\n",
       " 'tour': 0.0,\n",
       " 'brisban': 0.0,\n",
       " 'took': 0.0,\n",
       " 'boat': 0.0,\n",
       " 'cruis': 0.0,\n",
       " 'park': 0.0,\n",
       " 'eprintsuq': 0.0,\n",
       " 'submit': 0.0,\n",
       " 'public': 0.0,\n",
       " 'latest': 0.0,\n",
       " 'prototyp': 0.0,\n",
       " 'australasian': 0.0,\n",
       " 'crc': 0.0,\n",
       " 'product': 0.0,\n",
       " 'softwar': 0.0,\n",
       " 'download': 0.0,\n",
       " 'onto': 0.0,\n",
       " 'phone': 0.0,\n",
       " 'instant': 0.0,\n",
       " 'discuss': 0.0,\n",
       " 'forum': 0.0,\n",
       " 'wireless': 0.0,\n",
       " 'symbol': 0.0,\n",
       " 'increasingli': 0.0,\n",
       " 'brought': 0.0,\n",
       " 'transport': 0.0,\n",
       " 'breakdown': 0.0,\n",
       " 'neighbourhood': 0.0,\n",
       " 'glocalis': 0.0,\n",
       " 'gone': 0.0,\n",
       " 'text': 0.0,\n",
       " 'ani': 0.0,\n",
       " 'introduc': 0.0,\n",
       " 'enter': 0.0,\n",
       " 'social': 0.0,\n",
       " 'environ': 0.0,\n",
       " 'cours': 0.0,\n",
       " 'never': 0.0,\n",
       " 'watch': 0.0,\n",
       " 'comedi': 0.0,\n",
       " 'aerob': 0.0,\n",
       " 'film': 0.0,\n",
       " 'flow': 0.0,\n",
       " 'sad': 0.0,\n",
       " 'declin': 0.0,\n",
       " 'maryland': 0.0,\n",
       " 'movi': 0.0,\n",
       " 'segment': 0.0,\n",
       " 'seriou': 0.0,\n",
       " 'least': 0.0,\n",
       " 'session': 0.0,\n",
       " 'with': 0.0,\n",
       " 'ultrasound': 0.0,\n",
       " 'brachial': 0.0,\n",
       " 'arteri': 0.0,\n",
       " 'vessel': 0.0,\n",
       " 'run': 0.0,\n",
       " 'shoulder': 0.0,\n",
       " 'elbow': 0.0,\n",
       " 'indic': 0.0,\n",
       " 'extract': 0.0,\n",
       " 'distress': 0.0,\n",
       " ...}"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tfidf_science"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}