Lecture06_Sub_LSA_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Latent Semantic Analysis &#49892;&#49845; 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(&#50696;&#51228;) &#45796;&#51020;&#44284; &#44057;&#51060; document&#50752; query&#44032; &#51452;&#50612;&#51276;&#45796;.\n",
    "\n",
    "$d_1$ : \"Machine learning is super fun\"\n",
    "\n",
    "$d_2$ : \"Python is super, super cool\"\n",
    "\n",
    "$d_3$ : \"Statistics is cool, too\"\n",
    "\n",
    "$d_4$ : \"Data science is fun\"\n",
    "\n",
    "$d_5$ : \"Python is great for machine learning\"\n",
    "\n",
    "$d_6$ : \"I like football\"\n",
    "    \n",
    "$d_7$ : \"Football is great to watch\"\n",
    "\n",
    "$query$ : \"machine learning python\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Matrix&#47484; &#49373;&#49457;&#54616;&#44592; &#50948;&#54616;&#50668;, index&#51032; &#44256;&#51221;&#51060; &#54596;&#50836;&#54616;&#44256;, &#51060;&#47484; &#50948;&#54644;&#49436; &#49692;&#49436;&#47484; &#51221;&#54869;&#55176; &#48372;&#51109;&#54616;&#44592; &#50948;&#54644; OrderDict&#51012; &#49324;&#50857;&#54620;&#45796;.\n",
    "from collections import OrderedDict\n",
    "\n",
    "# Corpus&#47484; &#49373;&#49457;&#54620;&#45796;.\n",
    "corpus = OrderedDict(\n",
    "    {\n",
    "        'cool':0, \n",
    "        'data':1,\n",
    "        'football':2,\n",
    "        'fun':3,\n",
    "        'great':4,\n",
    "        'learning':5,\n",
    "        'like':6,\n",
    "        'machine':7,\n",
    "        'python':8,\n",
    "        'science':9,\n",
    "        'statistics':10,\n",
    "        'super':11,\n",
    "        'watch':12\n",
    "    }\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('cool', 0),\n",
       "             ('data', 1),\n",
       "             ('football', 2),\n",
       "             ('fun', 3),\n",
       "             ('great', 4),\n",
       "             ('learning', 5),\n",
       "             ('like', 6),\n",
       "             ('machine', 7),\n",
       "             ('python', 8),\n",
       "             ('science', 9),\n",
       "             ('statistics', 10),\n",
       "             ('super', 11),\n",
       "             ('watch', 12)])"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "d1 = \"Machine learning is super fun\"\n",
    "d2 = \"Python is super, super cool\"\n",
    "d3 = \"Statistics is cool, too\"\n",
    "d4 = \"Data science is fun\"\n",
    "d5 = \"Python is great for machine learning\"\n",
    "d6 = \"I like football\"\n",
    "d7 = \"Football is great to watch\"\n",
    "\n",
    "docs = [d1, d2, d3, d4, d5, d6, d7]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Machine learning is super fun',\n",
       " 'Python is super, super cool',\n",
       " 'Statistics is cool, too',\n",
       " 'Data science is fun',\n",
       " 'Python is great for machine learning',\n",
       " 'I like football',\n",
       " 'Football is great to watch']"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "docs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Document-Word matrix &#49373;&#49457;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.1. &#51204;&#52376;&#47532; (Python &#51652;&#50689;&#50640;&#49436;&#45716; &#51204;&#47928;&#51201;&#51064; &#54805;&#53468;&#49548;&#48516;&#49437;&#44592; konlpy&#47484; &#49324;&#50857;&#54616;&#47732; &#46108;&#45796;.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Machine', 'learning', 'is', 'super', 'fun']"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# &#44033; &#47928;&#49436;&#47484; &#48712; &#50668;&#48177;&#51012; &#44592;&#51456;&#51004;&#47196; &#45208;&#45576;&#45796;.\n",
    "d1_tokens = d1.split(' ')\n",
    "d1_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['machine', 'learning', 'is', 'super', 'fun']"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# &#45824;&#49548;&#47928;&#51088;&#51032; &#44396;&#48516;&#51012; &#50630;&#50532;&#45796;.\n",
