explore-data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0101ab30",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pathlib import Path\n",
    "\n",
    "data_path = Path('data/train_val_test-WITH_POSTS/train_anonymized-WITH_POSTS.json')\n",
    "data_path.exists()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f3dbcba4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "with open(data_path) as f:\n",
    "    data = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "87dd64a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import chain, combinations\n",
    "\n",
    "def iter_json_sample(data_):\n",
    "    for sample in data_.values():\n",
    "        annos = chain(*sample['Annotations'].values())\n",
    "        emos = {anno['Emotion'] for anno in annos if anno['Emotion'] != 'NA'}\n",
    "        for emo1, emo2 in combinations(emos, 2):\n",
    "            yield emo1, emo2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "6211f6ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['anger', 'disgust', 'fear', 'joy', 'sadness', 'trust', 'anticipation']"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from data_preprocess import EMO_LIST\n",
    "\n",
    "EMO_LIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "0995271d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'anger': 0,\n",
       " 'disgust': 1,\n",
       " 'fear': 2,\n",
       " 'joy': 3,\n",
       " 'sadness': 4,\n",
       " 'trust': 5,\n",
       " 'anticipation': 6}"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_emos = len(EMO_LIST)\n",
    "emo2idx = dict(zip(EMO_LIST, range(n_emos)))\n",
    "emo2idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "69a5d83a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  0., 146., 263.,  30., 198.,  19., 300.],\n",
       "       [146.,   0., 105.,  13.,  78.,  12., 112.],\n",
       "       [263., 105.,   0.,  52., 227.,  67., 550.],\n",
       "       [ 30.,  13.,  52.,   0.,  27.,  22.,  80.],\n",
       "       [198.,  78., 227.,  27.,   0.,  21., 220.],\n",
       "       [ 19.,  12.,  67.,  22.,  21.,   0.,  74.],\n",
       "       [300., 112., 550.,  80., 220.,  74.,   0.]])"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "emo_combs = np.zeros([n_emos, n_emos])\n",
    "\n",
    "for emos in iter_json_sample(data):\n",
    "    i, j = map(emo2idx.get, emos)\n",
    "    emo_combs[i][j] += 1\n",
    "    emo_combs[j][i] += 1\n",
    "\n",
    "emo_combs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "c1ba92f9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "ax.imshow(emo_combs, cmap='hot')\n",
    "ax.set_xticks(np.arange(n_emos), labels=EMO_LIST)\n",
    "ax.set_yticks(np.arange(n_emos), labels=EMO_LIST)\n",
    "\n",
    "plt.setp(ax.get_xticklabels(), rotation=45, ha='right', rotation_mode='anchor')\n",
    "\n",
    "for i in range(n_emos):\n",
    "    for j in range(n_emos):\n",
    "        ax.text(j, i, emo_combs[i, j], ha='center', va='center', color='c')\n",
    "\n",
    "fig.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "e7fed918",
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import chain\n",
    "\n",
    "def iter_json_data(data_):\n",
    "    for sample in data_.values():\n",
    "        annos = chain(*sample['Annotations'].values())\n",
    "        emos = {anno['Emotion'] for anno in annos if anno['Emotion'] != 'NA'}\n",
    "        for emo in emos:\n",
    "            yield emo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "027d3e95",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'anger': 470,\n",
       " 'disgust': 192,\n",
       " 'fear': 765,\n",
       " 'joy': 134,\n",
       " 'sadness': 360,\n",
       " 'trust': 99,\n",
       " 'anticipation': 873}"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "emo_freq = dict.fromkeys(EMO_LIST, 0)\n",
    "\n",
    "for emo in iter_json_data(data):\n",
    "    emo_freq[emo] += 1\n",
    "\n",
    "emo_freq"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "31b86086",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict = {'input': [], 'summary': []}\n",
    "instr = 'Generate a summary of what triggered {} in this post: {}'\n",
    "emo_cnt = dict.fromkeys(EMO_LIST, 0)\n",
