project1/IR_9831011.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ba9d9b6e",
   "metadata": {},
   "source": [
    "# Information Retrieval Project (Phase 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de687bac",
   "metadata": {},
   "source": [
    "## import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "089d3bfc",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import json\n",
    "import pickle\n",
    "import re\n",
    "import math\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from string import punctuation\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from parsivar import Normalizer, Tokenizer, FindStems"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f647e1b",
   "metadata": {},
   "source": [
    "### Structure of dataset\n",
    "- dataset \n",
    "    - docID\n",
    "        - title\n",
    "        - content\n",
    "        - tags\n",
    "        - date\n",
    "        - url\n",
    "        - category"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b5cfc204",
   "metadata": {},
   "source": [
    "## Load Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b36ba4ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "docs_path = './IR_data_news_12k.json'\n",
    "\n",
    "def load_docs(docs_path):\n",
    "    result = {}\n",
    "    with open(docs_path) as f:\n",
    "        docs = json.load(f)\n",
    "        for docID, body in docs.items():\n",
    "            result[docID] = {}\n",
    "            result[docID]['title'] = body['title']\n",
    "            result[docID]['content'] = body['content']\n",
    "            result[docID]['url'] = body['url']\n",
    "    return result\n",
    "\n",
    "docs = load_docs(docs_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91968765",
   "metadata": {},
   "source": [
    "## Preprocess Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4938ac7d",
   "metadata": {},
   "source": [
    "### Remove Punctuation\n",
    "<div dir=\"rtl\">\n",
    "    حذف علائم نگارشی موجود در زبان برای جلوگیری از ورود آنها به دیکشنری\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2e936f31",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "به گزارش ایسنا سمینار شیمی آلی از امروز ۱۱ شهریور ۱۳۹۶ در دانشگاه علم و صنعت ایران آغاز به کار کرد این سمینار تا ۱۳ شهریور ادامه می یابد\n"
     ]
    }
   ],
   "source": [
    "tmp_sentence = 'به گزارش ایسنا سمینار شیمی آلی از امروز ۱۱ شهریور ۱۳۹۶ در دانشگاه علم و صنعت ایران آغاز به کار کرد. این سمینار تا ۱۳ شهریور ادامه می یابد.'\n",
    "\n",
    "pure_sentence = re.sub(f'[{punctuation}؟،٪×÷»«]+', '', tmp_sentence)\n",
    "print(pure_sentence)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdb26cb4",
   "metadata": {},
   "source": [
    "### Normalize Text\n",
    "<div dir=\"rtl\">\n",
    "    هم فرم و هم شکل کردن قسمت های مختلف تکست برای ایجاد توانایی مقایسه میان تکست های مختلف و قالب های متفاوت\n",
    "    <br>\n",
    "    برای مثال انگلیسی کردن تمامی اعداد\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "166fbdf2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "به گزارش ایسنا سمینار شیمی آلی از امروز 11 شهریور 1396 در دانشگاه علم و صنعت ایران آغاز به کار کرد این سمینار تا 13 شهریور ادامه می‌یابد\n"
     ]
    }
   ],
   "source": [
    "normalizer = Normalizer()\n",
    "normalized_sentence = normalizer.normalize(pure_sentence)\n",
    "print(normalized_sentence)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0d72e6c2",
   "metadata": {},
   "source": [
    "### Tokenize Text\n",
    "<div dir=\"rtl\">\n",
    "    تبدیل متن به توکن ها\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b8322f7f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['به', 'گزارش', 'ایسنا', 'سمینار', 'شیمی', 'آلی', 'از', 'امروز', '11', 'شهریور', '1396', 'در', 'دانشگاه', 'علم', 'و', 'صنعت', 'ایران', 'آغاز', 'به', 'کار', 'کرد', 'این', 'سمینار', 'تا', '13', 'شهریور', 'ادامه', 'می\\u200cیابد']\n"
     ]
    }
   ],
   "source": [
    "tokenizer = Tokenizer()\n",
    "tokens = tokenizer.tokenize_words(normalized_sentence)\n",
    "print(tokens)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eafea569",
   "metadata": {},
   "source": [
    "### Word Stemming\n",
    "<div dir=\"rtl\">\n",
    "    ریشه یابی کلمات موجود در متن. این کار باعث می شود که با تغییر شکل یک کلمه آن کلمه مجزا در نظر گرفته نشده و بتوان به شکل بهتری میزان تکرار آن کلمه در متن را به دست آورد.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ad6b26d8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['به', 'گزارش', 'ایسنا', 'سمینار', 'شیمی', 'آلی', 'از', 'امروز', '11', 'شهریور', '1396', 'در', 'دانشگاه', 'علم', 'و', 'صنعت', 'ایران', 'آغاز', 'به', 'کار', 'کرد', 'این', 'سمینار', 'تا', '13', 'شهریور', 'ادامه', 'یافت&یاب']\n"
     ]
    }
   ],
   "source": [
    "stemmer = FindStems()\n",
    "stemmed_tokens = list(map(stemmer.convert_to_stem, tokens))\n",
    "print(stemmed_tokens)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2bfd5087",
   "metadata": {},
   "source": [
    "### Remove Stopwords\n",
    "<div dir=\"rtl\">\n",
    "    حذف کلمات با تکرار زیاد. این کلمات با تکرار زیاد باعث جاگیری اضافه در دیکشنری و ایندکس شده و از آنجایی که در غالب سند ها نیز وجود دارند، در فرایند بازیابی سند کمک چندانی نمی کنند.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "47fa8e62",
   "metadata": {},
   "outputs": [],
   "source": [
    "persian_stopwords_path = './persian-stopwords.txt'\n",
    "\n",
    "def load_stopwords(stopwords_path):\n",
    "    stopwords_set = None\n",
    "    with open(stopwords_path, 'r') as f:\n",
    "        stopwords_set = set(f.read().split('\\n'))\n",
    "    return stopwords_set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "630db888",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'آخرین',\n",
       " 'آقای',\n",
       " 'آمد',\n",
       " 'آمده',\n",
       " 'آمده_است',\n",
       " 'آن',\n",
       " 'آنان',\n",
       " 'آنجا',\n",
       " 'آنها',\n",
       " 'آنچه',\n",
       " 'آنکه',\n",
       " 'آورد',\n",
       " 'آوری',\n",
       " 'آیا',\n",
       " 'ابتدا',\n",
       " 'اثر',\n",
       " 'اجرا',\n",
       " 'اخیر',\n",
       " 'از',\n",
       " 'است',\n",
       " 'اش',\n",
       " 'اغلب',\n",
       " 'افراد',\n",
       " 'افرادی',\n",
       " 'افزود',\n",
       " 'البته',\n",
       " 'اما',\n",
       " 'امر',\n",
       " 'امکان',\n",
       " 'اند',\n",
       " 'او',\n",
       " 'اول',\n",
       " 'اولین',\n",
       " 'اکنون',\n",
       " 'اگر',\n",
       " 'ایشان',\n",
       " 'این',\n",
       " 'اینجا',\n",
       " 'اینکه',\n",
       " 'با',\n",
       " 'بار',\n",
       " 'باره',\n",
       " 'باز',\n",
       " 'باشد',\n",
       " 'باشند',\n",
       " 'باعث',\n",
       " 'بالا',\n",
       " 'باید',\n",
       " 'بخش',\n",
       " 'بخشی',\n",
       " 'بدون',\n",
       " 'بر',\n",
       " 'برابر',\n",
       " 'براساس',\n",
       " 'برای',\n",
       " 'برخی',\n",
       " 'برداری',\n",
       " 'بروز',\n",
       " 'بزرگ',\n",
       " 'بسیار',\n",
       " 'بسیاری',\n",
       " 'بعد',\n",
       " 'بعضی',\n",
       " 'بلکه',\n",
       " 'بنابراین',\n",
       " 'بندی',\n",
       " 'به',\n",
       " 'بهتر',\n",
       " 'بهترین',\n",
       " 'بود',\n",
       " 'بودن',\n",
       " 'بودند',\n",
       " 'بوده',\n",
       " 'بوده_است',\n",
       " 'بی',\n",
       " 'بیان',\n",
       " 'بیرون',\n",
       " 'بیش',\n",
       " 'بیشتر',\n",
       " 'بیشتری',\n",
       " 'بین',\n",
       " 'تا',\n",
       " 'تاکنون',\n",
       " 'تبدیل',\n",
       " 'تحت',\n",
       " 'ترتیب',\n",
       " 'تعداد',\n",
       " 'تعیین',\n",
       " 'تغییر',\n",
       " 'تمام',\n",
       " 'تمامی',\n",
       " 'تنها',\n",
       " 'تهیه',\n",
       " 'تو',\n",
       " 'جا',\n",
       " 'جاری',\n",
       " 'جای',\n",
       " 'جایی',\n",
       " 'جدی',\n",
       " 'جدید',\n",
       " 'جریان',\n",
       " 'جز',\n",
       " 'جمع',\n",
       " 'جمعی',\n",
       " 'حال',\n",
       " 'حالا',\n",
       " 'حالی',\n",
       " 'حتی',\n",
       " 'حد',\n",
       " 'حداقل',\n",
       " 'حدود',\n",
       " 'حل',\n",
       " 'خاص',\n",
       " 'خاطرنشان',\n",
       " 'خصوص',\n",
       " 'خطر',\n",
       " 'خواهد_بود',\n",
       " 'خواهد_شد',\n",
       " 'خواهد_کرد',\n",
       " 'خوب',\n",
       " 'خوبی',\n",
       " 'خود',\n",
       " 'خودش',\n",
       " 'خویش',\n",
       " 'خیلی',\n",
       " 'داد',\n",
       " 'دادن',\n",
       " 'دادند',\n",
       " 'داده',\n",
       " 'داده_است',\n",
       " 'دار',\n",
       " 'دارای',\n",
       " 'دارد',\n",
       " 'دارند',\n",
       " 'داریم',\n",
       " 'داشت',\n",
       " 'داشتن',\n",
       " 'داشتند',\n",
       " 'داشته',\n",
       " 'داشته_است',\n",
       " 'داشته_باشد',\n",
       " 'داشته_باشند',\n",
       " 'دانست',\n",
       " 'در',\n",
       " 'درباره',\n",
       " 'درون',\n",
       " 'دسته',\n",
       " 'دهد',\n",
       " 'دهند',\n",
       " 'دهه',\n",
       " 'دو',\n",
       " 'دوباره',\n",
       " 'دور',\n",
       " 'دوم',\n",
       " 'دچار',\n",
       " 'دیگر',\n",
       " 'دیگران',\n",
       " 'دیگری',\n",
       " 'را',\n",
       " 'راه',\n",
       " 'رسید',\n",
       " 'رسیدن',\n",
       " 'رشد',\n",
       " 'رفت',\n",
       " 'رو',\n",
       " 'روبه',\n",
       " 'روش',\n",
       " 'روند',\n",
       " 'روی',\n",
       " 'ریزی',\n",
       " 'زاده',\n",
       " 'زیاد',\n",
       " 'زیادی',\n",
       " 'زیر',\n",
       " 'زیرا',\n",
       " 'ساز',\n",
       " 'سازی',\n",
       " 'ساله',\n",