    "d1_tokens = map(str.lower, d1_tokens)  # &#46608;&#45716; [s.lower for s in d1_tokens]&#47484; &#49324;&#50857;\n",
    "d1_tokens = list(d1_tokens)\n",
    "d1_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['machine', 'learning', 'is', 'super', 'fun']"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# '-'&#47484; &#51228;&#50808;&#54620; &#44396;&#46160;&#51216;&#47564; &#51228;&#44144;&#54620;&#45796;. (&#50500;&#51452; &#44036;&#45800;&#54616;&#44172;)\n",
    "d1_tokens = [re.sub(r'[^a-zA-Z0-9-]', '', token) for token in d1_tokens]\n",
    "d1_tokens"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "(Advanced)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'teststringwowgood'"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# &#44396;&#46160;&#51216;&#51012; &#51228;&#44144;&#54616;&#45716; &#48169;&#48277;&#51008; &#45796;&#51020;&#44284; &#44057;&#51060; &#54616;&#47732; &#46108;&#45796;.\n",
    "import string\n",
    "print(string.punctuation)\n",
    "\n",
    "s = 'test-string,wow.good!'\n",
    "translator=str.maketrans('','',string.punctuation)\n",
    "s.translate(translator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'asd-'"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# a-z and A-Z and 0-9 and -&#47484; &#51228;&#50808;&#54620;(^) &#45800;&#50612;&#45716; empty string&#44284; &#44368;&#52404;&#54620;&#45796;.\n",
    "re.sub(r'[^a-zA-Z0-9-]', '','asd-')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# &#54632;&#49688;&#47196; &#51221;&#51032;&#54620;&#45796;.\n",
    "def parse_document(doc):\n",
    "    tokens = doc.split(' ')\n",
    "    tokens = map(str.lower, tokens)\n",
    "    tokens = list(tokens)\n",
    "    tokens = [re.sub(r'[^a-zA-Z0-9-]', '', token) for token in tokens]\n",
    "    \n",
    "    return tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['machine', 'learning', 'is', 'super', 'fun']\n",
      "['python', 'is', 'super', 'super', 'cool']\n",
      "['statistics', 'is', 'cool', 'too']\n",
      "['data', 'science', 'is', 'fun']\n",
      "['python', 'is', 'great', 'for', 'machine', 'learning']\n",
      "['i', 'like', 'football']\n",
      "['football', 'is', 'great', 'to', 'watch']\n"
     ]
    }
   ],
   "source": [
    "# &#49324;&#49892; &#51060; &#44284;&#51221;&#50640;&#49436;, &#44284;&#44144;&#54805;&#51008; &#54788;&#51116;&#54805;&#51004;&#47196;, &#52629;&#50557;&#54805; that's&#45716; that is&#47196; &#48320;&#44221;&#54644;&#51452;&#45716; &#46321;\n",
    "# \"&#54805;&#53468;&#49548; &#48516;&#49437;\" &#44284;&#51221;&#51060; &#54596;&#50836;&#54616;&#45208; &#44536;&#44163;&#51008; NLP &#51204;&#52376;&#47532; &#44284;&#51221;&#50640; &#49549;&#54632;&#51004;&#47196; &#50864;&#49440;&#51008; &#50668;&#44592;&#44620;&#51648;&#47564; &#51089;&#50629;&#54620;&#45796;.\n",
    "d2_tokens = parse_document(d2)\n",
    "d3_tokens = parse_document(d3)\n",
    "d4_tokens = parse_document(d4)\n",
    "d5_tokens = parse_document(d5)\n",
    "d6_tokens = parse_document(d6)\n",
    "d7_tokens = parse_document(d7)\n",
    "\n",
    "print(d1_tokens)\n",
    "print(d2_tokens)\n",
    "print(d3_tokens)\n",
    "print(d4_tokens)\n",
    "print(d5_tokens)\n",
    "print(d6_tokens)\n",
    "print(d7_tokens)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.2. Document vector&#47484; &#49373;&#49457;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cool 0\n",
      "data 1\n",
      "football 2\n",
      "fun 3\n",
      "great 4\n",
      "learning 5\n",
      "like 6\n",
      "machine 7\n",
      "python 8\n",
      "science 9\n",
      "statistics 10\n",
      "super 11\n",
      "watch 12\n"
     ]
    }
   ],
   "source": [
    "# &#47676;&#51200; corpus&#45236;&#51032; iterator &#45236;&#50857;&#51012; &#54869;&#51064;&#54616;&#51088;.\n",
    "for v, i in corpus.items():\n",