    "emo_list = set(EMO_LIST)\n",
    "MAX_SIZE = min(emo_freq.values())\n",
    "\n",
    "for sample in data.values():\n",
    "    annos = chain(*sample['Annotations'].values())\n",
    "    annos = list(filter(lambda a: a['Emotion'] in emo_list, annos))\n",
    "    anno = min(annos, key=lambda a: emo_freq[a['Emotion']], default=None)\n",
    "    if anno is not None:\n",
    "        emo = anno['Emotion']\n",
    "        data_dict['input'].append(instr.format(emo, sample['Reddit Post']))\n",
    "        data_dict['summary'].append(anno['Abstractive'])\n",
    "        emo_cnt[emo] += 1\n",
    "        if emo_cnt[emo] == MAX_SIZE:\n",
    "            emo_list.remove(emo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "1fe650d3",
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generate a summary of what triggered sadness in this post: In my area we have a super high vaccination rate. In the sf Metro area we have 65 fully vaccinated, and some parts of it like sf have 68.9 fully vaccinated of whole population. And yet the delta is still surging here. The cdc just said in areas with high transmission masks should be mandated again and I feel completely hopeless. It's so far unknown if the bay area will reimplement masks but I'm sure they will. It's been close too 2 weeks since LA reinstated masks and the cases are still exploding there which is pretty hopeless. I can just see another lockdown coming maybe in the winter.\n",
      "[SUMMARY]The person feels defeated and without expectations of having to bear wearing masks again.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered anticipation in this post: Basically I looked at the Bloomberg tracker and saw that the USA will take another *8 months* to reach herd while Canada will only take four more *weeks*. I have a super close friend who could write a letter of recommendation for an extended visa, and a job that could get me priority (though not sure if my USA certification would carry over), but I dont have a passport. Pros of this plan: Id be living somewhere where my mental health could finally recover My career should carry over for work, as my practice is also a chain Two different high priority check marks for the visa I could possibly be roommates w/my BFF for that time Cons: Ive never lived alone before and my family definitely won be able to uproot their lives to follow My current job is in a town I know/really like with people I really like, by the time I come back the position might not be available there/they my be soured on me from bailing The headaches of packing my life up into my car and driving that far, once again alone &amp;#x200B; Basically I need either talked down from this idea or given assurance/tips to pull it off. Thank you for reading this far.\n",
      "[SUMMARY]I'm looking for any advice about whether I should give up my life in the U.S. and move to Canada because they'll reach herd immunity faster.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered joy in this post: I was feeling hopeful for a while. For a month, cases were dropping and I was feeling like we were gonna get out of this. As soon as I started feeling decent and like things were going good, cases start rising dramatically and I later find out about delta. This crushed me. I especially felt dejected when we started going backwards again on the guidance and now even if vaccinated we have to wear masks. It makes me want to look into how to invent a time machine, override paradoxes and go back to when my parents were kids, or when I was 12 in 2004. Life was much better at those times and our generation would not have gotten duped. I feel like rubbish again.\n",
      "[SUMMARY]I'm happy because covid cases are dropping\n",
      "\n",
      "\n",
      "Generate a summary of what triggered anger in this post: It's been so nice not to have these itchy t-shirt tags on our faces at all times the last few months, now they're back just in time for the hottest part of summer in the hottest part of the country because hardly anyone where I live has gotten vaccinated. I'm incredibly annoyed and feel like Covid is never, ever going to end, it'll just be in and out of lockdowns, masking, and no concerts for the rest of our lives.\n",
