Day16- 計數方法詞向量介紹_作業.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 作業目標: 透過思考與回答以更加了解計數方法的詞向量"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 請問詞庫手法會有什麼樣的優缺點？\n",
    "\n",
    "詞庫手法為創建包含大量字詞的詞庫，將相同意思字詞(同義字)或相似意思字詞(相似字)分類在相同群組。\n",
    "\n",
    "優點:\n",
    "將字庫轉換成向量較容易\n",
    "\n",
    "缺點:\n",
    "建立及維護較花費時間及人力，新的字詞無法定義"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 請問共現矩陣有什麼樣的優缺點？ \n",
    "\n",
    "根據分佈假說，相似的字詞會有類似的上下文，因此我們可以透過計數周圍(window)的字詞來表達特定字詞的向量。\n",
    "\n",
    "優點:\n",
    "較簡單\n",
    "\n",
    "缺點:\n",
    "矩陣維度大，較佔內存/高頻詞效果較差效果較差"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 實作cosine similarity\n",
    "\n",
    "在比較兩個詞向量的相似度時可以使用cosine similarity:\n",
    "$$\n",
    "similarity(x,y) = \\frac{x \\cdot y}{||x||||y||} = \\frac{x_1y_1+...+x_ny_n}{\\sqrt{x_1^2+...+x_n^2}\\sqrt{y_1^2+...+y_n^2}}\n",
    "$$\n",
    "\n",
    "請實作cosine similarity並計算共現矩陣課程範例中you向量([0,1,0,0,0,0,0])與I([0,1,0,1,0,0,0])向量的相似度"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": [
      "Similarity: 0.7071067811865475\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "I = np.array([0,1,0,0,0,0,0])\n",
    "You = np.array([0,1,0,1,0,0,0])\n",
    "\n",
    "def cos_similarity(x, y, eps=1e-8):\n",
    "    ### your code ###\n",
    "    dot=sum(a*b for a, b in zip(I, You))\n",
    "    norm_I=sum(x*x for x in I)**0.5\n",
    "    norm_You=sum(y*y for y in You)**0.5\n",
    "    Similarity=dot/(norm_I*norm_You)\n",
    "    return Similarity\n",
    "\n",
    "print(f\"Similarity: {cos_similarity(I, You)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3.7.6 64-bit (conda)",
   "metadata": {
    "interpreter": {
     "hash": "a8f61be024eba58adef938c9aa1e29e02cb3dece83a5348b1a2dafd16a070453"
    }
   }
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6-final"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}