    "    print(v, i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color=red>Document&#47484; &#44396;&#49457;&#54616;&#45716; vector&#51032; &#44600;&#51060;&#45716; corpus&#51032; &#44600;&#51060;&#50752; &#44057;&#45796;.</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "odict_values([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12])"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus.values()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- &#52395;&#48264;&#51704; &#47928;&#51109; `Machine learning is super fun`&#51008;,\n",
    "\n",
    "- &#51204;&#52376;&#47532; &#54980; `'machine', 'learning', 'is', 'super', 'fun'` tokens&#44032; &#46104;&#47728;,\n",
    "\n",
    "- &#51204;&#52404; corpus&#51012; entry&#47196; &#44032;&#51652; vector `(cool:&#50630;&#51020;) (data:&#50630;&#51020;) (football:&#50630;&#51020;) fun (great:&#50630;&#51020;) learning (like:&#50630;&#51020;) machine (python:&#50630;&#51020;) (science:&#50630;&#51020;) (statistics:&#50630;&#51020;) super (watch:&#50630;&#51020;)`&#50640; &#45824;&#51025;&#46104;&#44256;,\n",
    "\n",
    "- &#52572;&#51333;&#51201;&#51004;&#47196; document vector `0 0 0 1 0 1 0 1 0 0 0 1 0`&#50640; &#45824;&#51025;&#46108;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "odict_items([('cool', 0), ('data', 1), ('football', 2), ('fun', 3), ('great', 4), ('learning', 5), ('like', 6), ('machine', 7), ('python', 8), ('science', 9), ('statistics', 10), ('super', 11), ('watch', 12)])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus.items()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['machine', 'learning', 'is', 'super', 'fun']"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d1_tokens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# d1 vector&#47484; &#49373;&#49457;&#54644;&#48372;&#51088;.\n",
    "d1_vec = []\n",
    "for v, _ in corpus.items():\n",
    "    if v in d1_tokens:\n",
    "        d1_vec.append(1)\n",
    "    else:\n",
    "        d1_vec.append(0)\n",
    "d1_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Document vector&#47484; &#49373;&#49457;&#54616;&#45716; &#53076;&#46300;&#47484; &#54632;&#49688;&#47196; &#48148;&#45012;&#48372;&#51088;.\n",
    "def get_doc_vector(tokens):\n",
    "    doc_vec = [1 if v in tokens else 0 for v, _ in corpus.items()]\n",
    "    return doc_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0]\n",
      "[1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0]\n",
      "[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0]\n",
      "[0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0]\n",
      "[0, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0]\n",
      "[0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]\n",
      "[0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1]\n"
     ]
    }
   ],
   "source": [
    "d2_vec = get_doc_vector(d2_tokens)\n",
    "d3_vec = get_doc_vector(d3_tokens)\n",
    "d4_vec = get_doc_vector(d4_tokens)\n",
    "d5_vec = get_doc_vector(d5_tokens)\n",
    "d6_vec = get_doc_vector(d6_tokens)\n",
    "d7_vec = get_doc_vector(d7_tokens)\n",
    "\n",
    "print(d1_vec)\n",
    "print(d2_vec)\n",
    "print(d3_vec)\n",
    "print(d4_vec)\n",
    "print(d5_vec)\n",
    "print(d6_vec)\n",
    "print(d7_vec)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.3. Document-Word matrix&#47484; &#49373;&#49457;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0],\n",
       "       [1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0],\n",
       "       [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
       "       [0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0],\n",
       "       [0, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0],\n",
       "       [0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],\n",
       "       [0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1]])"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# &#52572;&#51333; matrix Word-Document matrix\n",