      "[SUMMARY]I thought covid was far gone, but it got me angry knowing for fact that covid is back.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered trust in this post: I got fully vaccinated the first moment I could. Ive been so anxious about covid from the beginning of the pandemic, as a lot of people have been. My boyfriend is unvaccinated - hes not anti-vax or anything, just a little bit wary of a vaccine thats so new. I really want him to get vaccinated with the new delta variant going around and covid cases going up again in our area. Hes not in the at-risk group, but Ive still had nightmares about the what-if scenarios. When Ive talked to him he reiterates that hes unsure about how new the covid vaccine is. I understand that completely, but personally, Id rather be vaccinated especially with the lack of masks in our area. I dont want to push the subject too much as I know getting vaccinated is a personal decision, but I still dont sleep well at night knowing hes going to work unmasked, unvaccinated, and interacting with people. Maybe I just need to accept the fact he doesnt want to get the vaccine and find ways to combat my anxiety surrounding his vaccination - but Id still like to know if anyone has suggestions on how to talk to loved ones who arent vaccinated about getting the vaccine without being pushy or overbearing. Or if anyone has successfully convinced a loved one to get the vaccine?\n",
      "[SUMMARY]I accept that this is a personal choice that everyone has to make for themselves and I know that I need to move forward with getting my fears under control. \n",
      "\n",
      "\n",
      "Generate a summary of what triggered fear in this post: Hello, first of all I hope you are doing well. So, I'm taking oral birth control almost 4-5 years straight. I wanted to do the vaccine finally but my gynecologist said I should stop taking the pill for at least 2 months for possibly blood clots. I just want your opinion on this topic from women who did the vaccine and was taking birth control. Did you stop it, did it cause something etc.\n",
      "[SUMMARY]I'm apprehensive about whether I should go off the pill for 2 months prior to getting my COVID vaccine since my gynecologist said that I could possibly get blood clots if I don't.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered fear in this post: Im fully vaccinated and live with my parents who are also fully vaccinated but theyre both at high risk for complications if they were to get Covid. Im set to see a doctor this Tuesday for an ultrasound that Ive waited a month for, but with the delta variant and huge surge, Im scared to go. But again, dont want to delay the appointment further as it took me a while to get this one. Is it dangerous to go to the doctors now? If I wear a K95 mask should I be well protected?\n",
      "[SUMMARY]I'm afraid of getting the ultrasound that I need to get because seeing the doctor could be risky given the huge surge in COVID cases from the Delta variant, especially when I live with my parents who are at high risk for COVID complications.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered fear in this post: Hi everyone I had my 1st <ORG> vaccine 2 days ago and have had some side effects (aching arm, nausea, slight headache and fatigue) but I have had dizziness, it started soon after the injection and lasted most of the day, then it's been intermittent from there. Just wondering if anyone else has had this side effect?\n",
      "[SUMMARY]This is not a normal side effect that I know of, which makes me scared that there could be something wrong.