       " 'سالهای',\n",
       " 'سال\\u200cهای',\n",
       " 'سایر',\n",
       " 'سبب',\n",
       " 'سراسر',\n",
       " 'سعی',\n",
       " 'سمت',\n",
       " 'سه',\n",
       " 'سهم',\n",
       " 'سوم',\n",
       " 'سوی',\n",
       " 'سپس',\n",
       " 'سی',\n",
       " 'شامل',\n",
       " 'شان',\n",
       " 'شاید',\n",
       " 'شخصی',\n",
       " 'شد',\n",
       " 'شدن',\n",
       " 'شدند',\n",
       " 'شده',\n",
       " 'شده_است',\n",
       " 'شده_اند',\n",
       " 'شده_بود',\n",
       " 'شروع',\n",
       " 'شش',\n",
       " 'شما',\n",
       " 'شمار',\n",
       " 'شود',\n",
       " 'شوند',\n",
       " 'صرف',\n",
       " 'ضمن',\n",
       " 'طبق',\n",
       " 'طرف',\n",
       " 'طور',\n",
       " 'طول',\n",
       " 'طی',\n",
       " 'ع',\n",
       " 'عالی',\n",
       " 'عدم',\n",
       " 'علاوه',\n",
       " 'علت',\n",
       " 'علیه',\n",
       " 'عهده',\n",
       " 'عین',\n",
       " 'غیر',\n",
       " 'فرد',\n",
       " 'فردی',\n",
       " 'فقط',\n",
       " 'فوق',\n",
       " 'فکر',\n",
       " 'قابل',\n",
       " 'قبل',\n",
       " 'لازم',\n",
       " 'لحاظ',\n",
       " 'لذا',\n",
       " 'ما',\n",
       " 'مانند',\n",
       " 'متاسفانه',\n",
       " 'متر',\n",
       " 'متفاوت',\n",
       " 'مثل',\n",
       " 'محسوب',\n",
       " 'مدت',\n",
       " 'مربوط',\n",
       " 'مشخص',\n",
       " 'ممکن',\n",
       " 'من',\n",
       " 'مناسب',\n",
       " 'منظور',\n",
       " 'مهم',\n",
       " 'مواجه',\n",
       " 'موجب',\n",
       " 'مورد',\n",
       " 'می',\n",
       " 'میان',\n",
       " 'می\\u200cآید',\n",
       " 'می\\u200cباشد',\n",
       " 'می\\u200cتوان',\n",
       " 'می\\u200cتواند',\n",
       " 'می\\u200cتوانند',\n",
       " 'می\\u200cدهد',\n",
       " 'می\\u200cدهند',\n",
       " 'می\\u200cرسد',\n",
       " 'می\\u200cرود',\n",
       " 'می\\u200cشد',\n",
       " 'می\\u200cشود',\n",
       " 'می\\u200cشوند',\n",
       " 'می\\u200cکرد',\n",
       " 'می\\u200cکردند',\n",
       " 'می\\u200cکند',\n",
       " 'می\\u200cکنم',\n",
       " 'می\\u200cکنند',\n",
       " 'می\\u200cکنیم',\n",
       " 'می\\u200cگوید',\n",
       " 'می\\u200cگویند',\n",
       " 'می\\u200cگیرد',\n",
       " 'می\\u200cیابد',\n",
       " 'ناشی',\n",
       " 'نباید',\n",
       " 'نبود',\n",
       " 'نحوه',\n",
       " 'نخست',\n",
       " 'نخستین',\n",
       " 'ندارد',\n",
       " 'ندارند',\n",
       " 'نزدیک',\n",
       " 'نسبت',\n",
       " 'نشست',\n",
       " 'نظر',\n",
       " 'نظیر',\n",
       " 'نمی\\u200cشود',\n",
       " 'نه',\n",
       " 'نوع',\n",
       " 'نوعی',\n",
       " 'نیاز',\n",
       " 'نیز',\n",
       " 'نیست',\n",
       " 'نیستند',\n",
       " 'نیمه',\n",
       " 'هایی',\n",
       " 'هر',\n",
       " 'هستند',\n",
       " 'هستیم',\n",
       " 'هم',\n",
       " 'همان',\n",
       " 'همه',\n",
       " 'همواره',\n",
       " 'همچنان',\n",
       " 'همچنین',\n",
       " 'همچون',\n",
       " 'همیشه',\n",
       " 'همین',\n",
       " 'هنوز',\n",
       " 'هنگام',\n",
       " 'هیچ',\n",
       " 'و',\n",
       " 'وارد',\n",
       " 'وجود',\n",
       " 'وقتی',\n",
       " 'ولی',\n",
       " 'وگو',\n",
       " 'وی',\n",
       " 'ویژه',\n",
       " 'پخش',\n",
       " 'پر',\n",
       " 'پس',\n",
       " 'پشت',\n",
       " 'پنج',\n",
       " 'پی',\n",
       " 'پیدا',\n",
       " 'پیش',\n",
       " 'چرا',\n",
       " 'چند',\n",
       " 'چنین',\n",
       " 'چه',\n",
       " 'چهار',\n",
       " 'چهارم',\n",
       " 'چون',\n",
       " 'چگونه',\n",
       " 'چیز',\n",
       " 'چیزی',\n",
       " 'کافی',\n",
       " 'کامل',\n",
       " 'کاملا',\n",
       " 'کدام',\n",
       " 'کرد',\n",
       " 'کردم',\n",
       " 'کردن',\n",
       " 'کردند',\n",
       " 'کرده',\n",
       " 'کرده_است',\n",
       " 'کرده_اند',\n",
       " 'کرده_بود',\n",
       " 'کسانی',\n",
       " 'کسی',\n",
       " 'کل',\n",
       " 'کلی',\n",
       " 'کم',\n",
       " 'کمی',\n",
       " 'کنار',\n",
       " 'کند',\n",
       " 'کنم',\n",
       " 'کنند',\n",
       " 'کننده',\n",
       " 'کنندگان',\n",
       " 'کنید',\n",
       " 'کنیم',\n",
       " 'که',\n",
       " 'کوچک',\n",
       " 'گاه',\n",
       " 'گذاری',\n",
       " 'گردد',\n",
       " 'گرفت',\n",
       " 'گرفته',\n",
       " 'گرفته_است',\n",
       " 'گروهی',\n",
       " 'گفت',\n",
       " 'گفته',\n",
       " 'گونه',\n",
       " 'گیرد',\n",
       " 'گیری',\n",
       " 'یا',\n",
       " 'یابد',\n",
       " 'یافت',\n",
       " 'یافته',\n",
       " 'یافته_است',\n",
       " 'یعنی',\n",
       " 'یک',\n",
       " 'یکدیگر',\n",
       " 'یکی'}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stopwords_set = load_stopwords(persian_stopwords_path)\n",
    "stopwords_set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "cf05243b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_stopwords(words, persian_stopwords_path='./persian-stopwords.txt'):\n",
    "    persian_stopwords = load_stopwords(persian_stopwords_path)\n",
    "    return [word for word in words if word not in persian_stopwords]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6b830d23",
   "metadata": {},
   "outputs": [],
   "source": [
    "stemmed_tokens_without_stopwords = remove_stopwords(stemmed_tokens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3a4108e8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "total words with stopwords: 28\n",
      "total words without stopwords: 20\n"
     ]
    }
   ],
   "source": [
    "print(f'total words with stopwords: {len(stemmed_tokens)}')\n",
    "print(f'total words without stopwords: {len(remove_stopwords(stemmed_tokens_without_stopwords))}')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ddcb9c6",
   "metadata": {},
   "source": [
    "### Stacking These components to make a Pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "58a68085",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['سلام', 'امروز', 'خواست&خواه', 'تابع', '3', 'امتحان', 'کرد&کن']\n"
     ]
    }
   ],
   "source": [
    "def preprocess(text, stopwords_remove=True, stemming=True):\n",
    "    normalizer = Normalizer()\n",
    "    tokenizer = Tokenizer()\n",
    "    stemmer = FindStems()\n",
    "    \n",
    "    pure_text = re.sub(f'[{punctuation}؟،٪×÷»«]+', '', text)\n",
    "    normal_text = normalizer.normalize(pure_text)\n",
    "    res = tokenizer.tokenize_words(normal_text)\n",
    "    if stemming:\n",
    "        res = list(map(stemmer.convert_to_stem, res))\n",
    "    if stopwords_remove:\n",
    "        res = remove_stopwords(res, persian_stopwords_path)\n",
    "    \n",
    "    return res\n",
    "\n",
    "print(preprocess('سلام من امروز می خواهم این تابع را ۳ بار امتحان کنم.'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "f08c9626",
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess_contents(docs_dict, stopwords_remove=True, stemming=True):\n",
    "    normalizer = Normalizer()\n",
    "    tokenizer = Tokenizer()\n",
    "    stemmer = FindStems()\n",
    "    \n",
    "    for docID, body in docs_dict.items():\n",
    "        body['content'] = preprocess(body['content'], stopwords_remove, stemming)\n",
    "        \n",
    "    return docs_dict\n",
    "\n",
    "preprocessed_docs = preprocess_contents(load_docs(docs_path))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab1aea4f",
   "metadata": {},
   "source": [
    "## Indexing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e03373a7",
   "metadata": {},
   "source": [
    "### Dictionary (Vocabulary)\n",
    "At first, we must create a dictionary of all words that have been seen in the docs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "04e58f01",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_dictionary(docs):\n",
    "    dictionary = set()\n",
    "    for body in docs.values():\n",
    "        dictionary.update(body['content'])\n",
    "    return dictionary\n",
    "\n",
    "dictionary = create_dictionary(preprocessed_docs)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e084f7cb",
   "metadata": {},
   "source": [
    "### Positional Index\n",
    "Structure of Index:\n",
    "- Token\n",
    "    - Total Appearance\n",
    "    - docs\n",
    "        - list of docs\n",
    "            - total incidences in every doc\n",
    "            - list of each position in every doc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "9ac41c40",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 have been processed.\n",
      "1000 have been processed.\n",
      "2000 have been processed.\n",
      "3000 have been processed.\n",
      "4000 have been processed.\n",
      "5000 have been processed.\n",
      "6000 have been processed.\n",
      "7000 have been processed.\n",
      "8000 have been processed.\n",
      "9000 have been processed.\n",
      "10000 have been processed.\n",
      "11000 have been processed.\n",
      "12000 have been processed.\n"
     ]
    }
   ],
   "source": [
    "def create_index(docs):\n",
    "    dictionary = set()\n",
    "    index = {}\n",
    "    for docID, body in docs.items():\n",
    "        if int(docID) % 1000 == 0:\n",
    "            print(f'{int(docID)} have been processed.')\n",
    "        content = body['content']\n",
    "        for token_i in range(len(content)):\n",
    "            token = content[token_i]\n",
    "            if token not in dictionary:\n",
    "                dictionary.add(token)\n",
    "                index[token] = {\n",