    "A = np.array([d1_vec, d2_vec, d3_vec, d4_vec, d5_vec, d6_vec, d7_vec])\n",
    "A"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Doc-Doc matrix &#50752; Word-Word matrix&#47484; &#49373;&#49457;&#54620;&#45796;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1. Doc-Doc matrix $AA^T$ :  &#47928;&#49436; $i$&#50752; &#47928;&#49436; $j$&#44032; $b$&#44060;&#51032; &#44277;&#53685; &#45800;&#50612;&#47484; &#44032;&#51648;&#44256; &#51080;&#51004;&#47732; $B[i, j]=b$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[4, 1, 0, 1, 2, 0, 0],\n",
       "       [1, 3, 1, 0, 1, 0, 0],\n",
       "       [0, 1, 2, 0, 0, 0, 0],\n",
       "       [1, 0, 0, 3, 0, 0, 0],\n",
       "       [2, 1, 0, 0, 4, 0, 1],\n",
       "       [0, 0, 0, 0, 0, 2, 1],\n",
       "       [0, 0, 0, 0, 1, 1, 3]])"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "doc_mat = np.dot(A, A.T)\n",
    "doc_mat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2. Word-Word matrix $A^TA$ : &#45800;&#50612; $i$&#50752; &#45800;&#50612; $j$&#44032; &#47928;&#49436; $c$&#44060; &#50640;&#49436; &#54632;&#44760; &#48156;&#49373;&#54664;&#51004;&#47732; $C[i,j]=c$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[2, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0],\n",
       "       [0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0],\n",
       "       [0, 0, 2, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1],\n",
       "       [0, 1, 0, 2, 0, 1, 0, 1, 0, 1, 0, 1, 0],\n",
       "       [0, 0, 1, 0, 2, 1, 0, 1, 1, 0, 0, 0, 1],\n",
       "       [0, 0, 0, 1, 1, 2, 0, 2, 1, 0, 0, 1, 0],\n",
       "       [0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],\n",
       "       [0, 0, 0, 1, 1, 2, 0, 2, 1, 0, 0, 1, 0],\n",
       "       [1, 0, 0, 0, 1, 1, 0, 1, 2, 0, 0, 1, 0],\n",
       "       [0, 1, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0],\n",
       "       [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0],\n",
       "       [1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 2, 0],\n",
       "       [0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1]])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "word_mat = np.dot(A.T, A)\n",
    "word_mat"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. SVD(Singular Value Decomposition)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### $$A=U \\Sigma V^T$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "U, sig, V_T = np.linalg.svd(A)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1. $U$&#51032; &#52972;&#47100;&#51008; $AA^T$&#51032; &#44256;&#50976;&#48289;&#53552;&#46308;&#51060;&#45796;.  &#51593; Doc-Doc matrix &#45936;&#51060;&#53552;&#47484; &#49373;&#49457;&#54616;&#45716; &#52629;&#46308;&#51060;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.63555233,  0.3116475 , -0.23588417,  0.0923833 , -0.51003069,\n",
       "         0.29231986, -0.29864837],\n",
       "       [-0.35219853,  0.13826479,  0.61288984, -0.36051167, -0.04467366,\n",
       "        -0.56373943, -0.17730917],\n",
       "       [-0.07303416,  0.07324226,  0.43497297, -0.43325945,  0.1676411 ,\n",
       "         0.73984157,  0.19213866],\n",
       "       [-0.16627132,  0.35104399, -0.57669202, -0.55020151,  0.4027139 ,\n",
       "        -0.16590483,  0.155318  ],\n",
       "       [-0.63765049, -0.26214197,  0.05160136,  0.40140898,  0.39896807,\n",
       "        -0.03886768,  0.44744312],\n",
       "       [-0.03657726, -0.38783282, -0.12179755, -0.3522012 , -0.60221382,\n",
       "        -0.113457  ,  0.57778327],\n",
       "       [-0.17638947, -0.73214047, -0.17161637, -0.29306375,  0.16048032,\n",