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered sadness in this post: Trigger warning, just in case: death, delta variant talk, vaccine resistant talk? Been a while since I've posted here. I got my first dose of the vaccine yesterday. I wasn't able to get the jab sooner for reasons outside of my control, but I work from home and haven't left the house in months, so I was able to not catch COVID or spread it, thankfully. My arm hurts so much more than I thought it would, but other than that, I can't say anything of note has happened. I'm grateful for that. That said... All this news about the delta variant and vaccine resistant strains being a possibility is really beginning to scare and upset me. I'm having a really hard time keeping my cool. I do truly believe this can't go on forever, and I'm willing to force myself to get through the emotional toll of lockdowns and mask mandates... but I'm so scared of losing my life to a vaccine resistant strain. It's hard not to be furious at the antivaxxers all the time. Someone in my extended family is going to die very soon as a result of catching the delta variant as an unvaccinated person. They are already brain dead, but still on life support as of last night. Their spouse also has it, and is refusing to go to the hospital despite probably needing to be put on oxygen. It's so terrible. All of this is so terrible. I'm really scared of dying from a new strain. I know they're talking boosters, but what if it's not enough? I can handle the world standing still... in theory, for as long as it takes, but I don't want it all to be for nothing.\n",
      "[SUMMARY]I'm sad that one of my family members is about to die from COVID and that their spouse is still refusing to go to the hospital despite also having COVID.\n",
      "\n",
      "\n",
      "Generate a summary of what triggered fear in this post: I know that things may change in the next month, but what are your current thoughts on returning your kids to school in the Fall? Personally, I do not want to send my 10yo back until he is able to get the vaccine. However, I dont know any other parents who feel the same way. Most are EAGER to send their kids back to school. Right now, my state (NJ) is not allowing a virtual option. If I feel it is in his best interest, I will homeschool for the fall and re-enroll him when hes fully vaccinated. But with higher vaccination rates in NJ, precautions in school like masks and desk shields, etc., am I overreacting and putting my kid at a social disadvantage to his peers for what is a relatively low risk of infection? Saying that sounds crazy to me, though, considering the disadvantage he would be at if he did contract the virus and suffered long-term effects or death, God forbid. Thank you for sharing your thoughts.\n",
      "[SUMMARY]I'm afraid of sending my 10 year old son back to school in person until he's able to be vaccinated since he would be at risk for contracting COVID and could have long term effects or even die.\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for i in range(10):\n",
    "    print(data_dict['input'][i], f'[SUMMARY]{data_dict[\"summary\"][i]}', sep='\\n')\n",
    "    print('\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f715780f",
   "metadata": {},
   "source": [
    "## Explore Balanced Dataset Properties"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "059c0f58",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "from pathlib import Path\n",
    "\n",
    "data_path = Path('data/train_val_test-WITH_POSTS/train_anonymized-WITH_POSTS.json')\n",
    "assert data_path.exists()\n",
    "\n",
    "with open(data_path) as f:\n",
    "    raw_data = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f15c57c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import chain\n",
    "from data_preprocess import EMO_LIST\n",
    "\n",
    "emo_freq = {\n",
    "    'anger': 470,\n",
    "    'disgust': 192,\n",
    "    'fear': 765,\n",
    "    'joy': 134,\n",
    "    'sadness': 360,\n",
    "    'trust': 99,\n",
    "    'anticipation': 873\n",
    "}\n",
    "data_dict = {'post': [], 'emo': [], 'summary': []}\n",
    "emo_cnt = dict.fromkeys(EMO_LIST, 0)\n",
    "valid_emos = set(EMO_LIST)\n",
    "MAX_SIZE = min(emo_freq.values())\n",
    "\n",
    "for sample in raw_data.values():\n",
    "    annos = chain(*sample['Annotations'].values())\n",
    "    annos = list(filter(lambda a: a['Emotion'] in valid_emos, annos))\n",
    "    anno = min(annos, key=lambda a: emo_freq[a['Emotion']], default=None)\n",
    "    if anno is not None:\n",
    "        emo = anno['Emotion']\n",
    "        data_dict['emo'].append(emo)\n",
    "        data_dict['post'].append(sample['Reddit Post'])\n",
    "        data_dict['summary'].append(anno['Abstractive'])\n",
    "        emo_cnt[emo] += 1\n",
    "        if emo_cnt[emo] == MAX_SIZE:\n",