    "                    'total_app': 1,\n",
    "                    'docs': [\n",
    "                        {\n",
    "                            docID:{\n",
    "                                'total_inc': 1,\n",
    "                                'position': [token_i,]\n",
    "                            }\n",
    "                        }\n",
    "                    ]\n",
    "                }\n",
    "            else:\n",
    "                index[token]['total_app'] += 1\n",
    "                \n",
    "                docs_ID = list(map(lambda x : list(x.keys())[0], index[token]['docs']))\n",
    "                docs_ID_set = set(map(lambda x : list(x.keys())[0], index[token]['docs']))\n",
    "                if docID not in docs_ID_set:\n",
    "                    index[token]['docs'].append({\n",
    "                        docID:{\n",
    "                            'total_inc': 1,\n",
    "                            'position': [token_i,]\n",
    "                        }\n",
    "                    })\n",
    "                else:\n",
    "                    doc_index = docs_ID.index(docID)\n",
    "                    index[token]['docs'][doc_index][docID]['total_inc'] += 1\n",
    "                    index[token]['docs'][doc_index][docID]['position'].append(token_i)\n",
    "    return index, dictionary\n",
    "\n",
    "index, dictionary = create_index(preprocessed_docs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "34b681e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def save_index(index, filename):\n",
    "    with open(filename, 'wb') as outp:  # Overwrites any existing file.\n",
    "        pickle.dump(index, outp, pickle.HIGHEST_PROTOCOL)\n",
    "        print(f'index saved in {filename}')\n",
    "    \n",
    "def load_index(filename):\n",
    "    index = None\n",
    "    with open(filename, 'rb') as inp:\n",
    "        index = pickle.load(inp)\n",
    "    return index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f723c995",
   "metadata": {},
   "outputs": [],
   "source": [
    "index = load_index('index_v2.pkl')\n",
    "dictionary = load_index('dictionary_v2.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "b69e00a9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'total_app': 34,\n",
       " 'docs': [{'2267': {'total_inc': 1, 'position': [338]}},\n",
       "  {'2511': {'total_inc': 1, 'position': [23]}},\n",
       "  {'2897': {'total_inc': 1, 'position': [131]}},\n",
       "  {'2980': {'total_inc': 1, 'position': [28]}},\n",
       "  {'3059': {'total_inc': 2, 'position': [105, 144]}},\n",
       "  {'3084': {'total_inc': 1, 'position': [45]}},\n",
       "  {'3131': {'total_inc': 1, 'position': [31]}},\n",
       "  {'3140': {'total_inc': 1, 'position': [53]}},\n",
       "  {'3241': {'total_inc': 1, 'position': [62]}},\n",
       "  {'3323': {'total_inc': 1, 'position': [60]}},\n",
       "  {'3334': {'total_inc': 1, 'position': [43]}},\n",
       "  {'3451': {'total_inc': 2, 'position': [43, 54]}},\n",
       "  {'3495': {'total_inc': 1, 'position': [281]}},\n",
       "  {'3567': {'total_inc': 1, 'position': [25]}},\n",
       "  {'3687': {'total_inc': 2, 'position': [25, 49]}},\n",
       "  {'3950': {'total_inc': 2, 'position': [175, 191]}},\n",
       "  {'4447': {'total_inc': 3, 'position': [196, 329, 346]}},\n",
       "  {'4939': {'total_inc': 2, 'position': [62, 112]}},\n",
       "  {'4945': {'total_inc': 2, 'position': [5, 81]}},\n",
       "  {'4951': {'total_inc': 2, 'position': [105, 124]}},\n",
       "  {'5134': {'total_inc': 1, 'position': [660]}},\n",
       "  {'5484': {'total_inc': 1, 'position': [69]}},\n",
       "  {'5781': {'total_inc': 1, 'position': [144]}},\n",
       "  {'5827': {'total_inc': 1, 'position': [157]}},\n",
       "  {'8002': {'total_inc': 1, 'position': [336]}}]}"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "index['معاوضه']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc0d6ad1",
   "metadata": {},
   "source": [
    "## Query Processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "36467369",
   "metadata": {},
   "outputs": [],
   "source": [
    "def process_query(query):\n",
    "    \n",
    "    intersect = set()\n",
    "    \n",
    "    query_index = dict()\n",
    "    exception_index = dict()\n",
    "    phrase_index = dict()\n",
    "    \n",
    "    token_docID = dict()\n",
    "    exception_token_docID = dict()\n",
    "    phrase_token_docID = dict()\n",
    "    \n",
    "    exception_tokens = list()\n",
    "    phrasal_tokens = list()\n",
    "    \n",
    "    # find tokens immediately after exclamation's mark\n",
    "    exception_tokens = re.findall(r'\\!\\s(\\w+)', query)\n",
    "    # find tokens in between double quotation\n",
    "    phrasal_tokens = re.findall(r'\"([^\"]*)\"', query)\n",
    "    # remove two types of tokens extracted before from main query\n",
    "    raw_tokens = re.sub(r'\\!\\s\\w+', '', query)\n",
    "    raw_tokens = re.sub(r'\"[^\"]*\"', '', raw_tokens)\n",
    "            \n",
    "#     print(raw_tokens)\n",
    "#     print(\" \".join(exception_tokens))\n",
    "#     print(\" \".join(phrasal_tokens))\n",
    "    \n",
    "    # preprocess each type of token in query\n",
    "    if len(raw_tokens):\n",
    "        preprocessed_query = preprocess(raw_tokens)\n",
    "        preprocessed_query = [token for token in preprocessed_query if token in dictionary]\n",
    "        raw_tokens = preprocessed_query\n",
    "    if len(exception_tokens) > 0:\n",
    "        preprocessed_exception = preprocess(\" \".join(exception_tokens))\n",
    "        preprocessed_exception = [token for token in preprocessed_exception if token in dictionary]\n",
    "        exception_tokens = preprocessed_exception\n",
    "    if len(phrasal_tokens):\n",
    "        preprocessed_phrasals = list()\n",
    "        for phrase in phrasal_tokens:\n",
    "            preprocessed_phrasals.append(preprocess(phrase))\n",
    "    \n",
    "    # separate docID related to extracted raw tokens\n",
    "    if len(raw_tokens):\n",
    "        for token in preprocessed_query:\n",
    "            query_index[token] = index[token]\n",
    "            token_docID[token] = set(map(lambda x: int(list(x.keys())[0]), index[token]['docs']))\n",
    "\n",
    "    # separate docID related to extracted negated tokens\n",
    "    if len(exception_tokens) > 0:\n",
    "        for token in preprocessed_exception:\n",
    "            exception_index = index[token]\n",
    "            exception_token_docID[token] = set(map(lambda x: int(list(x.keys())[0]), index[token]['docs']))\n",
    "            \n",
    "    if len(phrasal_tokens):\n",
    "        for phrase in preprocessed_phrasals:\n",
    "            phrase_token_docID[tuple(phrase)] = dict()\n",
    "            for token in phrase:\n",
    "                phrase_index = index[token]\n",
    "                phrase_token_docID[tuple(phrase)][token] = set(map(lambda x: int(list(x.keys())[0]), phrase_index['docs']))\n",
    "         \n",
    "#         print(phrase_token_docID)\n",
    "        phrase_docID = dict()\n",
    "        for phrase in preprocessed_phrasals:\n",
    "            phrase_docID[tuple(phrase)] = set()\n",
    "            phrase_intersect = phrase_token_docID[tuple(phrase)][phrase[0]]\n",
    "            for token in phrase:\n",
    "                phrase_intersect = phrase_intersect.intersection(phrase_token_docID[tuple(phrase)][token])\n",
    "            \n",
    "            for docID in list(phrase_intersect):\n",
    "#                 print(index[phrase[0]]['docs'])\n",
    "                sequential_position = None\n",
    "                for i in range(len(phrase)):\n",
    "                    current_token = phrase[i]\n",
    "                    for doc in index[current_token]['docs']:\n",
    "                        if int(list(doc.keys())[0]) == docID:\n",
    "                            current_position = set(doc[f'{docID}']['position'])\n",
    "                            if i == 0:\n",
    "                                sequential_position = current_position\n",
    "                                continue\n",
    "                            current_position = set(map(lambda x : x - i, current_position))\n",
    "                            sequential_position = sequential_position.intersection(current_position)\n",
    "                            \n",
    "                if len(sequential_position) > 0:\n",
    "                    phrase_docID[tuple(phrase)].add(docID)\n",
    "                    \n",
    "        final_phrase_docID = phrase_docID[tuple(preprocessed_phrasals[0])]\n",
    "        for phrase in preprocessed_phrasals:\n",
    "            final_phrase_docID = final_phrase_docID.intersection(phrase_docID[tuple(phrase)])\n",
    "                \n",
    "    if len(raw_tokens) > 0:\n",
    "        intersect = token_docID[preprocessed_query[0]]\n",
    "        for token in preprocessed_query:\n",
    "            intersect = intersect.intersection(token_docID[token])\n",
    "        \n",
    "    if len(phrasal_tokens) > 0:\n",
    "        if len(intersect) > 0:\n",
    "            intersect = intersect.intersection(final_phrase_docID)\n",
    "        else: \n",
    "            intersect = final_phrase_docID\n",
    "        \n",
    "    if len(exception_tokens) > 0:\n",
    "        for token in  preprocessed_exception:\n",