       "         0.08645119, -0.53318928]])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "U"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> $AA^T$ &#51032; eigenvector (&#52972;&#47100;&#46308;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# np.linalg.eig(np.dot(A, A.T))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2. $V$&#51032; &#52972;&#47100;&#51008; $AA^T$&#51032; &#44256;&#50976;&#48289;&#53552;&#46308;&#51060;&#45796;. &#51593; Word-Word matrix &#45936;&#51060;&#53552;&#47484; &#49373;&#49457;&#54616;&#45716; &#52629;&#46308;&#51060;&#45796;."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.16280159,  0.10726903,  0.56753028, -0.47167366,  0.09339555,\n",
       "         0.15827037,  0.01428834, -0.11369133, -0.20704032,  0.03298642,\n",
       "        -0.46399237, -0.33237781,  0.02832841],\n",
       "       [-0.06365747,  0.17803733, -0.31234068, -0.32694004,  0.30586704,\n",
       "        -0.14910562,  0.14965023,  0.36329084, -0.49577965, -0.35164399,\n",
       "         0.26687418, -0.19166032, -0.11061159],\n",
       "       [-0.08153494, -0.56801159, -0.15891516, -0.38342851, -0.335503  ,\n",
       "        -0.02427125,  0.04296669,  0.21060964,  0.23009486,  0.06441144,\n",
       "        -0.04761734, -0.16005546, -0.50409511],\n",
       "       [-0.30698056,  0.33609413, -0.4400973 , -0.27204415, -0.08150866,\n",
       "         0.11361448, -0.13810005, -0.30264492,  0.40962447, -0.41313829,\n",
       "        -0.22319577,  0.043474  ,  0.06863482],\n",
       "       [-0.31165761, -0.50426555, -0.06500102,  0.06438076,  0.42490916,\n",
       "         0.04276529, -0.08261717, -0.31394715, -0.29811439, -0.13523936,\n",
       "        -0.1814318 ,  0.44723235, -0.11115103],\n",
       "       [-0.48744948,  0.02510749, -0.09980894,  0.29342062, -0.08435367,\n",
       "         0.22778809,  0.14336504, -0.37964684, -0.1478498 ,  0.19818597,\n",
       "         0.33701334, -0.491527  , -0.16032123],\n",
       "       [-0.01400371, -0.19669535, -0.06596646, -0.20928454, -0.45739012,\n",
       "        -0.10196856,  0.55669919, -0.21060964, -0.23009486, -0.06441144,\n",
       "         0.04761734,  0.16005546,  0.50409511],\n",
       "       [-0.48744948,  0.02510749, -0.09980894,  0.29342062, -0.08435367,\n",
       "         0.22778809,  0.14336504,  0.63109722, -0.0114254 ,  0.09249607,\n",
       "        -0.36669574,  0.07998492,  0.19574354],\n",
       "       [-0.37896663, -0.06282621,  0.35989337,  0.02430195,  0.26909174,\n",
       "        -0.54158824,  0.2602764 ,  0.06249678,  0.45738959, -0.15544268,\n",
       "         0.2111142 , -0.03569028,  0.07572872],\n",
       "       [-0.06365747,  0.17803733, -0.31234068, -0.32694004,  0.30586704,\n",
       "        -0.14910562,  0.14965023, -0.06064591,  0.08615518,  0.76478228,\n",
       "        -0.04367841,  0.14818633,  0.04197677],\n",
       "       [-0.02796134,  0.03714593,  0.23558459, -0.25745088,  0.12732584,\n",
       "         0.66492659,  0.18512726,  0.11369133,  0.20704032, -0.03298642,\n",
       "         0.46399237,  0.33237781, -0.02832841],\n",
       "       [-0.37816335,  0.2281799 ,  0.20418907, -0.15932689, -0.421306  ,\n",
       "        -0.24393613, -0.45858921,  0.05119455, -0.25034927,  0.12245626,\n",
       "         0.25287817,  0.36806808, -0.10405713],\n",
       "       [-0.06753123, -0.37131624, -0.0929487 , -0.17414397,  0.12188713,\n",
       "         0.0776973 , -0.51373249,  0.10333751,  0.06801952,  0.07082792,\n",
       "         0.22904914, -0.28717689,  0.61524614]])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# V&#51032; &#44033; &#52972;&#47100;&#51008; A^T*A&#51032; eigenvector &#51060;&#45796;.\n",