    "            valid_emos.remove(emo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "4f289693",
   "metadata": {},
   "outputs": [],
   "source": [
    "trust_summ = [\n",
    "    data_dict['summary'][i] for i, emo in enumerate(data_dict['emo'])\n",
    "    if emo == 'trust'\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "7705b2ad",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I trust that the vaccines will do their job and help end the pandemic.\n",
      "\n",
      "I trust that people here have knowledge they are willing to share that can help me make choices that won't put my loved ones at risk of catching Covid.\n",
      "\n",
      "I trust scientific evidence and know that it will help relieve my anxiety about catching Covid and with it, I will be able to lesson my obsessive behaviors. \n",
      "\n",
      "I know that taking these vitamins and minerals will help my body fight this virus and I trust that my immune system will get a boost. \n",
      "\n",
      "I trust that someone here can tell me what the chances are that I have been re-infected. Which may help relief some of my fears. \n",
      "\n",
      "I trust that I will be able to get helpful advice and learn about others that have had similar experiences, which will likely ease my anxiety.\n",
      "\n",
      "I trust that people here will give me good advice that will help me make a good decision on whether or not to get the vaccine.\n",
      "\n",
      "I accept that double masking will be necessary and feel the recommendations here can help me do it as comfortably as possible.\n",
      "\n",
      "I have a lot of trust in the vaccine's efficacy although knowing some additional information about COVID's spread between vaccinated individuals would be helpful.\n",
      "\n",
      "I trust that you can give me information that can help me make this decision.\n"
     ]
    }
   ],
   "source": [
    "print(*[*filter(lambda s: 'help' in s, trust_summ)][:10], sep='\\n\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a8bda538",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anger         vocab size: 858, mean: 3.191, variance: 62.215\n",
      "disgust       vocab size: 770, mean: 3.235, variance: 67.112\n",
      "fear          vocab size: 824, mean: 3.689, variance: 100.236\n",
      "joy           vocab size: 635, mean: 3.030, variance: 61.172\n",
      "sadness       vocab size: 788, mean: 3.268, variance: 84.759\n",
      "trust         vocab size: 621, mean: 3.176, variance: 82.077\n",
      "anticipation  vocab size: 863, mean: 3.169, variance: 80.889\n"
     ]
    }
   ],
   "source": [
    "from collections import Counter\n",
    "import numpy as np\n",
    "\n",
    "def get_emo_summ_vocab(tgt_emo):\n",
    "    emo_summ = [\n",
    "        summ for summ, emo in zip(data_dict['summary'], data_dict['emo'])\n",
    "        if emo == tgt_emo\n",
    "    ]\n",
    "    vocab = Counter(chain(*[summ.split() for summ in emo_summ]))\n",
    "    return emo_summ, vocab\n",
    "\n",
    "for tgt_emo in EMO_LIST:\n",
    "    _, vocab = get_emo_summ_vocab(tgt_emo)\n",
    "    vocab_size = len(vocab)\n",
    "\n",
    "    vocab = np.float32(list(vocab.values()))\n",
    "    vocab_mean = vocab.mean()\n",
    "    vocab_var = vocab.var()\n",
    "\n",
    "    print(f'{tgt_emo:<14}vocab size: {vocab_size}, mean: {vocab_mean:.3f}, variance: {vocab_var:.3f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7a54b4af",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['anger', 'disgust', 'fear', 'joy', 'sadness', 'trust', 'anticipation']"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "EMO_LIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e8bf0ca6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_emo_words(emo, k=20):\n",
    "    _, vocab = get_emo_summ_vocab(emo)\n",
    "    words, _ = zip(*vocab.most_common(k))\n",
    "    return set(words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "e33f1545",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "73"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "k = 400\n",
    "common_words = get_emo_words(EMO_LIST[0], k)\n",
    "\n",
    "for emo in EMO_LIST[1:]:\n",