    "            intersect = intersect - exception_token_docID[token]\n",
    "            \n",
    "#     print(f'intersect: {intersect}')\n",
    "        \n",
    "    related_docID = list(intersect)\n",
    "#     print(related_docID)\n",
    "#     print(query_index)\n",
    "    related_docs = list()\n",
    "    for docID in related_docID:\n",
    "        total_token_inc = 0\n",
    "        if len(raw_tokens) > 0:\n",
    "            for token in preprocessed_query:\n",
    "                for doc in query_index[token]['docs']:\n",
    "                    if int(list(doc.keys())[0]) == docID:\n",
    "                        total_token_inc += doc[f'{docID}']['total_inc']\n",
    "        if len(phrasal_tokens) > 0:\n",
    "            for phrase in preprocessed_phrasals:\n",
    "                for token in phrase:\n",
    "                    for doc in index[token]['docs']:\n",
    "                        if int(list(doc.keys())[0]) == docID:\n",
    "                            total_token_inc += doc[f'{docID}']['total_inc']\n",
    "        related_docs.append((docID, total_token_inc))\n",
    "        \n",
    "    related_docs.sort(key=lambda x : x[1], reverse=True)\n",
    "        \n",
    "    return related_docs[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "2e8ab0a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_query_result(query, with_content=False):\n",
    "    res = process_query(query)\n",
    "    for doc, rank in res:\n",
    "        print(20*'*')\n",
    "        print(f'Rank: {rank}')\n",
    "        print(f'docID: {doc}')\n",
    "        print(f'Title: {docs[f\"{doc}\"][\"title\"]}')\n",
    "        print(f'URL: {docs[f\"{doc}\"][\"url\"]}')\n",
    "        if with_content:\n",
    "            print(f'{docs[f\"{doc}\"][\"content\"]}')\n",
    "        print(20*'*')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a8b19f5a",
   "metadata": {},
   "source": [
    "# Sample Queries"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4969f2e3",
   "metadata": {},
   "source": [
    "## Query 1\n",
    "<div dir=\"rtl\">\n",
    "    تحریم‌های آمریکا علیه ایران\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "e304d886",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************\n",
      "Rank: 169\n",
      "docID: 9742\n",
      "Title: مرکز پژوهش‌های مجلس: مذاکرات وین به توافقی زودهنگام منجر نمی‌شود\n",
      "URL: https://www.farsnews.ir/news/14000924000773/مرکز-پژوهش‌های-مجلس-مذاکرات-وین-به-توافقی-زودهنگام-منجر-نمی‌شود\n",
      "********************\n",
      "********************\n",
      "Rank: 149\n",
      "docID: 6929\n",
      "Title: توضیحات یک منبع آگاه درباره وقفه مذاکرات وین\n",
      "URL: https://www.farsnews.ir/news/14001222000450/توضیحات-یک-منبع-آگاه-درباره-وقفه-مذاکرات-وین\n",
      "********************\n",
      "********************\n",
      "Rank: 107\n",
      "docID: 11864\n",
      "Title: اهرم‌سازی از افغانستان در برجام/ نقطه عزیمت آمریکا در مذاکرات جامع با ایران چیست؟\n",
      "URL: https://www.farsnews.ir/news/14000803000676/اهرم‌سازی-از-افغانستان-در-برجام-نقطه-عزیمت-آمریکا-در-مذاکرات-جامع-با\n",
      "********************\n",
      "********************\n",
      "Rank: 95\n",
      "docID: 9496\n",
      "Title: گفت‌وگوی مشروح | ترقی: آمریکا شروط ایران را نپذیرد، پشت در مذاکرات می‌ماند/ روحانی کشور را به بن‌بست کشاند\n",
      "URL: https://www.farsnews.ir/news/14000926000385/گفت‌وگوی-مشروح-|-ترقی-آمریکا-شروط-ایران-را-نپذیرد-پشت-در-مذاکرات\n",
      "********************\n",
      "********************\n",
      "Rank: 87\n",
      "docID: 10757\n",
      "Title: راهبرد «مذاکره هسته‌ای نداریم» چگونه تأمین کننده منافع کشور خواهد بود؟/ غرب در دوراهی سخت\n",
      "URL: https://www.farsnews.ir/news/14000821000097/راهبرد-مذاکره-هسته‌ای-نداریم-چگونه-تأمین-کننده-منافع-کشور-خواهد-بود-\n",
      "********************\n"
     ]
    }
   ],
   "source": [
    "print_query_result('تحریم‌های آمریکا علیه ایران')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "89d1381f",
   "metadata": {},
   "source": [
    "docID = 9742\n",
    "<div dir=\"rtl\">\n",
    "    با گذشت نزدیک به یک سال از روی‌کار آمدن دولت بایدن، \n",
    "    <u>آمریکا</u>\n",
    "    به برجام بازنگشته و از رفع مؤثر تحریم‌‌ها خودداری می‌‌ورزد. ایده اصلی دولت بایدن این بود که با اعمال «فشار هوشمند» یا همان «ترکیب دیپلماسی و چندجانبه‌گرایی با کمپین فشار حداکثری» \n",
    "    <u>ایران</u>\n",
    "    را وادار به پذیرش «توافق قوی‌تر و طولانی‌تر» کند.  به همین دلیل، دولت بایدن بیش از آنکه به‌دنبال بازگشت فوری به برجام باشد در پی طراحی سازوکاری بود که چگونه می‌‌توان توافق هسته‌ای را به بستری برای دستیابی به تبدیل توافقی جامع برای مهار ایران کرد. تیم مذاکره‌کننده آمریکا با چنین هدفی در فروردین‌ماه 1400 وارد مذاکرات وین شد و پیش‌نویسی را برای گفتگو درباره دو موضوع اصلی مذاکره یعنی «تحریم» و « برنامه هسته‌‌ای» ارائه کرد که طبق آن از یکسو آمریکا با حفظ بخش مهمی از \n",
    "    <u>تحریم‌‌ها</u>\n",
    "    به‌دنبال استمرار اهرم فشار \n",
    "    <u>علیه</u>\n",
    "    ایران و ازسوی دیگر به بهانه «آستانه گریز یک‌ساله» خواستار محدودیت‌‌های فنی و نظارتی فرابرجامی ازجمله تخریب سانتریفیوژهای نسل جدید و زنجیره ساخت این ماشین‌‌ها بود.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a0dfbda",
   "metadata": {},
   "source": [
    "docID = 6929\n",
    "<div dir=\"rtl\">\n",
    "    * <u>تحریم‌های آمریکا علیه ایران</u>\n",
    "\n",
    "آمریکا پس از تسخیر لانه‌ جاسوسی آمریکا در تهران، «کارتر» در تاریخ 8 نوامبر 1979م. با استناد به قانون‌ «کنترل صدور تسلیحات نظامی»، کشتی حامل‌ لوازم یدکی نظامی متعلق به ایران را توقیف کرد. ارزش این لوازم 300 میلیون دلار بود. با اوج‌گیری کشمکش‌های سیاسی بر سر مسئله‌ تصرف سفارت، دولت موقت، اعلام کرد تمام دارایی‌های‌ خود را از بانک‌های آمریکا خارج خواهد کرد. کارتر با لحاظ این احتمال، در کشور شرایط اضطراری اعلام کرد و با استناد به قانون شرایط اضطراری اقتصاد بین‌الملل (IEEPA) و قانون‌ شرایط اضطراری ملی (NEA)، با صدور دستور ویژه‌ی شماره‌ی (12170)، تمام دارایی‌های ایران در آمریکا را به تصرف خود درآورد.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "93f32ac3",
   "metadata": {},
   "source": [
    "docID = 11864\n",
    "<div dir=\"rtl\">\n",
    "    FATF مشکل ما را حل نمی‌کرد بلکه مشکلات را بر ما رسوب می داد چون ما متاثر از \n",
    "    <u>تحریم‌های</u>\n",
    "    اولیه و ثانویه آمریکایی‌ها هستیم که نه به برجام ربطی دارد و نه بهFATF و هیچ سازمان بین المللی. یک تصمیم ملی داخلی ایالات متحده \n",
    "    <u>آمریکا</u>\n",
    "    برای \n",
    "    <u>ایران</u>\n",
    "    است تا تحریم‌ها را موثر بکنند.\n",
    "    <br>\n",
    "    ظهره وند: در آذربایجان هم این فضا را به وجود می آورند. ایران کشوری مادر است و عمق استراتژیک آن در محیط پیرامونی خودش و وسیع‌تر از جغرافیای طبیعی اش هست و امروز مردم منطقه متوجه ظرفیت ها و موقعیت ایران و هم‌راستایی در تهدید و فرصت با ایران هستند. این هم‌راستایی به یک تهدید معتبر \n",
    "    <u>علیه</u>\n",
    "    غرب تبدیل شده و آمریکایی ها اگر نتوانند ایران و چین را در افغانستان متوقف کنند دیگر نمی توانند در محیط داخلی خودشان هم منسجم شوند.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4ba8602",
   "metadata": {},
   "source": [
    "docID = 9496\n",
    "<div dir=\"rtl\">\n",
    "    اکنون در مذاکرات وین، 4+1 به عنوان یک تیم بدعهد و بدهکار  جلوه می‌کند و \n",
    "    <u>ایران</u>\n",
    "    به عنوان کشوری که کاملاً به تمام تعهدات خود عمل کرده و طلبکار است و ابتکار عمل را در دست دارد، در مذاکرات دست بالاتر را به نمایش گذاشته است. به نظر می‌رسد اینها در وضعیت مذاکرات تأثیر قابل توجهی دارد و وقایعی که طی مدت خروج \n",
    "    <u>آمریکا</u>\n",
    "    از برجام رخ داده، برای مردم جهان قابل نادیده گرفتن نیست. طرف مذاکره کننده نمی‌تواند ترور سردار سلیمانی، شهید فخری‌زاده (با جایگاهی که شهید فخری‌زاده در انرژی هسته‌ای و مخالفت با برجام داشت) 4 حمله دشمنان به تأسیسات هسته ای ایران و زوایای متعدد به سایت ها و سانترفیوژها، تشدید \n",
    "    <u>تحریم‌ها</u>\n",
    "    و ادامه تحریم‌ها توسط آمریکا و غرب را نادیده بگیرد. اینها وقایعی است که در طول این مدت رخ داده و این رخدادهای طبیعی، وضعیتی را پدیدآورده که تیم مذاکره کننده غربی با موضع منفعلانه برخورد کند.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d60cc460",
   "metadata": {},
   "source": [
    "docID = 10757\n",
    "<div dir=\"rtl\">\n",
    "    علی رغم اینکه هیچ علامت مثبتی از دولت بایدن مبنی بر پذیرش شروط \n",
    "    <u>ایران</u>\n",
    "    دیده نمی‌شد، دولت حسن روحانی دور جدیدی از مذاکرات هسته‌ای را با اروپایی‌ها موسوم به مذاکرات وین آغاز کرد. مذاکراتی که با هدف احیای برجام صورت گرفت و بنا بود زمینه بازگشت \n",
    "    <u>آمریکا</u>\n",
    "    را به برجام فراهم کند. با این حال بدون اینکه این مذاکرات به نتیجه‌ای برسد، دولت روحانی به پایان رسید.\n",
    "\n",
    "دولت سیّد ابراهیم رئیسی که روی کار آمد اعلام کرد میز مذاکره را ترک نخواهد کرد، با این وجود، زمین بازی را از «مذاکره برای احیای برجام» به «مذاکره برای رفع تحریم» تغییر داد و به موضعی بازگشت که قبلاً به‌عنوان سیاست قطعی نظام اعلام شده بود؛ یعنی رفع همه تحریم‌ها و راستی‌آزمایی از سوی ایران.\n",
    "\n",
    "در راستای همین سیاست است که علی باقری معاون سیاسی وزیر امور خارجه که این روزها در حال  فراهم کردن مقدمات مذاکرات ۸ آذر با اروپایی‌هاست، اعلام می‌کند که زین پس مذاکره هسته‌ای نداریم و آنچه به جریان خواهد افتاد، مذاکره برای رفع \n",
    "    <u>تحریم‌های</u>\n",
    "    ظالمانه \n",
    "    <u>علیه</u>\n",
    "    ایران است. سیاستی که توپ را در زمین طرف غربی مذاکرات قرار می‌دهد. چرا که ایران پس از امضای برجام، دغدغه یا بهانه غربی‌ها مبنی بر صلح‌آمیز بودن برنامه هسته‌ای این کشور را برطرف کرده بود و آژانس بین‌المللی انرژی اتمی نیز با راستی‌آزمایی انجام تعهدات ایران در ۱۵ مرحله بر صحت انجام آن، مهر تأیید زده بود. این اهرمی بود در دست ایران تا با اتکا به آن، طرف مقابل را وادار به رفع تحریم‌ها کند یا حداقل بهانه را از دست آن‌ها خارج کند.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "035f2c92",
