hfSentiment.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a1fdd8fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import plotly.graph_objects as go # for fancy interactive plot\n",
    "from transformers import pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "25b49e89",
   "metadata": {},
   "outputs": [],
   "source": [
    "sentiment_pl = pipeline(\"sentiment-analysis\", model=\"distilbert-base-uncased-finetuned-sst-2-english\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4f512bc0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3000, 4)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the two data sets needed\n",
    "\n",
    "facetPlotData = pd.read_csv('./Data/genData/facetPlotData.csv')\n",
    "df = pd.read_csv('./Data/genData/fullEmotionData.csv', encoding='utf_8') # apostrophes aren't being read correctly\n",
    "# Pandas reads date as string, cast to datetime object\n",
    "df['date'] = pd.to_datetime(df['date'], format='%Y-%m-%d')\n",
    "facetPlotData['date'] = pd.to_datetime(facetPlotData['date'], format='%Y-%m-%d')\n",
    "facetPlotData.shape # Should be (3000, 4) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "3276d436",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>file</th>\n",
       "      <th>text_oba</th>\n",
       "      <th>text_wsj</th>\n",
       "      <th>text_nyt</th>\n",
       "      <th>emo_oba</th>\n",
       "      <th>emo_nyt</th>\n",
       "      <th>emo_wsj</th>\n",
       "      <th>date</th>\n",
       "      <th>obafear</th>\n",
       "      <th>obaanger</th>\n",
       "      <th>...</th>\n",
       "      <th>wsjfear</th>\n",
       "      <th>wsjanger</th>\n",
       "      <th>wsjtrust</th>\n",
       "      <th>wsjsurprise</th>\n",
       "      <th>wsjpositive</th>\n",
       "      <th>wsjnegative</th>\n",
       "      <th>wsjsadness</th>\n",
       "      <th>wsjdisgust</th>\n",
       "      <th>wsjjoy</th>\n",
       "      <th>wsjanticipation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2008-06-04-ObamaNomination.txt</td>\n",
       "      <td>Thank you....Thank you. What a -- What a wonde...</td>\n",
       "      <td>Sen. Barack Obama captured enough convention d...</td>\n",
       "      <td>Senator Barack Obama claimed the Democratic pr...</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09f2a0eda0&gt;</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09e5416260&gt;</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09e2782260&gt;</td>\n",
       "      <td>2008-06-04</td>\n",
       "      <td>0.095798</td>\n",
       "      <td>0.055462</td>\n",
       "      <td>...</td>\n",
       "      <td>0.051282</td>\n",
       "      <td>0.036630</td>\n",
       "      <td>0.168498</td>\n",
       "      <td>0.047619</td>\n",
       "      <td>0.326007</td>\n",
       "      <td>0.095238</td>\n",
       "      <td>0.054945</td>\n",
       "      <td>0.025641</td>\n",
       "      <td>0.106227</td>\n",
       "      <td>0.087912</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2008-11-05-ObamaElected.txt</td>\n",
       "      <td>Hello, Chicago.\\r\\n\\r\\nIf there is anyone out ...</td>\n",
       "      <td>Sen. Barack Obama was elected the nation's fir...</td>\n",
       "      <td>Barack Hussein Obama was elected the 44th pres...</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09f2a0db10&gt;</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09e5416350&gt;</td>\n",
       "      <td>&lt;nrclex.NRCLex object at 0x7f09e2782350&gt;</td>\n",
       "      <td>2008-11-05</td>\n",
       "      <td>0.060000</td>\n",
       "      <td>0.037778</td>\n",
       "      <td>...</td>\n",
       "      <td>0.079114</td>\n",
       "      <td>0.063291</td>\n",
       "      <td>0.175633</td>\n",
       "      <td>0.049051</td>\n",
       "      <td>0.216772</td>\n",
       "      <td>0.140823</td>\n",
       "      <td>0.068038</td>\n",
       "      <td>0.030063</td>\n",
       "      <td>0.071203</td>\n",
       "      <td>0.106013</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 38 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                             file  \\\n",