    "    words = get_emo_words(emo, k)\n",
    "    common_words = common_words.intersection(words)\n",
    "\n",
    "len(common_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "60c67243",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anger       : {'annoyed': 30, 'was': 26, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'was': 17, 'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'if': 27, 'expect': 32}\n"
     ]
    }
   ],
   "source": [
    "for emo in EMO_LIST:\n",
    "    words = get_emo_words(emo)\n",
    "    _, vocab = get_emo_summ_vocab(emo)\n",
    "    unique = words - common_words\n",
    "    cnt = {word: vocab[word] for word in unique}\n",
    "    print(f'{emo:<12}: {cnt}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "f7bc69ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "def count_unique_words(k):\n",
    "    print(f'{k=}')\n",
    "    common_words = get_emo_words(EMO_LIST[0], k)\n",
    "\n",
    "    for emo in EMO_LIST[1:]:\n",
    "        words = get_emo_words(emo, k)\n",
    "        common_words = common_words.intersection(words)\n",
    "\n",
    "    for emo in EMO_LIST:\n",
    "        words = get_emo_words(emo)\n",
    "        _, vocab = get_emo_summ_vocab(emo)\n",
    "        unique = words - common_words\n",
    "        cnt = {word: vocab[word] for word in unique}\n",
    "        print(f'{emo:<12}: {cnt}')\n",
    "    print('-'*20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "c6352e2d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "k=50\n",
      "anger       : {'because': 26, 'was': 26, 'annoyed': 30, 'angry': 23, \"I'm\": 57}\n",
      "disgust     : {'this': 21, 'people': 29, 'disgusted': 64, \"I'm\": 43, 'see': 20}\n",
      "fear        : {'because': 36, 'afraid': 36, 'from': 27, 'apprehensive': 27, \"I'm\": 99}\n",
      "joy         : {'because': 13, 'was': 17, \"I'm\": 37, 'happy': 48}\n",
      "sadness     : {'because': 29, 'has': 21, 'sad': 69, \"I'm\": 69}\n",
      "trust       : {'help': 20, 'will': 57, 'trust': 67, 'can': 17, 'from': 19, 'not': 16}\n",
      "anticipation: {'if': 27, 'expect': 32, 'will': 26, 'from': 24}\n",
      "--------------------\n",
      "k=100\n",
      "anger       : {'annoyed': 30, 'was': 26, 'angry': 23}\n",
      "disgust     : {'disgusted': 64, 'see': 20, 'this': 21}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'was': 17, 'happy': 48}\n",
      "sadness     : {'has': 21, 'sad': 69}\n",
      "trust       : {'help': 20, 'will': 57, 'can': 17, 'trust': 67}\n",
      "anticipation: {'if': 27, 'expect': 32, 'will': 26}\n",
      "--------------------\n",
      "k=150\n",
      "anger       : {'annoyed': 30, 'was': 26, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'was': 17, 'happy': 48}\n",
      "sadness     : {'has': 21, 'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'if': 27, 'expect': 32}\n",
      "--------------------\n",
      "k=200\n",
      "anger       : {'annoyed': 30, 'was': 26, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'was': 17, 'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'if': 27, 'expect': 32}\n",
      "--------------------\n",
      "k=300\n",
      "anger       : {'annoyed': 30, 'was': 26, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'was': 17, 'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'expect': 32}\n",
      "--------------------\n",
      "k=400\n",
      "anger       : {'annoyed': 30, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'expect': 32}\n",
      "--------------------\n",
      "k=500\n",
      "anger       : {'annoyed': 30, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'can': 17, 'trust': 67}\n",
      "anticipation: {'expect': 32}\n",
      "--------------------\n",
      "k=None\n",
      "anger       : {'annoyed': 30, 'angry': 23}\n",
      "disgust     : {'see': 20, 'disgusted': 64}\n",
      "fear        : {'afraid': 36, 'apprehensive': 27}\n",
      "joy         : {'happy': 48}\n",
      "sadness     : {'sad': 69}\n",
      "trust       : {'help': 20, 'trust': 67}\n",
      "anticipation: {'expect': 32}\n",
      "--------------------\n"
     ]
    }
   ],
   "source": [
    "for k in [50, 100, 150, 200, 300, 400, 500, None]:\n",