   "metadata": {},
   "source": [
    "## Query 2\n",
    "<div dir=\"rtl\">\n",
    "تحریم‌های آمریکا ! ایران\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "8f496f34",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************\n",
      "Rank: 14\n",
      "docID: 7615\n",
      "Title: یک مأموریت‌ فتنه 88 کاهش قدرت چانه‌زنی مسؤولان در مذاکرات بود\n",
      "URL: https://www.farsnews.ir/news/14001129000373/یک-مأموریت‌-فتنه-کاهش-قدرت-چانه‌زنی-مسؤولان-در-مذاکرات-بود\n",
      "********************\n",
      "********************\n",
      "Rank: 13\n",
      "docID: 7261\n",
      "Title: سود مافیای اسلحه‌سازی آمریکا در ناامن بودن جهان است\n",
      "URL: https://www.farsnews.ir/news/14001211000898/سود-مافیای-اسلحه‌سازی-آمریکا-در-ناامن-بودن-جهان-است\n",
      "********************\n",
      "********************\n",
      "Rank: 10\n",
      "docID: 8282\n",
      "Title: خطیب جمعه تهران: سازمان‌ همکاری اسلامی در برابر کشتار مردم یمن سکوت کرده است\n",
      "URL: https://www.farsnews.ir/news/14001108000202/خطیب-جمعه-تهران-سازمان‌-همکاری-اسلامی-در-برابر-کشتار-مردم-یمن-سکوت\n",
      "********************\n",
      "********************\n",
      "Rank: 9\n",
      "docID: 7491\n",
      "Title: مسائل اقتصادی کشور را معطل نتیجه مذاکرات نمی‌‌کنیم\n",
      "URL: https://www.farsnews.ir/news/14001203000383/مسائل-اقتصادی-کشور-را-معطل-نتیجه-مذاکرات-نمی‌‌کنیم\n",
      "********************\n",
      "********************\n",
      "Rank: 7\n",
      "docID: 9994\n",
      "Title: امام جمعه موقت تهران: مذاکره کنونی با گذشته که نقد دادند و نسیه گرفتند، تفاوت دارد\n",
      "URL: https://www.farsnews.ir/news/14000919000137/امام-جمعه-موقت-تهران-مذاکره-کنونی-با-گذشته-که-نقد-دادند-و-نسیه-گرفتند\n",
      "********************\n"
     ]
    }
   ],
   "source": [
    "print_query_result('تحریم‌های آمریکا ! ایران')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba7b9597",
   "metadata": {},
   "source": [
    "docID = 7615\n",
    "<div dir=\"rtl\">\n",
    "    حجت‌الاسلام مجتبی ذوالنور نماینده مردم قم در مجلس، در گفت‌وگو با خبرنگار پارلمانی خبرگزاری فارس، با اشاره به مسأله \n",
    "    <u>تحریم‌های</u>\n",
    "    ضدایرانی در حوزه‌های مختلف و ریشه‌های اصلی وقوع پدیده تحریم اظهار داشت: فلسفه وضع تحریم‌ها، افزایش فشار بر مردم و نظام بوده است تا تاب‌آوری مردم از بین برود و مقاومت آنها بشکند و در نهایت تسلیم شوند.\n",
    "    <br>\n",
    "    نماینده مردم قم در مجلس شورای اسلامی، با اشاره به تاثیر فتنه 88 بر گسترش فشارهای تحریمی تصریح کرد: یکی از مأموریت‌های فتنه 88، کاهش قدرت چانه‌زنی مسؤولان در پای میز مذاکره بود و حتی اطلاعاتی در دست داریم که برخی آقازاده‌ها در آن سال به \n",
    "    <u>آمریکا</u>\n",
    "    رفت و آمد می‌کردند و به غربی‌ها و آمریکایی‌ها پیغام می‌دادند که ما نفس نظام را در داخل گرفته‌ایم و اگر شما هم فشارهای تحریمی را افزایش دهید، کار نظام تمام است.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "773c1d94",
   "metadata": {},
   "source": [
    "docID = 7261\n",
    "<div dir=\"rtl\">\n",
    "    وی افزود: \n",
    "    <u>آمریکا</u>\n",
    "    با سوءاستفاده از شرایط به وجود آمده بین روسیه و اوکراین، کشورهای اروپای شرقی را از روسیه می ترساند تا برای حفاظت از خودشان در برابر حمله احتمالی روسیه، به آمریکا و مافیای حاکم بر آن باج دهند.  \n",
    "    <br>\n",
    "نماینده ادوار مجلس ضمن تاکید به هماهنگ بودن کشورهای غربی گفت: کشورهای غربی به صورت گروهی عمل می کنند و در بزنگاه ها کاملا هماهنگ هستند همچنان که شاهد هستیم در جریان نبرد میان روسیه و اوکراین در اوج اتحاد \n",
    "    <u>تحریم هایی</u>\n",
    "    را به روسیه تحمیل کردند.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ca211084",
   "metadata": {},
   "source": [
    "docID = 8282\n",
    "<div dir=\"rtl\">\n",
    "    وی به مذاکرات لغو \n",
    "    <u>تحریم‌ها</u>\n",
    "    هم اشاره کرد و گفت: آمریکا در پی غارتگری است، نه مذاکره؛ \n",
    "    <u>آمریکا</u>\n",
    "    خواب و خیال غارتگری قبل از انقلاب را دارد و این خواب هرگز تعبیر نشده و نخواهد شد.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a39cad67",
   "metadata": {},
   "source": [
    "docID = 7491\n",
    "<div dir=\"rtl\">\n",
    "    وی افزود: ما باید بدانیم که \n",
    "    <u>آمریکا</u>\n",
    "    به عنوان یک دشمن همواره در صدد ضربه زدن به کشور، نظام و ملت ماست و طبیعتا بنایش برای آزادسازی \n",
    "    <u>تحریم‌ها</u>\n",
    "    و ایجاد گشایش اقتصادی در کشور ما نخواهد بود.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "00581d6c",
   "metadata": {},
   "source": [
    "docID = 9994\n",
    "<div dir=\"rtl\">\n",
    "    وی در ادامه با بیان اینکه \n",
    "    <u>آمریکا</u>\n",
    "    در دوره‌های مختلف و در زمان ترامپ و قبل از او در سه جبهه علیه نظام جمهوری اسلامی توطئه کرد، افزود: این سه جبهه شامل وزارت خزانه‌داری آمریکا که مرتب \n",
    "        <u>تحریم‌های</u>\n",
    "        به قول خودشان فلج‌کننده را آماده می‌کند، وزارت خارجه آمریکا که با مسافرت‌های مختلف دولت‌ها و شرکت‌ها را از معامله با ما مرعوب می‌کنند و نیز سازمان سیا است که از ابتدای انقلاب تاکنون در صدد ایجاد نارضایتی در کشور بوده است.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6d68648",
   "metadata": {},
   "source": [
    "# Query 3\n",
    "<div dir=\"rtl\">\n",
    "    کنگره ضدتروریست\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "b3bf16aa",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********************\n",
      "Rank: 5\n",
      "docID: 6929\n",
      "Title: توضیحات یک منبع آگاه درباره وقفه مذاکرات وین\n",
      "URL: https://www.farsnews.ir/news/14001222000450/توضیحات-یک-منبع-آگاه-درباره-وقفه-مذاکرات-وین\n",
      "********************\n"
     ]
    }
   ],
   "source": [
    "print_query_result('کنگره ضدتروریست')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5bb5676",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    ریگان که نمی‌خواست کمتر از \n",
    "    <u>کنگره ضدتروریست</u>\n",
    "    جلوه کند، 3هفته بعد با صدور دستور ویژه‌ی (12613) ورود هرگونه کالا و خدمات از ایران را ممنوع کرد. وی برای صدور این دستور به بند 505 «قانون همکاری‌های‌ بین‌المللی امنیتی و توسعه»، مصوب سال‌ 1985م. استناد کرد. در واقع زمانی که ایران درگیر جنگ تحمیلی بود، ایالات متحده به تعریف سیاست‌های تحریمی پرداخت که بتواند بر نتیجه‌ی جنگ ایران و عراق تأثیری شگرف بگذارد و رقیب نوپای اسلامی خود را در جنگی نابرابر از میدان به در کند. اما پایان جنگ، تافته‌ی بافته‌ی آمریکا را ریش ریش کرد.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "382ee35d",
   "metadata": {},
   "source": [
    "## Query 4\n",
    "<div dir=\"rtl\">\n",
    "    \"تحریم هسته‌ای\" آمریکا ! ایران\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "338e1d22",
   "metadata": {},
   "outputs": [],
   "source": [
    "print_query_result('\"تحریم هسته‌ای\" آمریکا ! ایران')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4397061",
   "metadata": {},
   "source": [
    "there is no output for this query"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81739a08",
   "metadata": {},
   "source": [
    "## Query 5\n",
    "<div dir=\"rtl\">\n",
    "    اورشلیم ! صهیونیست\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "1b3c3162",
   "metadata": {},
   "outputs": [],
   "source": [
    "print_query_result('اورشلیم ! صهیونیست')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7878a88b",
   "metadata": {},
   "source": [
    "there is no output for this query"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "57c106e5",
   "metadata": {},
   "source": [
    "# Zipf Law\n",
    "در این قسمت، می خواهیم رابطه بین تکرار توکن ها در کل اسناد و رتبه توکن بپردازیم\n",
    "$$\\log cf_i = \\log K - \\log i$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2043c974",
   "metadata": {},
   "outputs": [],
   "source": [
    "def token_incidence(docs):\n",
    "    token_inc = dict()\n",
    "    for body in docs.values():\n",
    "        for token in body['content']:\n",
    "            if token_inc.get(token):\n",
    "                token_inc[token] += 1\n",
    "            else: \n",
    "                token_inc[token] = 1\n",
    "    \n",
    "    return [(token, inc) for token, inc in token_inc.items()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "33c0457a",
   "metadata": {},
   "outputs": [],
   "source": [
    "dictionary_with_stopwords = token_incidence(preprocess_contents(load_docs(docs_path), stopwords_remove=False))\n",