       "0  2008-06-04-ObamaNomination.txt   \n",
       "1     2008-11-05-ObamaElected.txt   \n",
       "\n",
       "                                            text_oba  \\\n",
       "0  Thank you....Thank you. What a -- What a wonde...   \n",
       "1  Hello, Chicago.\\r\\n\\r\\nIf there is anyone out ...   \n",
       "\n",
       "                                            text_wsj  \\\n",
       "0  Sen. Barack Obama captured enough convention d...   \n",
       "1  Sen. Barack Obama was elected the nation's fir...   \n",
       "\n",
       "                                            text_nyt  \\\n",
       "0  Senator Barack Obama claimed the Democratic pr...   \n",
       "1  Barack Hussein Obama was elected the 44th pres...   \n",
       "\n",
       "                                    emo_oba  \\\n",
       "0  <nrclex.NRCLex object at 0x7f09f2a0eda0>   \n",
       "1  <nrclex.NRCLex object at 0x7f09f2a0db10>   \n",
       "\n",
       "                                    emo_nyt  \\\n",
       "0  <nrclex.NRCLex object at 0x7f09e5416260>   \n",
       "1  <nrclex.NRCLex object at 0x7f09e5416350>   \n",
       "\n",
       "                                    emo_wsj       date   obafear  obaanger  \\\n",
       "0  <nrclex.NRCLex object at 0x7f09e2782260> 2008-06-04  0.095798  0.055462   \n",
       "1  <nrclex.NRCLex object at 0x7f09e2782350> 2008-11-05  0.060000  0.037778   \n",
       "\n",
       "   ...   wsjfear  wsjanger  wsjtrust  wsjsurprise  wsjpositive  wsjnegative  \\\n",
       "0  ...  0.051282  0.036630  0.168498     0.047619     0.326007     0.095238   \n",
       "1  ...  0.079114  0.063291  0.175633     0.049051     0.216772     0.140823   \n",
       "\n",
       "   wsjsadness  wsjdisgust    wsjjoy  wsjanticipation  \n",
       "0    0.054945    0.025641  0.106227         0.087912  \n",
       "1    0.068038    0.030063  0.071203         0.106013  \n",
       "\n",
       "[2 rows x 38 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "619be030",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "ff95c9e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "meansents = []\n",
    "for text in df['text_oba']:\n",
    "    hfsent = sentiment_pl([sentence.strip() for sentence in text.split('.') if len(sentence)>2])\n",
    "    # hfsent is a list of dictionaries, one for each sentence. Find the mean for the whole document\n",
    "    mean = np.mean([ value[1] if key[1]=='POSITIVE' else -1*value[1] for key, value in (dictnry.items() for dictnry in hfsent) ])\n",
    "    meansents.append(mean)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "d533f0c5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'2016-02-16-ScaliaDies.txt'"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Find index of maximum\n",
    "maxidx = max( (v, i) for i, v in enumerate(meansents) )[1]\n",
    "df['file'][maxidx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "4fd1722b",
   "metadata": {},
   "outputs": [],
   "source": [
    "meandf = pd.DataFrame(meansents)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "767e7360",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot: >"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "meandf.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45cbe865",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "99372e0c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "distilbert-base-uncased-finetuned-sst-2-english/resolve/main/config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\4e60bb8efad3d4b7dc9969bf204947c185166a0a3cf37ddb6f481a876a3777b5.9f8326d0b7697c7fd57366cdde57032f46bc10e37ae81cb7eb564d66d23ec96b.json \n",
      "\n",
      "\n",
      "distilbert-base-uncased-finetuned-sst-2-english/resolve/main/tokenizer_config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\d44ec0488a5f13d92b3934cb68cc5849bd74ce63ede2eea2bf3c675e1e57297c.627f9558061e7bc67ed0f516b2f7efc1351772cc8553101f08748d44aada8b11.json \n",
      "\n",
      "\n",