    "    count_unique_words(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "2180eab2",
   "metadata": {},
   "outputs": [],
   "source": [
    "unique_vocab_size_hist = {emo: [] for emo in EMO_LIST}\n",
    "\n",
    "def count_unique_vocab_size(k):\n",
    "    common_words = get_emo_words(EMO_LIST[0], k)\n",
    "\n",
    "    for emo in EMO_LIST[1:]:\n",
    "        words = get_emo_words(emo, k)\n",
    "        common_words = common_words.intersection(words)\n",
    "\n",
    "    for emo in EMO_LIST:\n",
    "        words = get_emo_words(emo)\n",
    "        _, vocab = get_emo_summ_vocab(emo)\n",
    "        unique = words - common_words\n",
    "        cnt = {word: vocab[word] for word in unique}\n",
    "        unique_vocab_size_hist[emo].append(len(cnt))\n",
    "\n",
    "for k in [50, 100, 150, 200, 300, 400, None]:\n",
    "    count_unique_vocab_size(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "9c2c917b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "k_vals = [50, 100, 150, 200, 300, 400, 'Whole Dataset']\n",
    "k_axis = np.arange(len(k_vals))\n",
    "bar_width = 0.1\n",
    "offset = 0.18\n",
    "fig, ax = plt.subplots(figsize=[12, 6])\n",
    "\n",
    "for dx, emo in enumerate(EMO_LIST):\n",
    "    ax.bar(k_axis + dx * bar_width - offset, unique_vocab_size_hist[emo], width=bar_width, label=emo)\n",
    "\n",
    "ax.legend(loc='upper right')\n",
    "ax.set_xlabel('K')\n",
    "ax.set_ylabel('Unique Vocabulary Size')\n",
    "ax.set_xticks(k_axis + bar_width, k_vals)\n",
    "# ax.set_title('Vocabulary Size of Top-K Most Frequent Unique Words')\n",
    "\n",
    "fig.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "9484bc10",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--------anger--------\n",
      "('to', 'the', 'I', 'and', 'that', 'my', \"I'm\", 'a', 'of', 'COVID', 'is', 'annoyed', 'are', 'because', 'was', 'have', 'me', 'angry', 'for', 'since')\n",
      "--------disgust--------\n",
      "('to', 'the', 'I', 'disgusted', 'that', 'and', 'my', \"I'm\", 'a', 'of', 'COVID', 'since', 'people', 'in', 'for', 'are', 'is', 'be', 'this', 'see')\n",
      "--------fear--------\n",
      "('I', \"I'm\", 'the', 'to', 'my', 'and', 'a', 'that', 'of', 'COVID', 'have', 'about', 'because', 'afraid', 'since', 'for', 'in', 'from', 'apprehensive', 'is')\n",
      "--------joy--------\n",
      "('I', 'to', 'the', 'and', 'that', 'happy', \"I'm\", 'a', 'my', 'of', 'be', 'am', 'is', 'was', 'with', 'in', 'are', 'COVID', 'about', 'because')\n",
      "--------sadness--------\n",
      "('I', 'to', 'the', 'and', 'that', 'my', 'sad', \"I'm\", 'of', 'COVID', 'a', 'have', 'me', 'because', 'since', 'be', 'in', 'am', 'has', 'for')\n",
      "--------trust--------\n",
      "('I', 'that', 'the', 'trust', 'and', 'to', 'will', 'my', 'me', 'be', 'it', 'of', 'help', 'a', 'from', 'have', 'can', 'COVID', 'in', 'not')\n",
      "--------anticipation--------\n",
      "('I', 'to', 'the', 'and', 'that', 'my', 'a', 'of', 'COVID', 'be', 'in', 'expect', 'am', 'for', 'if', 'will', 'is', 'get', 'from', 'are')\n"
     ]
    }
   ],
   "source": [
    "for emo in EMO_LIST:\n",
    "    _, vocab = get_emo_summ_vocab(emo)\n",
    "    print(f'--------{emo}--------')\n",
    "    words, freq = zip(*vocab.most_common(20))\n",
    "    print(words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "feb3fcf9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anger        min=8.0 max=51.0 mean=27.657 std=10.548\n",
      "disgust      min=10.0 max=59.0 mean=25.162 std=9.141\n",
      "fear         min=11.0 max=56.0 mean=30.707 std=9.255\n",
      "joy          min=4.0 max=44.0 mean=19.434 std=8.912\n",
      "sadness      min=7.0 max=51.0 mean=26.010 std=9.763\n",
      "trust        min=7.0 max=47.0 mean=19.919 std=7.442\n",
      "anticipation min=9.0 max=56.0 mean=27.626 std=10.305\n"
     ]
    }
   ],
   "source": [
    "for emo in EMO_LIST:\n",
    "    summ, _ = get_emo_summ_vocab(emo)\n",
    "    summ_size = np.float32([len(s.split()) for s in summ])\n",
    "    print(f'{emo:<13}min={summ_size.min()} max={summ_size.max()} mean={summ_size.mean():.3f} std={summ_size.std():.3f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad58e9b0",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}