    "dictionary_without_stopwords = token_incidence(preprocess_contents(load_docs(docs_path), stopwords_remove=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "4bdb9620",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEaCAYAAAAcz1CnAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABCBElEQVR4nO3dd1zVZfvA8c/FEJyICu69FcGBouLWcqSZo6FmjtQy59O2PX75VGZpaVnmKDPLkSNz5Az3Jje5FSeSuJV1//74Ig+aJuMczoFzvV8vXnG+nO99X4fwXOfeYoxBKaWU63JzdABKKaUcSxOBUkq5OE0ESinl4jQRKKWUi9NEoJRSLk4TgVJKuThNBMrliMhiEemVyucWFpEwEbksIqPtHVtqiEgzEYl0dBwq+9BEoLIVEekhIlfu8mVE5C0AY0xbY8x3qSxyAHAeyGeMeeEedQaLyEIRuSAiMSKyV0Q+EBFfG70spexKE4HKVowx040xeVJ+AcOBs8DEdBRZGthr7rHyUkQaAquBdUAVY0x+oA0QDwTd4x6PdMShlN1oIlDZmojUAsYATxhjTiddWy0i/ZK+7y0i60RknIhcFJH9ItIy6WdTgV7Ay0mtilZ3qeJjYIox5r/GmLMAxpjjxpi3jTGr76jjMxGJBt4RkfIislJEokXkvIhMF5H8KeI+KiIjkloXF0Rkioh43/HaXhCRcyJyWkT62PQXp1yKJgKVbSW9sc4G3r/1pnwPIcAhoBDwNvCLiBQwxvQGpgMfJ7Uult9Rfm6gATAnFeGEAIeBwsAHgAD/BYoBVYGSwDt33NMDaA2UByoBb6T4WRHABygOPA2M164olV6aCFS2JCICfA/sxvrU/m/OAWOMMXHGmJ+BCOChVFTji/Vv6EyKej9OGie4KiIp37hPGWO+MMbEG2OuG2MOGmOWGWNuGmOigE+BpneUP84Yc8IY8zdW8uiW4mdxwHtJMS8CrgCVUxGzUv+giUBlV68A1YFe9+rfT+HkHc85hvVJ/X4uAIlA0VsXjDEvJ40TzAVSjgWcSHlj0mykn0TkpIhcAn7AapFwj3vujCnaGBOf4vE1IE8qYlbqHzQRqGxHRJoBrwNdjTExqbileFIL4pZSwKn73WSMuQpsAjqnoo47k9HIpGs1jDH5gCexuotSKpnWmJRKD00EKlsRkaLAT8BwY8yOVN7mDwwVEU8ReRSrz35RKu99GegrIq+KiH9SDCWAsve5Ly9Wd85FESkOvHSX5wwSkRIiUgArsf2cypiUShNNBCq76Y81IDv2LmsJJtzjnk1ARaz1Ah9gtSSiU1OZMWYt0AJoAvwlIjHAEqwppV/8y63vArWBi8BvwC93ec6PwO9Yg8yHgP9LTUxKpZXowTTKlYlIb6CfMaaRo2NJSUSOYsW1/H7PVSqjtEWglFIuThOBUkq5OO0aUkopF6ctAqWUcnFZcvOrQoUKmTJlyjg6DKWUylK2bdt23hjjd+f1LJkIypQpw9atWx0dhlJKZSkicuxu17VrSCmlXJwmAqWUcnGaCJRSysVlyTECpZxFXFwckZGR3Lhxw9GhKJXM29ubEiVK4OnpmarnayJQKgMiIyPJmzcvZcqU4fYNTJVyDGMM0dHRREZGUrbs/fY+tGjXkFIZcOPGDQoWLKhJQDkNEaFgwYJpaqVqIlAqgzQJKGeT1r9Jl+oaWrn/LIfOXaVB+YJULZoPdzf9B6yUUi7VIlgdEcUHi/bR/ou11H5/GQO+38qUdUeIOHMZ3XNJZVXu7u7UrFmT6tWrExQUxOjRo0lMTARg69atDB069F/vHzly5G2PGzZsaLdYbWnBggV8+OGHAMybN4+9e/cm/6xZs2b3XXSamJjI0KFDCQgIoEaNGtStW5cjR44A//ydOIOjR48SEBBgl7Kz5KZzwcHBJr0ri89cvMHGw9GsP3SeDYejOfH3dQAK5s5B/XIFqV++IA3KFaS8X25t8qv72rdvH1WrVnVoDHny5OHKlSsAnDt3ju7duxMaGsq7776b5vszQ3x8PB4etu2M6N27N+3bt6dr166AlQg++eQTgoOD73nPjBkzmDNnDjNnzsTNzY3IyEhy586Nr69vpv9O7ubO39PRo0dp3749u3fvTtX9d/vbFJFtxph//FIyrUUgIpNF5JyI7E5xbZSI7BeRnSIyV0Ty2zuOIj7ePFKrOB93DWLNyy1Y+0pzRnUNpGklP7Yfv8Cb83bT6tM/CBm5gmE/7eCnzcc5Fn1VWwwqS/D39+ebb75h3LhxGGNYvXo17du3B+DKlSv06dOHGjVqEBgYyJw5c3j11Ve5fv06NWvWpEePHoCVGMCaffLSSy8lf2L++WfrpMzVq1fTrFkzunbtSpUqVejRo0fyv4/33nuPunXrEhAQwIABA5KvN2vWjOHDhxMcHMwHH3xA2bJliYuLA+DSpUu3PQZISEigbNmyGGOIiYnB3d2dsLAwAJo0acKBAweYOnUqgwcPZv369SxYsICXXnqJmjVrcujQIQBmzZpFvXr1qFSpEmvWrPnH7+r06dMULVoUNzfrbbBEiRL4+vre9Xfy6aefEhAQQEBAAGPGjAGsN+Zbr79q1ap07dqVa9eusWXLFjp3to6xnj9/Pjlz5iQ2NpYbN25Qrlw5AMLDw6lfvz6BgYF06tSJCxcu/OP3NHbsWLZt20ZQUBBBQUGMHz8+OfY9e/ZQr149atasSWBgIAcOHEjnX4wlM8cIpgLjgO9TXFsGjDDGxIvIR8AI4BW7RRB7DTCQI3fypRK+uXg0OBePBpfEGMOx6GtsOBzN+kPRrDsYzfxw67zwYj7eNChfiAblC9KgfEGK589ptzBV1vTur3vYe+qSTcusViwfb3eonqZ7ypUrR0JCAufOnbvt+vvvv4+Pjw+7du0C4MKFC3Tp0oVx48YRHh7+j3J++eUXwsPD+fPPPzl//jx169alSZMmAOzYsYM9e/ZQrFgxQkNDWbduHY0aNWLw4MG89dZbAPTs2ZOFCxfSoUMHAGJjY5O7a44ePcpvv/3GI488wk8//UTnzp1vm/Pu7u5O5cqV2bt3L0eOHKF27dqsWbOGkJAQTpw4QcWKFVm3bh1gdWU9/PDDt7UIwPpEvXnzZhYtWsS7777L8uW3H/b22GOP0ahRI9asWUPLli158sknqVWrFh9++OFtv5Nt27YxZcoUNm3ahDGGkJAQmjZtiq+vLxEREUyaNInQ0FD69u3Ll19+yfDhw5PvXbNmDQEBAWzZsoX4+HhCQkIAeOqpp/jiiy9o2rQpb731Fu+++25ygkn5ewoMDGTcuHE0adKEl17637HWEyZMYNiwYfTo0YPY2FgSEhJS8Zdxb5nWIjDGhAF/33Htd2NMfNLDjUAJuwax6gP4sj4cvPvpfyJCmUK56VavFF90q8WW11uy/PkmvN+xOkEl87Ny/1lenPUnoR+upM2YMMYuP8DBc5ftGrJStrJ8+XIGDRqU/NjX1/dfn7927Vq6deuGu7s7hQsXpmnTpmzZsgWAevXqUaJECdzc3KhZsyZHjx4FYNWqVYSEhFCjRg1WrlzJnj17kst7/PHHk7/v168fU6ZMAWDKlCn06dPnH/U3btyYsLAwwsLCGDFiBGvXrmXLli3UrVs3Va/31qfyOnXqJMeXUokSJYiIiOC///0vbm5utGzZkhUrVtz199CpUydy585Nnjx56Ny5c3ILo2TJkoSGhgLw5JNPsnbtWjw8PChfvjz79u1j8+bNPP/884SFhbFmzRoaN27MxYsXiYmJoWnTpgD06tUrubWT8vcUExNDTExMcvLt2bNn8nMaNGjAyJEj+eijjzh27Bg5c2bsg6kzzRrqC/x8rx+KyABgAECpUqXSV0PldvDXUvihC9R4DFqPhDz/2JE1ZZ1U8M9LBf+89GxQhsREQ8TZy6w7eJ6le84wZsVffLb8Lyr656FdjaK0q1GUSoXz6NiCi0rrJ3d7OXz4MO7u7vj7+7Nv3z671OHl5ZX8vbu7O/Hx8dy4cYPnnnuOrVu3UrJkSd55553b5rLnzv2/lnhoaChHjx5l9erVJCQk3HUQtEmTJnz11VecOnWK9957j1GjRrF69WoaN26cphhvxXev57Rt25a2bdtSuHBh5s2bR8uWLVNVPvxzmuatx02aNGHx4sV4enrSqlUrevfuTUJCAqNGjbpvmSl/T/fSvXt3QkJC+O2332jXrh1ff/01LVq0SHXcd3KKWUMi8joQD0y/13OMMd8YY4KNMcF+fvd+8/5XZUJh4Dpo+grsmQvj60L4j5DK/n83N6Fq0Xz0a1yOWc82ZOOIlrz7cHUK5M7B5ysP0HpMGK0+/YPRv0ew7/QlHVdQmS4qKopnn32WwYMH/+NN6oEHHritn/lWv7Snp+dt/fO3NG7cmJ9//pmEhASioqIICwujXr1696z71pt+oUKFuHLlCrNnz/7XWJ966im6d+9+19YAWK2O9evX4+bmhre3NzVr1uTrr79O/oScUt68ebl8OW2t8+3bt3PqlNX1m5iYyM6dOyldujRw+++kcePGzJs3j2vXrnH16lXmzp2bnIyOHz/Ohg0bAPjxxx9p1KhR8j1jxoyhQYMG+Pn5ER0dTUREBAEBAfj4+ODr65vcqpg2bVpy6yCl/Pnzkz9/ftauXQvA9On/e3s8fPgw5cqVY+jQoXTs2JGdO3em6bXfyeGJQER6A+2BHiYz3jk9vKD5a/DsWihUGeYNhO87QvShNBdVOJ83vRqW4ednGrDptZa8/0gA/nm9Gb/qIG3HrqHl6D+YvPaIJgRlV7cGNqtXr06rVq148MEHefvtt//xvDfeeIMLFy4QEBBAUFAQq1atAmDAgAEEBgYmD4ze0qlTJwIDAwkKCqJFixZ8/PHHFClS5J5x5M+fn/79+xMQEEDr1q3v24XTo0cPLly4QLdu3e76cy8vL0qWLEn9+vUB68318uXL1KhR4x/PfeKJJxg1ahS1atVKHiy+n3PnztGhQwcCAgIIDAzEw8ODwYMHA7f/TmrXrk3v3r2pV68eISEh9OvXj1q1agFQuXJlxo8fT9WqVblw4QIDBw4EICQkhLNnzyYnrcDAQGrUqJGcnL/77jteeuklAgMDCQ8PTx5XudOUKVMYNGgQNWvWvO19ZObMmQQEBFCzZk12797NU089larXfC+ZOn1URMoAC40xAUmP2wCfAk2NMVGpLScj00dvk5gI26bA8ncgIdZqKTQcAu6p26jpXs5fucnSPWeYv+MUm4/+zYPVCjP6sSDyemesXOV8nGH6aFY1e/Zs5s+fz7Rp0xwdSrqkdTpnZnPW6aMzgA1AZRGJFJGnsWYR5QWWiUi4iEzIrHgAcHODuk/DoE1Q8QFY8S580wwit2Wo2EJ5vOgRUpqfn6nPGw9VZcX+c3Qct44DZ3VgWSmAIUOG8Oqrr/Lmm286OhSFCy4o+1f7FsKil+DyaQh5Flq8Dl55M1zsxsPRDP5xO9diExjVNYiHAovaIFjlDLRFoJyVU7YIsoSq7a3WQd1+sGkCjK8PEUsyXGz9cgVZOKQxVYrkZdCP2/ngt73EJyTaIGCllMo4TQR38s4HD30CfZdarYEZj8Os3nD5bIaKLeLjzU8DGvBUg9JMXHOEJydt4vyVm7aJWSmlMkATwb2UCoFnwqD5G7D/N2uq6bap1gBzOuXwcOO9jgGMfjSIHcdjaDMmjJGL9rH75EWdWaSUchhNBP/GIwc0fQkGrofCNeDXYfBdezifsX09utQpwS/PNSSoRH4mrz1C+y/W0nL0H3y27C8ORTl2oyullOvRRJAahSpCr1+hw+dwdjd81RD++BjiY9NdZPViPkzqXZctr7diZKca+Ofz4vOVB2g5+g9afxbGfxftY93B89yMz9geIir7u7VJ3J169+5930Vd9/LOO+/wySef3PVnP/zwA4GBgcnbXvfr14+YmJh01XOnWxvJqczlTFtMODc3N6jTCyq1gSWvWvsW7Z5jJYdSIeku1jd3DrqHlKJ7SCnOXLzBb7tOs3L/WaasO8rXYYfJ6elOw/IF6RBUjDYBRfD2dLfhi1IqbZYsWcJnn33G4sWLKV68OAkJCXz33XecPXuW/Pnz3/bchIQE3N317zUr0BZBWuUtDI9Oge4z4eYVmNwaFj4PNy5muOgiPt483ags0/vVZ8dbDzCpVzCPBpdg/5nLDP85nHofLOfNebsJPxHDjThtKajbGWMYPHgwlStXplWrVrftPrpt2zaaNm1KnTp1aN26NadPnwZg4sSJ1K1bl6CgILp06cK1a9f+tY4PPviATz75hOLFiwPWPj59+/alcuXKAJQpU4ZXXnmF2rVrM2vWrHuW37t3b5599lmCg4OpVKkSCxcuTK7j1KlTtGnThooVK/Lyyy/b9Hek7k5bBOlVqbU11XTl/1lTTSMWQbtRULWDTYrP7eVBy6qFaVm1MO90MGw8HM3MrSeYufUE0zYeA8AnpydFfbx5qkEZuoekcyM+ZTuLX4Uzu2xbZpEa0PbDVD117ty5REREsHfvXs6ePUu1atXo27cvcXFxDBkyhPnz5+Pn58fPP//M66+/zuTJk+ncuTP9+/cHrC0oJk2axJAhQ+5Zx549e6hdu/a/xlGwYEG2b98OQHR09D3LP3r0KJs3b+bQoUM0b96cgwcPAtZe/Tt27MDLy4vKlSszZMgQSpYsmarfgUofTQQZ4ZXH+kda41H4dSj8/CRUaW8lhHzFbFaNm5vQsEIhGlYoxLvX41gdcY7IC9c5c/EG4SdieGPeLsr55aZ+uYI2q1NlPWFhYcnbRhcrVix5N8qIiAh2797NAw88AFhdNkWLWosad+/ezRtvvEFMTAxXrlyhdevWqa5v165d9OzZk8uXLzNy5Mjk7ZNTbjf9b+U/9thjuLm5UbFiRcqVK8f+/fsBaNmyJT4+PgBUq1aNY8eOaSKwM00EtlCiDgxYDRvGweoPYVw9aPU2BD9tjS3YkE9OTzrWLJ78+MrNeB7+Yi3DftrBoqGNKZjH61/uVnaVyk/umc0YQ/Xq1ZN3yUypd+/ezJs3j6CgIKZOncrq1av/tazq1auzfft2mjdvTo0aNQgPD2fw4MFcv349+Tkpt1H+t/LvtYXz3ba4VvalYwS24u4Jjf5jTTUtXhsWvQhT2sA5++wHf0seLw++6F6LC9fieH7mnyQm6noEV9WkSZPkbaNPnz6dvLto5cqViYqKSk4EcXFxyQfGXL58maJFixIXF3fbNsf3MmLECF588UUiIyOTr6VMAnf6t/JnzZpFYmIihw4d4vDhw8njDCrzaYvA1gqWh6fmw58zYOlrMKGxlSAavwCe3napsnoxH95sX4035+3mo6X7eaV1Fdzc9HAcV9OpUydWrlxJtWrVKFWqFA0aNAAgR44czJ49m6FDh3Lx4kXi4+MZPnw41atX5/333yckJAQ/Pz9CQkLuu6d/u3btiIqKom3btiQkJJA/f/7kbafv5t/KL1WqFPXq1ePSpUtMmDABb2/7/PtQ96ebztnT1fOwZATsmgkFK0KHsdbhOHZgjGHEL7v4acsJWlX1Z/RjNfHJqdte25tuOpc+vXv3/scZw8q2dNM5Z5G7EHSZCE/OgYSbMLUdLBgC1y/YvCoR4b+da/BOh2qsjoii47i1nL104/43KqVcniaCzFChFTy30Tr0ZscP1mDy7l9SfURmaokIvUPLMmNAfU5cuM5364/atHylbGXq1KnaGnAimggyS47c8OD/Qf9VkK8ozO4DM56AmBM2r6pumQI0q+THnO2Rut11JsiK3asqe0vr36QmgsxWrCb0WwkPfgBHwuDL+rBxAiTadqXwo8ElOXvpJmsOnLdpuep23t7eREdHazJQTsMYQ3R0dJoG33XWkCO4e0DDwdZBOAufhyWvWAPKHT6HIgE2qaJFFX8K5s7BzK0naF7F3yZlqn8qUaIEkZGRREWl+shtpezO29ubEiVKpPr5mggcybeMNZC8a7a1kd03Ta1xhKavgGfODBWdw8ONR2oV5/sNR/n7aiwFcuewTczqNp6enpQtW9bRYSiVIdo15GgiEPgoDN4CgY/D2s/gywZweHWGi34suCRxCYYvVh5g/cHz/HkihoPnLnPlpq7UVEr9j64jcDaH/4CFw+Hvw1CzhzXAnKtAuovr/OU6th+Pue1aXm8PpvSuS3CZ9JerlMp67rWOQBOBM4q7bh18s/5z8M4PbT6EGl2t1kMaXbkZz5Goq1yNjedabDyXb8QzdvkBzly6wbe9gmlYvpDt41dKOSVNBFnRmd3WrqYnt1lrER76FHxLZ7jYc5dv8OS3mzgWfY2ve9ahWWUdTFbKFTh8ZbGITBaRcyKyO8W1AiKyTEQOJP3XN7PiyRKKBMDTy6Dtx3B8ozXVdP0XkJCxPn7/vN78NKAB5f3yMOD7bSzfe9ZGASulsqLMHCyeCrS549qrwApjTEVgRdJjlZKbO4Q8Yx2CU7Yp/P4GfNsCToVnqNgCuXMwo399qhbNy8Dp21i1/9z9b1JKZUuZlgiMMWHA33dc7gh8l/T9d8AjmRVPluNTArrNgEe/g8tnYGILWPo6xF5Nf5G5PPm+bwiVCuflmWnbeP7ncP48EWO7mJVSWYKjp48WNsacTvr+DFD4Xk8UkQEislVEtrrs4h0RqP6I1Tqo9aR1EM6X9eHg8nQX6ZPLkx+eDqFLneIs33eWx7/ZwNajd+ZrpVR2lqmDxSJSBlhojAlIehxjjMmf4ucXjDH3HSdwmcHi+zm6Dn4dBtEHoMZj0Oa/1o6n6XT+yk0em7CBM5duUM4vNw3LF2Jg0/L46mI0pbIFhw8W38NZESkKkPRf7ahOizKhMHCdtRJ5z1wYFwzhP6Z7V9NCebz4/ul6tAkogk9OTyauOUyL0atZuV8Hk5XKzhydCBYAvZK+7wXMd2AsWZOHFzR/DZ5dC4UqwbyB8H1HiD6UruJK+Obi08dqMr1ffRYPa0wRn5z0nbqV9xfu5VqsrkhWKjvKtK4hEZkBNAMKAWeBt4F5wEygFHAMeMwYc98Oau0auofERNg2BZa/AwmxVkuh4RDrPOV0uhGXwMhF+/h+wzF8c3nSJ7QsjSoWolBuL4r75sRdj8RUKsvQBWWu5NIpWPwy7PsVCgdYu5qWqJOhIrcd+5vxqw6xMsU0U7+8XgxrWZFu9UppQlAqC9BE4Ir2LYRFL1rTTUOehRavg1feDBV55PxVjp6/ytlLN5gXfpKNh//GTaCoT04eDS7BUw3K6E6nSjkpTQSu6sZFWPEebJkE+YrDQ6Oh8p3r+tInMdGw4M9THIq6wq6TF1kdEUWhPF4Mbl6ecn55aFyxEJKO/ZGUUvahicDVHd9kTTWN2gfVO0GbjyDvPZdtpMveU5d49odtHP/7GgAV/fPQoHxBgssU4IGqhcmZw92m9Sml0kYTgYL4WFg3FsI+tg6+eeA9qPUUuNlu8lhCoiH66k1W7T/Hr3+eZvvxC1yLTaC8X24eqlGU2qV9dZM7pRxEE4H6n/MH4NfhcGwtlA6F9mPAr5JdqopPSGR1RBQfLdnPgXNXEIEhLSrSr3FZ8nmnfzaTUirtNBGo2yUmwo5psOxN6/yDJi9B6HDwsN9A77XYeIbOCGf5vrM0KFeQ75+uh6e7o5eyKOU6nHVlsXIUNzeo0wsGbYEq7WHVB/B1Y2sswU5y5fDg217BvNS6MhsOR9Nr8mYSErPeBxGlshtNBK4ub2F4dAp0nwk3r8Dk1rDweWu2kZ0816w8L7WuzPpD0Xy6LIKL1+LsVpdS6v40EShLpdbWrqYhz8LWyTA+xFqQZgciwsCm5QkpW4Dxqw7RdmwYJ2Ou26UupdT96RiB+qfIbdYRmWd3W91G7UZBvmI2ryYh0fDHX+cYNH0H1+OsmUVVi+bjgWqFaRtQlBwe+jlFKVvSwWKVNglx1rGYf3wEbp7Q6m0IftqmU01viThzmcW7T7Mr8iJ/Rl7k/JWbhFYoyKRedfH21LUHStmKJgKVPtGHYOF/4MgfUDIEOowF/6p2qy4x0TBm+V98vvIghfN58XLrKrSs6k/+XLpthVIZpYlApZ8x8OdPsHSENaDc6D/Q+AXw9LZTdYbl+87xydIIIs5exsNNmNS7Lk0r+dmlPqVchSYClXFXz8PS12Dnz1CwotU6KBNqt+oSEg1bj/7NkBk7SEg0tKtRlJfbVCavLkRTKl10HYHKuNyFoPM38OQcSLgJU9vBgiFw/YJdqnN3E0LKFeTLHrWpWjQf0zYe47W5u/n7aqxd6lPKVWmLQKVP7FVY/SFsGA+5CkLbj6zN7Oy42+jTU7ewYv853N2EUgVy0bKKP081KEOpgrnsVqdS2Yl2DSn7OP2n1So4/SdUagPtPoH8Je1SVXxCYvJ214t3n+avs1fw9nSjYflC1C6Vn76NypIrh4dd6lYqO9BEoOwnIR42TbC2qRA3aPEm1OsPbvad+nky5jpfrjrIin3nOHPpBjnc3ejZoDQtq/rToFxBPQtBqTtoIlD2d+EY/PY8HFwOxetYR2QWCbB7tcYYVkdEMXndEdYdPE+igZ71S/P+I/avW6msRAeLlf35loYes6HLJCspfNMUlr9j7W5qRyJC8yr+THs6hPC3H6RpJT+mbTzGvB0nyYofdJTKbJoIlG2JQI2uMHgLBD4Baz+DLxvA4dWZUn0+b08+7FKDakXzMfzncNqMWcO0DUeJjU/MlPqVyoo0ESj7yFUAHhkPTy2wksP3HWHuQLj2t92rLuqTk9kDG/BS68okGMOb8/fQbeJGth37m7gETQhK3UnHCJT9xV2HsFHWMZne+aHNh1arIRMGc40xzAs/ybu/7iXmWhzFfLwZ0rIinWoV132MlMtx6sFiEfkP0A8wwC6gjzHmxr2er4kgizqzG34dBie3QvmW0P5T8C2TKVXHXIvlj7+iGLv8AIfPX8UnpyevtavC43VLZUr9SjkDp00EIlIcWAtUM8ZcF5GZwCJjzNR73aOJIAtLTIAtk2DFu2ASoflrEDIQ3DNn/v/N+ASW7jnLmOV/cTjqKo/ULMZbHapTILduaqeyP2efNeQB5BQRDyAXcMrB8Sh7cXOHkAHWIThlm8Lvb8C3LeBUeKZU7+XhzsNBxfh1cCMaVyzEvPBThIxczqfL/uJGXEKmxKCUs3F4iwBARIYBHwDXgd+NMT3u8pwBwACAUqVK1Tl27FjmBqlszxjYtwAWvWRtaNfgOWg2AnLkzrQQ9p+5xKe//8Xve8+SO4c7g1pU4Jkm5XF308VoKvtx5q4hX2AO8DgQA8wCZhtjfrjXPdo1lM1cj7HWG2ybAvlLQfvPoEKrTKveGEPYgfNM33iM3/eepUzBXHSsWZy+jcrik1N3OlXZhzN3DbUCjhhjoowxccAvQEMHx6QyU8780GEM9FkMHt7wQxeY0x+uRGVK9SJC00p+fN2zDl90q0XJArkYu+IADf67gs9XHNBFaSrbc4YWQQgwGaiL1TU0FdhqjPniXvdoiyAbi78Jaz6FNaPBKw+0HglB3TJlqmlKf56IYeyKA6zcf46K/nnoE1qWdjWK6ElpKktz2q4hABF5F6trKB7YAfQzxty81/M1EbiAqAhYMBRObISyTaD9GChYPlNDSEg0zNp6gklrj3Dg3BXc3YQG5QrSqqo/HWsWx1dnGqksxqkTQVppInARiYmwfSosexsSYqHpK9BwCLhnbr+9MYadkRf5fe8Z5oefIvLCdYrnz8mb7avyYLUiuOnAssoiNBGorOvSaVj8sjXDqHCAtatpiToOCcUYw+q/ohj52z4OnLtCQPF8/KdVJVpU8ddtr5XT00Sgsr79v8FvL8Ll0xDyLLR4HbzyOiSUuIREZm2NZPTvEURfjSW4tC8fdqlBBX/HxKNUamgiUNnDjUuw8n3YPBHyFYeHRkPlNo4LJy6B+eEn+WhJBJdvxNGxZnHefKgaPrl02qlyPs48fVSp1PPOB+1GwdO/W62BGY/DrN5w+axjwvF05/G6pVgyrDHd65Vi7o6TdBi3lj/+itJppyrL0ESgsqaS9eCZMGjxBuxfBOPrwrap1gCzA/jn8+bdjgHMfKY+icbQa/Jmeny7iVMx9j2URylb0K4hlfWdPwgLh8PRNVA61Jpq6lfJYeHciEtgxubjfLwkAjeBvo3KMqh5Bd32Wjmcdg2p7KtQBej1Kzw8Ds7ugQmh8MfHEB/rkHC8Pd3pE1qW3//ThKaV/fhi5UGe/m4LZy7ec2d1pRxKE4HKHkSgdk/riMyqHWDVB/B1Yzi+yWEhlSyQiy971OG9jtXZcuQCD32+hi1H7X9Cm1JppYlAZS95/KHrZOg+C2KvwuQHYeHzcOOiw0J6qkEZFg5thLenO49O2MDrc3fpltfKqWgiUNlTpQfhuY1Qf5C1q+n4ENj3q+PCKZyXZc83oV+jskzfdJyHx60l7K/M2VRPqfvRRKCyL6880GYk9FsBuQvBz0/CTz3gkmPOPcqVw4M32ldjUq9g4hIMT03ezIhfdnLlZrxD4lHqFk0EKvsrXhv6r4IH3oODK2BcPWtBmoOmmrasWphFQxvTr1FZftpygrZjw1h/8LxDYlEKUpEIRGRs0n9z2j8cpezE3RNCh8Fz6619iha9CFPawLl9DgknZw533mhfje/71iMxEbp/u4kRv+zi/JV7brqrlN2kpkXQJOm/a+0ZiFKZokA56DkPOn0N5w/AhMaw8gOIc8zUzsYV/VjxQlP6hJZh1tYTPPzFWnZGxjgkFuW6UpMIVojIBqCIiPQVkToi4mXvwJSyGxEIesKaahrQBcI+hgmN4Og6h4Tj7enO2x2qM2dgQ2ITDJ2+XM+opfuJT3BM15VyPfdNBMaYF4EngQSgLPAmsFtE9ojIz3aOTyn7yV0IOn8NPeda5x1MbQcLhsD1Cw4JJ6hkfla80JROtYozftUh+n+/lb+vOmZRnHItqd5iQkQqGWP+SvE4DxBgjNlor+DuRbeYUDYXew1W/xc2jIdcBaHtR1C9U6YfkXnLtA1HeX/hPnxze/LZ4zVpWL6QQ+JQ2YsttpjIJyK5bz0wxlxxRBJQyi5y5IIH34cBqyBfMZjdB2Y8ATEnHBJOzwZl+OW5huT28qDHt5v4eMl+bsbrIjRlH2lJBN9jdQ8BICKFRKS97UNSyoGKBlnrDlqPhCNh8GV92DgBEjP/TTiguA8LhzTi8eCSfLn6EO3GriH8REymx6Gyv7QkghvGmOSpFcaY88B7tg9JKQdz94AGg6yVyaXqw5JXYNIDcGZ3poeSK4cHH3YJZGqfulyPTaDrV+v5bv1R4nQgWdlQWhLBYRFpe8e1HLYMRimn4lsaesyGLpMg5jh80xSWvwNxmX/GQLPK/vw2tDF1yxTg7QV7tHWgbCotg8VFgcXAXmAjUB0oYozpaL/w7k4Hi1Wmu/Y3LHsTdvwAvmWhwxgo1yzTwzDGsGT3Gd75dQ9/X41laIuKPNe8Au5ujhnUVllLhgeLjTGngTrAHMAP2Al0t1mESjmzXAWg43jr3AMR+L4jzB1oJYhMJCK0rVGUpcOb0LSSH6OX/cUz07bpQLLKEKc4oUxE8gPfAgGAAfoaYzbc6/naIlAOFXcdwkbBurHgnR/afAg1ujpkqunktUd4b+FeWlbx5/Nutcjt5ZHpMaisw9lPKBsLLDHGVAGCAMdsAKNUanjmhJZvWWcm+5aBX/rBD13gwtFMD6Vvo7K83aEaqyLO8fg3G4i6rHsVqbRLyxjB83e5fBHYZowJT3cAIj5AOFDOpDIYbREop5GYAFsmwYp3wSRC89cgZKA18ygT/b7nDEN/2kEF/zxM6xuCb26dx6H+yRYtgmDgWaB40tczQBtgooi8nIHYygJRwBQR2SEi36ZcuHaLiAwQka0isjUqSg/0UE7CzR1CBsCgTdbg8e9vwLct4FR4pobxYPUifNGtNhFnLvPw+LVsP+6YbTJU1pSWRFACqG2MecEY8wLWwLE/1u6kvTMQgwdQG/jKGFMLuAq8eueTjDHfGGOCjTHBfn5+GahOKTvwKQFP/AiPfgeXz8DE5rD0deu4zEzyQLXC/DSgAYmJ8NiEDSzZfSbT6lZZW1oSgT+QsgMyDihsjLl+x/W0igQijTG3ThmfjZUYlMpaRKD6IzBoM9TuBRvGWSuTDy7PtBDqlPZlweBQqhXLx8Dp2/hx0/FMq1tlXWlJBNOBTSLytoi8A6wDfkzqxtmb3gCMMWeAEyJSOelSy4yUp5TD5cxvrTPosxg8vK2B5Dn94ErmdGkWzOPFTwPqU69MAV6bu4v/LtpHYqLjZwcq55Wm6aMiEgyEJj1cZ4yxyYitiNTEmj6aAzgM9DHG3LOTUweLVZYRfxPWfAprRltnKD/4AdTsnilTTW/GJ/DG3N3M2hbJA9UK88mjQfjk9LR7vcp52Wr6aByQiLX5XJwtAgMwxoQn9f8HGmMe+bckoFSW4uEFzUfAwHVQqDLMfw6+fxiiD9m9ai8Pdz7uGsjwVhVZvu8sj05Yr0dhqrtKdSIQkWFY3UOFsMYLfhCRIfYKTKlsxa+y1VXU/jNrRtFXDa2WQoLNPk/dlYgwvFUlpvUN4Vj0NTqOW8fpi5m/V5JybmlZR7ATaGCMuZr0ODewwRgTaMf47kq7hlSWduk0LH4Z9i0A/+rw8OdQ4h+tdZvbcCia3lM2U9w3J9P7hVDUJ6fd61TOxRZdQ0KK8wiSvtedrpRKq3xF4fFp1nTT6xfg21aw+BW4edmu1TYoX5BJvepy7tJNekzcROSFa3atT2UdaUkEU7BmDb0jIu8Cm4DJ9glLKRdQ5SFrIVq9/rDpaxhfHyKW2LXKRhULMbl3Xc5cusEj49ex7/Qlu9ansoa07D76KdAHiAbOA72MMZ/ZKzClXIJ3Pmg3Cp7+HbzywozHYWYva1GandQrW4A5AxsSl2DoNnEj+89oMnB19x0jEJHLWDuCJl9K8b0xxuSzR2D/RscIVLYUHwvrx8Ifo6z1Bw++B7WeAjf77A158Nxluk3chDGGGf3rU7FwXrvUo5xHuscIjDF5jTH5UnzlTfGV6UlAqWzLIwc0eQkGroeigfDrMPiuPUT9ZZfqKvjnZWqfusQnWi0D7SZyXc6yDbVS6pZCFawDcB4eB2f3wIRQWP2R1WKwserFfPiuTz3cROg2cSM7dLM6l6SJQClnJAK1e8LgLVC1A6weCV83huMbbV5VUMn8/Ng/BA83N7pN3Miq/edsXodybpoIlHJmefyh62ToPsvayXRya1j4H7hx0abVVPDPy8IhjSiePyfPTNvGkt2nbVq+cm6aCJTKCio9CM9thPqDYNtUGFcP9i6waRVFfLz5ZWAo5f3z8OwP25m7I9Km5SvnpYlAqazCKw+0GQn9VkBuP5jZE37qAZdO2awKn1ye/PB0PQrn8+I/P//JrK0nbFa2cl6aCJTKaorXhgGr4IH34OAKq3WweSIkJtqk+IJ5vPh1cCMAXpq9k7/O2nfFs3I8TQRKZUXunhA6DJ5bDyXqwKIXrfGDc/tsUrx/Pm9+HlAfgHZj13A46opNylXOSROBUllZgXLQcx50+hqiD8KExrDy/yDuRoaLDilXkAlP1iY+0dBi9B+6a2k2polAqaxOBIKegMFbIaALhI2y1h4cXZvhotsEFOWt9tUA6PDFOq7HJtznDpUVaSJQKrvIXRA6fw0951rnHEx9CBYMsXY4zYC+jcryTNNynL9yky5frddkkA1pIlAquynfwppqGjoMdky3BpN3z4E0HEt7pxFtq/LCA5XYe/oSXSdoMshuNBEolR3lyGXNKhqwCvIVg9l94cfHISb900GHtKzIGw9VZc+pS/T4diMJielPLMq5aCJQKjsrGmStO2g9Eo6ugfEhsPErSEzfJ/p+jcvxUuvKbD8ew4uz/rRxsMpRNBEold25e0CDQVZ3UemGsORV61S0M7vSVdyg5hV4pGYx5u44yetz01eGci6aCJRyFb6loccs6DIJLp6Ar5vC8ncgLu3TQkc/VpNqRfMxfdNxTQbZgCYCpVyJCNToCoM2Q81usPYz+LIBHFqVpmLc3YSpfesCMH3TceZs032JsjKnSQQi4i4iO0RkoaNjUSrby1UAOo63zj0QgWmPwNyBcO3vVBfhn9ebda+2AOCFWX+y6XC0nYJV9uY0iQAYBthmfbxSKnXKNrFORGv8AuyaCeOCYefMVE81LZ4/J1/1qA3A499s5FSMrj7OipwiEYhICeAh4FtHx6KUy/HMCS3fgmfCwLcs/NIffugMF46m6va2NYryatsqADT/ZDX3OwddOR+nSATAGOBlwDbbJyql0q5wdXj6d2g7Ck5shvH1Yd3nkBB/31ufbVqeumV8uRmfSK8pWzIhWGVLDk8EItIeOGeM2Xaf5w0Qka0isjUqKiqTolPKxbi5Q8gAGLQJyjeHZW/CxOZwasd9b53R39qtNOyvKEYu0l7erMThiQAIBR4WkaPAT0ALEfnhzicZY74xxgQbY4L9/PwyO0alXItPCXjiR3jse7hyFia2gKWvW8dl3oOHuxubXmsJwDdhh5m67khmRasyyOGJwBgzwhhTwhhTBngCWGmMedLBYSmlRKBaR2uqae1esGGc1V10YPk9bymcz5vlzzcB4J1f97L9eMY2vFOZw+GJQCnl5HLmhw5joM9i8PSG6V1gTj+4cvcu2gr+eZnwZB0AOn+5nis37z/GoBzLqRKBMWa1Maa9o+NQSt1F6Ybw7FpoNgL2zIPxda3dTe8yS6hNQBH6hpYFoNmoVTqTyMk5VSJQSjk5Dy9o9ioMXAeFKsP85+D7hyH60D+e+laHauTP5cn5K7EM/vH+g83KcTQRKKXSzq+y1VXU/jM4FQ5fNYQ1o60DcVK4NXj8267TLNx5ygGBqtTQRKCUSh83Nwjuaw0mV3wQVrxnbWQXuTX5KV4e7vw2tBEAg3/coSuPnZQmAqVUxuQrCo9Ps6abXr9gbXG96GW4eRmA6sV8eOOhqgA0/HAl12J18NjZaCJQStlGlYeshWj1+sPmb6xDcCIWA9aBNqEVCgLwwKdhjoxS3YUmAqWU7Xjng3aj4Oll4O0DM56Amb3g8hl+eDoEb083TsZc15XHTkYTgVLK9krWhQF/QIs3rVbBuHrI9u9Y+3IzwFp5vPlI6re8VvaliUApZR8eOaDJi9Y210UD4ddhFJrVmZldrC6ix77eQNTlmw4OUoEmAqWUvRWqYB2A8/A4OLeXeks6MLXsSnIQR9NRq4hP0E2HHU0TgVLK/kSgdk8YvAWqdqDZ6W/5PefrVIvbw5AZutjM0TQRKKUyTx5/6DoZus+iVF6Y7fUeofs/YPrqnY6OzKVpIlBKZb5KD+I2aBMxQf3p5r6SVqs6cGbjTEdH5bI0ESilHMMrD/k7fcKCutM4b3wosqQ/sdO7wSXdiiKzaSJQSjlUp/YdGFXqK0bGdSPxwHLMuHqweSIk6iByZtFEoJRyuClPN+TbxA48ePMjjnpXgUUvwuTWcE4XnmUGTQRKKYcTEda+0oLjpjDNzw3nZLMxEH0QJjSGlf8HcTccHWK2polAKeUUiuXPydgnagJChzUliHtuMwR0gbBRMCEUjq51dIjZliYCpZTTeDioGI0rFuLvq7EM//UEdP4aes61zjmY+hDMH2ztcKpsShOBUsppiAijHwvC3U34bedpJvxxCMq3gOc2QugwCP8RxtWD3XPuekSmSh9NBEopp+Kf15tJvYIB+HDxfnafvAg5csED78GAVZCvGMzuCz8+DjEnHBxt9qCJQCnldJpV9ueZJuUAeH3e7v/9oGgQ9FsBrUfC0TXWmQcbv4LEBAdFmj1oIlBKOaVX21ahZRV/dp+8SJ8pm//3A3cPaDDI6i4q3RCWvGqdinZml+OCzeI0ESilnJKIMLhFBQKK5WNVRBQHzl7GpBwX8C0NPWZBl0lw8YR1XvKytyFOz0VOK00ESimnVauUL0/UKwXAA5+FseNEzO1PEIEaXWHQZqjZDdaNgS8bwKFVmR5rVubwRCAiJUVklYjsFZE9IjLM0TEppZxH59rFGdmpBgDL9p7l3KW7LC7LVQA6jrfOPRCBaY/A3IFwTU9BSw2HJwIgHnjBGFMNqA8MEpFqDo5JKeUkvDzcaRNQBHc34avVh3gj5eDxnco2sU5Ea/wC7JoJ44Jh50ydanofDk8ExpjTxpjtSd9fBvYBxR0blVLKmRTInYM/XmpGUMn8nIy5zom/r90+XpCSZ05o+RY8Ewa+ZeGX/vBDZ7hwNFNjzkocnghSEpEyQC1g011+NkBEtorI1qioqEyPTSnlWCV8c1HSNyd7Tl2i8cerWLz7zL/fULg6PP07tB0FJzbD+Pqw7nNIiM+cgLMQp0kEIpIHmAMMN8ZcuvPnxphvjDHBxphgPz+/zA9QKeVwrz9UlY+7BAJwKiYVs4Pc3CFkAAzaBOWbw7I3YWJzOKXHY6bkFIlARDyxksB0Y8wvjo5HKeWcivrkpFNtq+d45tYTDJ2xg8gL1+5/o08JeOJHeOx7uHIWJraApa9D7FU7R5w1ODwRiIgAk4B9xphPHR2PUsq5ebq78VBgUeITDAv+PMWaA+dTd6MIVOtoTTWt3Qs2jLO6iw4st2/AWYDDEwEQCvQEWohIeNJXO0cHpZRyXuO71+aX5xoCcCMujdtL5MwPHcZAn8Xg6Q3Tu8CcfnDFdccePRwdgDFmLSCOjkMplbV4e7oDMHX9UZbtPUtohUIMal4h9QWUbgjProW1n0HYJ3BwOTz4AdTsbrUeXIgztAiUUirNvDzceDy4JP55vfjr7GV+3HQ87YV4eEGzV2HgOvCrAvOfg+8fhuhDtg/YiWkiUEplSSLCR10DmfVsQx6sXoSb8RnYgdSvMvReBO0/g1Ph8FVDWDPaOhDHBWgiUEpleV4eblyPTeBw1BUOR13hys10rBVwc4PgvtZgcqXWsOI9ayO7yK22D9jJaCJQSmV5+bw9uRqbQIvRf9Bi9B90+XJ9Bgorak0zfWKGdSzmt61g0ctw87LtAnYyDh8sVkqpjOobWpby/nkwxjBrayR7T/9jTWraVWkHZRrByvdh8zewfyE8NBoqt8142U5GWwRKqSzPJ5cnDwcVo2PN4lQsnIe4+ETbFOydD9qNgqeXgbcPzHgCZvaCy/fZ3iKL0USglMpWcni4cTPBRonglpJ1YcAf0OJNiFgM4+rB1imQaON6HES7hpRS2UoOdzdi4xP5ZGlE8rWCeXLQu2EZJCPrAzxyQJMXodojsHC49bVzJnQYC36VMhq2Q2kiUEplK5WL5CWHuxtf/WGtBUg0BmOgVdXClCyQK+MVFKpgHYCz4wf4/Q2YEAqNX4RGw611CVmQJgKlVLbSPrAY7QOLJT+eH36SYT+FE2fL7iIRqN3Tmma6ZASsHgm758DDn0Op+rarJ5PoGIFSKltzd7O6gxIS7XBKWR5/6DoJus+CuGswuTUs/A/cuGj7uuxIE4FSKlvzSEoEcQl2PK6y0oPw3EaoPwi2TbUGk/cusF99NqaJQCmVrXm4WW9zdmkRpOSVB9qMhH4rII8fzOwJP/WAiyftW68NaCJQSmVr7u5WiyA+s6Z6Fq8N/VfBA+/BwRUwPgQ2T3TqqaY6WKyUyta83K3Pu52/Wn/X/e5z5fBg3qCGVPDPa7tK3T0hdBhU7WCNGSx68X9TTQtXs109NqKJQCmVrdUu7csrbapwLfafG9GduXiDWdsiOfH3ddsmglsKlIOe82Dnz9bsoq+bWNNMG79oHYrjJDQRKKWyNW9PdwY2K3/Xn+2MjGHWtkgSjR3HD0Qg6Amo8AAsfQ3CRsGeuVbroEwj+9WbBjpGoJRyWW5ix6mld8pdEDp/DT3nWuccTH0I5g+Ga3/bv+770ESglHJZtxKBXVsEdyrfwppqGjoMwn+E8fWsxWiZGcMdNBEopVzWrcVmmdEguE2OXNasogGrIF9xmN0XfnwcYtJx3KYNaCJQSrmspDyQOV1Dd1M0yFp30HokHF0D4+vDhi8hMQPHbqaDJgKllMtyc3NA19Cd3D2gwSCru6h0Q1g6Ar5tCWd2ZVoImgiUUi7LIWME9+JbGnrMgi6T4GKkdV7ysrch9prdq3aKRCAibUQkQkQOisirjo5HKeUa3G8lAmdZ9CsCNbrCoM1QsxusGwNfNYBDq+xarcMTgYi4A+OBtkA1oJuION/SO6VUtpO0DREJztAiSClXAeg43jr3QNxg2iMw91m4Gm2X6pxhQVk94KAx5jCAiPwEdAT2OjQqpVS2d6tr6JOlEUwMO2yTMqf2rUfx/DltUhZlm8DA9dYitHVj4cDv8Oh3ULaxbcpP4gyJoDhwIsXjSCDkzieJyABgAECpUqUyJzKlVLZWJJ83vRuW4dzlGzYr09MtA8dh3rXAnNDyLQjoAktfh4J3XyWdEc6QCFLFGPMN8A1AcHCwk7XjlFJZkZub8M7D1R0dRuoUrg5PzbNL0Q4fIwBOAiVTPC6RdE0ppVQmcIZEsAWoKCJlRSQH8ASQdY72UUqpLM7hXUPGmHgRGQwsBdyBycaYPQ4OSymlXIbDEwGAMWYRsMjRcSillCtyhq4hpZRSDqSJQCmlXJwmAqWUcnGaCJRSysWJcbY9NlJBRKKAY+m8vRBw3obhOJK+FueTXV4H6GtxVhl5LaWNMX53XsySiSAjRGSrMSbY0XHYgr4W55NdXgfoa3FW9ngt2jWklFIuThOBUkq5OFdMBN84OgAb0tfifLLL6wB9Lc7K5q/F5cYIlFJK3c4VWwRKKaVS0ESglFIuzqUSgYi0EZEIETkoIq86Op70EJGSIrJKRPaKyB4RGebomDJKRNxFZIeILHR0LBkhIvlFZLaI7BeRfSLSwNExpZeI/Cfp72u3iMwQEW9Hx5RaIjJZRM6JyO4U1wqIyDIROZD0X19Hxpga93gdo5L+vnaKyFwRyW+LulwmEYiIOzAeaAtUA7qJSDXHRpUu8cALxphqQH1gUBZ9HSkNA/Y5OggbGAssMcZUAYLIoq9JRIoDQ4FgY0wA1vbwTzg2qjSZCrS549qrwApjTEVgRdJjZzeVf76OZUCAMSYQ+AsYYYuKXCYRAPWAg8aYw8aYWOAnoKODY0ozY8xpY8z2pO8vY73ZFHdsVOknIiWAh4BvHR1LRoiID9AEmARgjIk1xsQ4NKiM8QByiogHkAs45eB4Us0YEwb8fcfljsB3Sd9/BzySmTGlx91ehzHmd2NMfNLDjVgnOmaYKyWC4sCJFI8jycJvoAAiUgaoBWxycCgZMQZ4GUh0cBwZVRaIAqYkdXN9KyK5HR1UehhjTgKfAMeB08BFY8zvjo0qwwobY04nfX8GKOzIYGykL7DYFgW5UiLIVkQkDzAHGG6MueToeNJDRNoD54wx2xwdiw14ALWBr4wxtYCrZI3uh39I6j/viJXcigG5ReRJx0ZlO8aaM5+l582LyOtY3cTTbVGeKyWCk0DJFI9LJF3LckTEEysJTDfG/OLoeDIgFHhYRI5iddW1EJEfHBtSukUCkcaYW62z2ViJIStqBRwxxkQZY+KAX4CGDo4po86KSFGApP+ec3A86SYivYH2QA9jo4VgrpQItgAVRaSsiOTAGvxa4OCY0kxEBKsfep8x5lNHx5MRxpgRxpgSxpgyWP8/VhpjsuQnT2PMGeCEiFROutQS2OvAkDLiOFBfRHIl/b21JIsOfKewAOiV9H0vYL4DY0k3EWmD1ZX6sDHmmq3KdZlEkDTAMhhYivVHPdMYs8exUaVLKNAT69NzeNJXO0cHpQAYAkwXkZ1ATWCkY8NJn6RWzWxgO7AL630iy2zRICIzgA1AZRGJFJGngQ+BB0TkAFaL50NHxpga93gd44C8wLKkf/sTbFKXbjGhlFKuzWVaBEoppe5OE4FSSrk4TQRKKeXiNBEopZSL00SglFIuThOBUkq5OE0ESinl4jQRKJcmIlcyeP8zInJGRP4UkUMi8lQGy2spItMyUoZSaaWJQKmMqQG8Y4wJAroBGd32IwjYkeGolEoDTQRKASLyfNJpXLtFZHiK628mnWq3NumkrhfvuDUQiEj6/ggQm+LeriKyMam1sFZE/JKu/yIi/yciYSJyXERapSgvCNghIl4iMlVERibt96OU3WgiUC5PROoAfYAQrFPf+otILRGpC3TBenNuCwTf5fYaQETSm/Vg4PUUP1tljKmf1FpYBjyW4p4YY0wTrNPZeqS4JxBrZ8ylwHJjzGu22mFSqXvxcHQASjmBRsBcY8xVsD6xA42xPijNN8bcAG6IyK8pbxKRklgbgC3COuRoJ/BOiqf0FpHHAS+gCPCaiOQCfIDPkp7jCcQklecJlANmAM8YYzbY/JUqdRfaIlAq/WoAYcaYmkAloArQACBp0Lge0CKpRRAB7ME6L3ubMSYhqYxA4Nbh5FWxtkuPB279XCm700SgFKwBHknafz830Cnp2jqgg4h4J50I1/6O+wJJGtg1xlwAfsQ6fxmsJLHeGHNFRLpgHeyyK+l6+B1l7Ez6PghYj3U2wxQRyQ7HKaosQBOBcnnGmO3AVGAz1vnP3xpjdhhjtmAdaLIT62zYXcDFFLfW4PYZPr8Ct86GmAo8JyKbsc6VPpzU9XRnIgjgfy2CIGC3MeYv4BVgZlJ3kVJ2pecRKPUvRCRP0qf6XEAYMCApcSiVbehgsVL/7hsRqQZ4A99pElDZkbYIlFLKxekYgVJKuThNBEop5eI0ESillIvTRKCUUi5OE4FSSrk4TQRKKeXiNBEopZSL+39M1ZhdZjOK+QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def draw_zipf_plot(dictionary, plot_label):\n",
    "    dictionary.sort(key=lambda x: x[1], reverse=True)\n",
    "    incid = list(map(lambda x: math.log(x[1]), dictionary))\n",
    "    interval = range(1, len(dictionary)+1)\n",
    "    interval = list(map(math.log, interval))\n",
    "#     print(incid)\n",
    "    plt.title('Zipf Graph')\n",
    "    plt.xlabel(r'$\\log Rank$')\n",
    "    plt.ylabel(r'$\\log cf$')\n",
    "    plt.plot(interval, incid, label=plot_label)\n",
    "    plt.plot(range(13), range(13)[::-1], label='Ideal Graph')\n",
    "    plt.legend()\n",
    "draw_zipf_plot(dictionary_with_stopwords, plot_label='Dictionary with Stopwords')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "cf637bd3",
   "metadata": {},
   "outputs": [
    {
     "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": [
    "draw_zipf_plot(dictionary_without_stopwords, plot_label='Dictionary without Stopwords')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "513dd7a2",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    همانطور که مشاهده می شود، با افزایش جایگاه کلمات در دیکشنری تقریبا با شیب -۱ تعداد تکرار کلمه در کل داک ها کاهش پیدا می کند. چرا که در این نمودار داده ها بعد از اعمال لگاریتم محاسبه شده اند. پس قانون zipf در اینجا برقرار می باشد.\n",
    "    <br>\n",
    "    با توجه به نمودار می توان بیان کرد که با حذف stopwords خاصیت zipf به صورت بارزتری رخ داده و از آن پیروی می کند. به خصوص به ازای کلمات پرتکرار.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31ff76c9",
   "metadata": {},
   "source": [
    "# Heaps Law\n",
    "در این قسمت می خواهیم به ارتباط میان تعداد توکن های استخراج شده و سایز دیکشنری بپردازیم.\n",
    "$$\\log M = \\log K + b\\log T$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "5833aa93",
   "metadata": {},
   "outputs": [],
   "source": [
    "def token_heaps(docs):\n",
    "    seen_docs = 0\n",
    "    seen_token_num = 0\n",
    "    dictionary = set()\n",
    "    dictionary_size = []\n",
    "    dictionary_size_estimation = []\n",
    "    seen_token_num_list = []\n",
    "    for body in docs.values():\n",
    "        for token in body['content']:\n",
    "            seen_token_num += 1\n",
    "            dictionary.add(token)\n",
    "            dictionary_size.append(len(dictionary))\n",
    "        seen_docs += 1\n",
    "        if seen_docs in [500, 1000, 1500, 2000]:\n",
    "            dictionary_size_estimation.append(len(dictionary))\n",
    "            seen_token_num_list.append(seen_token_num)\n",
    "    \n",
    "    return dictionary_size, dictionary_size_estimation, seen_token_num_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "424a06dd",
   "metadata": {},
   "outputs": [
    {
     "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": [
    "def draw_heaps_plot(docs, title):\n",
    "    dictionary_size, dictionary_size_estimation, seen_token_num_list = token_heaps(docs)\n",
    "    model = LinearRegression()\n",
    "    X = np.array(list(map(math.log, seen_token_num_list))).reshape(-1, 1)\n",
    "    y = np.array(list(map(math.log, dictionary_size_estimation)))\n",
    "    model.fit(X, y)\n",
    "    plt.title(title)\n",
    "    plt.xlabel(r'$\\log T$')\n",
    "    plt.ylabel(r'$\\log M$')\n",
    "    plt.plot(list(map(math.log, range(1, len(dictionary_size)+1))), list(map(math.log, dictionary_size)), label='actual dictionary size')\n",
    "    estimation_list = list(map(math.log, [1] + seen_token_num_list + [len(dictionary_size)]))\n",
    "    plt.plot(estimation_list, list(map(lambda x: model.coef_ * x + model.intercept_, estimation_list)), linestyle='--', label='expected dictionary size')\n",
    "    plt.plot(list(map(math.log, seen_token_num_list)), list(map(math.log, dictionary_size_estimation)), 'ro', label='Sampled points')\n",
    "    plt.legend()\n",
    "    \n",
    "draw_heaps_plot(preprocessed_docs, 'Heaps Law after Stemming')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "28cd0914",
   "metadata": {},
   "outputs": [],
   "source": [
    "preprocessed_docs_without_stemming = preprocess_contents(docs, stemming=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "56ea83f2",
   "metadata": {},
   "outputs": [
    {
     "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": [
    "draw_heaps_plot(preprocessed_docs_without_stemming, 'Heaps Law before Stemming')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e233b31",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    در این قسمت نیز مشاهده می شود که با توجه به نقاطی که در آن نمونه برداری صورت گرفته است، سایز تخمینی دیکشنری نزدیک به سایز واقعی آن می باشد. بنابراین در اینجا نیز می توان بیان کرد که قانون heaps در هر دو حالت بعد از ریشه یابی و قبل از آن برقرار می باشد.\n",
    "    \n",
    "    در این حالت به دلیل اینکه برخی از کلمات که پیش از این پس از استخراج یک مدخل جدید در مجموعه دیکشنری ایجاد میکردند، حال به عنوان یک مدخل که در گذشته اضافه شده است، اضافه می شود و به همین دلیل در این حالت شیب خط کمتر است.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e4368f9a",
   "metadata": {},
   "source": [
    "# Challenges with Stemming"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "670adbb9",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    مشکل در تشخیص جمع کلمات با «ات»\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "34d09b6a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['مکرر']"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess('مکرر')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "2fcb6432",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['مکررا']"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess('مکررات')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2329502e",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    تفاوت قائل شدن میان دو کلمه با ریشه یکسان\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "5e581c34",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['آزمایشی']"
      ]
     },
     "execution_count": 126,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess('آزمایشی')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "id": "c8ec809b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['آزمایش']"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess('آزمایش')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8ad43f69",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\n",
    "    تلاش برای ریشه یابی کردن بر روی کلمات خارجی\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "id": "98485e4f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['آنت', 'هیستامین']"
      ]
     },
     "execution_count": 147,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocess('آنتی هیستامین')"
   ]
  }
 ],
 "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}