    "V_T.T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> $A^TA$&#51032; eigenvector (&#52972;&#47100;&#46308;)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# np.linalg.eig(np.dot(A.T, A))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Word vector&#50752; Document vector &#49373;&#49457;&#54616;&#44592;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Word vector&#50752; document vector&#47484; &#49373;&#49457;&#54624; &#46412;, &#49440;&#53469;&#54616;&#45716; singular value&#51032; &#49688;&#44032; &#51473;&#50836;&#54616;&#45796;. 2&#52264;&#50896;&#51004;&#47196; &#48372;&#50668;&#51452;&#44592; &#50948;&#54644; $k=2$&#47196; &#54364;&#54788;&#54616;&#51088;."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1. Words Vectors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_eigen_vec = V_T.T[:, :2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.16280159,  0.10726903],\n",
       "       [-0.06365747,  0.17803733],\n",
       "       [-0.08153494, -0.56801159],\n",
       "       [-0.30698056,  0.33609413],\n",
       "       [-0.31165761, -0.50426555],\n",
       "       [-0.48744948,  0.02510749],\n",
       "       [-0.01400371, -0.19669535],\n",
       "       [-0.48744948,  0.02510749],\n",
       "       [-0.37896663, -0.06282621],\n",
       "       [-0.06365747,  0.17803733],\n",
       "       [-0.02796134,  0.03714593],\n",
       "       [-0.37816335,  0.2281799 ],\n",
       "       [-0.06753123, -0.37131624]])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Word-Word matrix&#51032; eigenvectors\n",
    "word_eigen_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2.6119688 , 1.97174373, 1.84635579, 1.68288201, 1.31663057,\n",
       "       1.11266654, 1.03787339])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Singular value\n",
    "sig"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$\\text{word vector} = \\Sigma V^T$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[2.6119688 , 0.        ],\n",
       "       [0.        , 1.97174373]])"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.diag(sig[:2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.16280159, -0.06365747, -0.08153494, -0.30698056, -0.31165761,\n",
       "        -0.48744948, -0.01400371, -0.48744948, -0.37896663, -0.06365747,\n",
       "        -0.02796134, -0.37816335, -0.06753123],\n",
       "       [ 0.10726903,  0.17803733, -0.56801159,  0.33609413, -0.50426555,\n",
       "         0.02510749, -0.19669535,  0.02510749, -0.06282621,  0.17803733,\n",
       "         0.03714593,  0.2281799 , -0.37131624]])"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "word_eigen_vec.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "word_vec = np.dot(np.diag(sig[:2]), word_eigen_vec.T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.42523269,  0.21150705],\n",
       "       [-0.16627132,  0.35104399],\n",
       "       [-0.21296672, -1.11997329],\n",
       "       [-0.80182365,  0.66269149],\n",
       "       [-0.81403996, -0.99428243],\n",
       "       [-1.27320282,  0.04950553],\n",
       "       [-0.03657726, -0.38783282],\n",
       "       [-1.27320282,  0.04950553],\n",
       "       [-0.98984902, -0.12387718],\n",
       "       [-0.16627132,  0.35104399],\n",
       "       [-0.07303416,  0.07324226],\n",
       "       [-0.98775086,  0.44991228],\n",
       "       [-0.17638947, -0.73214047]])"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sigular value&#44032; weight&#47196; &#49324;&#50857;&#46108; word vector\n",
    "word_vec.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.42523269, -0.16627132, -0.21296672, -0.80182365, -0.81403996,\n",
       "        -1.27320282, -0.03657726, -1.27320282, -0.98984902, -0.16627132,\n",
       "        -0.07303416, -0.98775086, -0.17638947],\n",
       "       [ 0.21150705,  0.35104399, -1.11997329,  0.66269149, -0.99428243,\n",
       "         0.04950553, -0.38783282,  0.04950553, -0.12387718,  0.35104399,\n",
       "         0.07324226,  0.44991228, -0.73214047]])"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "word_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OrderedDict([('cool', 0),\n",
       "             ('data', 1),\n",
       "             ('football', 2),\n",
       "             ('fun', 3),\n",
       "             ('great', 4),\n",
       "             ('learning', 5),\n",