      "distilbert-base-uncased-finetuned-sst-2-english/resolve/main/vocab.txt \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\83261b0c74c462e53d6367de0646b1fca07d0f15f1be045156b9cf8c71279cc9.d789d64ebfe299b0e416afc4a169632f903f693095b4629a7ea271d5a0cf2c99.json \n",
      "\n",
      "\n",
      "google/pegasus-xsum/resolve/main/config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\f8db793080242073e939bf4bc066830a677ca5e1c2d3aa1fc2a79fe733ccf3c9.149318290a6d6f03f34bb735260994b16b4a7c8609973a4abc8e9315c7c5797c.json \n",
      "\n",
      "\n",
      "google/pegasus-xsum/resolve/main/special_tokens_map.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\fa42b967ba408eb6cada6772d2611447c50154335595911e7c51a386db9901e6.294ebaa4cd17bb284635004c92d2c4d522ec488c828dcce0c2471b6f28e3fe82.json \n",
      "\n",
      "\n",
      "google/pegasus-xsum/resolve/main/spiece.model \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\332f03e91a57dbbec20f967da851c4a988b93136d81acd1e170a728ec6b9e3ad.1acf68c74589da6c7fa3548093824dfc450a54637f4356929bbfea7e294a68f8.json \n",
      "\n",
      "\n",
      "google/pegasus-xsum/resolve/main/tokenizer.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\249fc759c5fa315dca06897d79493c1d27d999d4876ebb798d07d9d7a2ebbb29.b11a22b825f5f3714fac8c470a8602bd6f8fbce86cb56e588c571016e963bd4e.json \n",
      "\n",
      "\n",
      "google/pegasus-xsum/resolve/main/tokenizer_config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\27c4324d316dbe67537c2fe759928141fb29dfa1cc6c6b2674765b2bc7026f54.85037d315e8cfbde266939f8b6852c54e9482ce4c0aa513420289e0b5c0f3499.json \n",
      "\n",
      "\n",
      "nlptown/bert-base-multilingual-uncased-sentiment/resolve/main/config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\d9226eeac7b8b96d83ebc327cdd670490866d8c999505c1f83b6ef206ccb1604.a34960b447312b0727cb670d710444fcb41a6156eddcba062a19b3fc05d95251.json \n",
      "\n",
      "\n",
      "nlptown/bert-base-multilingual-uncased-sentiment/resolve/main/special_tokens_map.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\ed85e7bfaa7dfcf9924004400478a6426fcab28d3e427960549371a1729115d1.dd8bd9bfd3664b530ea4e645105f557769387b3da9f79bdb55ed556bdd80611d.json \n",
      "\n",
      "\n",
      "nlptown/bert-base-multilingual-uncased-sentiment/resolve/main/tokenizer_config.json \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\4d1409805ace1b4cc209352f82a0f0c59a015433d0a58e655394ffe7bbb755e9.13a045cad07359e6844c4f487af8e6323ad2308cac6357692d2359f1a9711443.json \n",
      "\n",
      "\n",
      "nlptown/bert-base-multilingual-uncased-sentiment/resolve/main/vocab.txt \n",
      " C:\\Users\\peter/.cache\\huggingface\\transformers\\d893db6e58fdc9c39ecb49aaba0fc940780c179e1fff5bfe27e322e1599c191f.92022aa29ab6663b0b4254744f28ab43e6adf4deebe0f26651e6c61f28f69d8b.json \n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Returns some info on downloaded hf models so they can be deleted\n",
    "import glob\n",
    "import json\n",
    "import re\n",
    "from collections import OrderedDict \n",
    "from transformers import TRANSFORMERS_CACHE\n",
    " \n",
    "metaFiles = glob.glob(TRANSFORMERS_CACHE + '/*.json')\n",
    "modelRegex = \"huggingface\\.co\\/(.*)(pytorch_model\\.bin$|resolve\\/main\\/tf_model\\.h5$)\"\n",
    "\n",
    "cachedModels = {}\n",
    "cachedTokenizers = {}\n",
    "for file in metaFiles:\n",
    "     with open(file) as j:\n",
    "         data = json.load(j)\n",
    "         isM = re.search(modelRegex, data['url'])\n",
    "         if isM:\n",
    "             cachedModels[isM.group(1)[:-1]] = file\n",
    "         else:\n",
    "             cachedTokenizers[data['url'].partition('huggingface.co/')[2]] = file\n",
    "\n",
    "cachedTokenizers = OrderedDict(sorted(cachedTokenizers.items(), key=lambda k: k[0]))\n",
    "[print(key,'\\n', value, '\\n\\n') for key, value in cachedTokenizers.items()];"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "78cc11c7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "67abab6b",
   "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": "3.9.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}