       "             ('like', 6),\n",
       "             ('machine', 7),\n",
       "             ('python', 8),\n",
       "             ('science', 9),\n",
       "             ('statistics', 10),\n",
       "             ('super', 11),\n",
       "             ('watch', 12)])"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corpus"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "odict_keys(['cool', 'data', 'football', 'fun', 'great', 'learning', 'like', 'machine', 'python', 'science', 'statistics', 'super', 'watch'])"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Plotting &#51012; &#50948;&#54644; &#45796;&#51020;&#44284; &#44057;&#51060; x&#52629; y&#52629;&#51012; &#44396;&#48516;&#54620;&#45796;.\n",
    "# x &#52629;\n",
    "w_x = word_vec[0]\n",
    "# y &#52629;\n",
    "w_y = word_vec[1]\n",
    "# Words&#47484; plotting&#50640; &#45208;&#53440;&#45236;&#44592; &#50948;&#54644;&#49436;\n",
    "words = corpus.keys()\n",
    "words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl\n",
    "\n",
    "mpl.rcParams['figure.figsize'] = (8, 8)\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "# Scatter\n",
    "ax.scatter(w_x, w_y, s=100)\n",
    "\n",
    "# Scatter&#50640; &#46972;&#48296;&#47553;&#51012; &#54620;&#45796;.\n",
    "for i, txt in enumerate(words):\n",
    "    ax.annotate(txt, (w_x[i], w_y[i]), fontsize=20)  # free&#50752; live&#44032; &#44217;&#52432;&#51648;&#45716; &#50948;&#52824;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2. Document Vectors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "doc_eigen_vec = U[:, :2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.63555233,  0.3116475 ],\n",
       "       [-0.35219853,  0.13826479],\n",
       "       [-0.07303416,  0.07324226],\n",
       "       [-0.16627132,  0.35104399],\n",
       "       [-0.63765049, -0.26214197],\n",
       "       [-0.03657726, -0.38783282],\n",
       "       [-0.17638947, -0.73214047]])"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Doc-Doc matrix&#51032; eigenvectors\n",
    "doc_eigen_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2.6119688 , 1.97174373, 1.84635579, 1.68288201, 1.31663057,\n",
       "       1.11266654, 1.03787339])"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Singular value\n",
    "sig"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Sigular value&#44032; weight&#47196; &#49324;&#50857;&#46108; doc vector\n",
    "doc_vec = np.dot(doc_eigen_vec, np.diag(sig[:2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-1.66004286,  0.614489  ],\n",
       "       [-0.91993157,  0.27262272],\n",
       "       [-0.19076294,  0.14441497],\n",
       "       [-0.43429549,  0.69216879],\n",
       "       [-1.6655232 , -0.51687678],\n",
       "       [-0.09553865, -0.76470694],\n",
       "       [-0.46072378, -1.44359338]])"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "doc_vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Plotting&#51012; &#50948;&#54644;&#49436; &#45796;&#51020;&#44284; &#44057;&#51060; &#45208;&#45576;&#45796;.\n",
    "doc_x = doc_vec[:, 0]\n",
    "doc_y = doc_vec[:, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "# Scatter\n",
    "ax.scatter(doc_x, doc_y, marker='*', s=100)\n",
    "\n",
    "# Scatter&#50640; &#46972;&#48296;&#47553;&#51012; &#54620;&#45796;.\n",
    "for i, txt in enumerate(['d1', 'd2', 'd3', 'd4', 'd5', 'd6', 'd7']):\n",
    "    ax.annotate(txt, (doc_x[i], doc_y[i]), fontsize=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Document&#50752; word&#47484; &#54632;&#44760; &#45208;&#53440;&#45236;&#48372;&#44192;&#45796;.\n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(w_x, w_y, s=100)\n",
    "for i, txt in enumerate(words):\n",
    "    ax.annotate(txt, (w_x[i], w_y[i]), fontsize=20)\n",
    "    \n",
    "ax.scatter(doc_x, doc_y, marker='*', s=100)\n",
    "for i, txt in enumerate(['d1', 'd2', 'd3', 'd4', 'd5', 'd6', 'd7']):\n",
    "    ax.annotate(txt, (doc_x[i], doc_y[i]), fontsize=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Query, document distance &#44396;&#54616;&#44592;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.1. Query"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "machine [-1.27320282  0.04950553]\n",
      "python [-0.98984902 -0.12387718]\n",
      "learning [-1.27320282  0.04950553]\n"
     ]
    }
   ],
   "source": [
    "# Query : die, dagger\n",
    "q1 = word_vec[:, 7]\n",
    "print('machine', q1)\n",
    "\n",
    "q2 = word_vec[:, 8]\n",
    "print('python', q2)\n",
    "\n",
    "q3 = word_vec[:, 5]\n",
    "print('learning', q3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "machine python learning : [-1.17875156 -0.00828871]\n"
     ]
    }
   ],
   "source": [
    "query = (q1 + q2 + q3) / 3\n",
    "print('machine python learning :', query)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2. Query&#50752; document&#51032; &#44144;&#47532; : Word embedding &#44277;&#44036;&#50640;&#49436;&#45716; cosine similarity&#47484; &#49324;&#50857;\n",
    "$$\\text{Cosine Similarity} = \\frac{d_i q}{|d_i| |q|}$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_consine_similarity(doc, query):\n",
    "    dot_prod = np.dot(doc, query)\n",
    "    len_doc = np.sqrt(sum(doc ** 2))\n",
    "    len_query = np.sqrt(sum(query ** 2))\n",
    "    \n",
    "    return dot_prod / (len_doc * len_query)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = {}\n",
    "for idx, doc in zip(docs, doc_vec):\n",
    "    result[get_consine_similarity(doc, query)] = idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Machine learning is super fun',\n",
       " 'Python is super, super cool',\n",
       " 'Statistics is cool, too',\n",
       " 'Data science is fun',\n",
       " 'Python is great for machine learning',\n",
       " 'I like football',\n",
       " 'Football is great to watch']"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "docs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0.130945501416146: 'I like football',\n",
       " 0.31073285924247535: 'Football is great to watch',\n",
       " 0.5255159978291322: 'Data science is fun',\n",
       " 0.7930342446275301: 'Statistics is cool, too',\n",
       " 0.9353474497542472: 'Machine learning is super fun',\n",
       " 0.9567621655734274: 'Python is super, super cool',\n",
       " 0.9571262278922466: 'Python is great for machine learning'}"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.3. &#52572;&#51333; &#44208;&#44284;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0.9571262278922466, 'Python is great for machine learning'),\n",
       " (0.9567621655734274, 'Python is super, super cool'),\n",
       " (0.9353474497542472, 'Machine learning is super fun'),\n",
       " (0.7930342446275301, 'Statistics is cool, too'),\n",
       " (0.5255159978291322, 'Data science is fun'),\n",
       " (0.31073285924247535, 'Football is great to watch'),\n",
       " (0.130945501416146, 'I like football')]"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import operator\n",
    "\n",
    "# &#52572;&#51333;&#44208;&#44284;&#47484; sorting&#54616;&#46972;!\n",
    "sorted(result.items(), key=operator.itemgetter(0), reverse=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(-1.17875,-0.00828871,'query')"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.scatter(w_x, w_y, s=100)\n",
    "for i, txt in enumerate(words):\n",
    "    ax.annotate(txt, (w_x[i], w_y[i]), fontsize=20)\n",
    "    \n",
    "ax.scatter(doc_x, doc_y, marker='*', s=100)\n",
    "for i, txt in enumerate(['d1', 'd2', 'd3', 'd4', 'd5', 'd6', 'd7']):\n",
    "    ax.annotate(txt, (doc_x[i], doc_y[i]), fontsize=20)\n",
    "    \n",
    "ax.scatter(query[0], query[1], marker='v', s=200, color='red')\n",
    "ax.annotate('query', (query[0], query[1]), fontsize=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}