Different Embeddings + Cosine Similarity + HeatMap illustration.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://medium.com/swlh/playing-with-word-vectors-308ab2faa519\n",
    "# https://github.com/makcedward/nlp/blob/master/sample/nlp-word_embedding.ipynb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "ss1 = 'The president greets the press in Chicago'\n",
    "ss2 = 'Obama speaks to the media in Illinois'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Count Vectorizer method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cosine_distance_countvectorizer_method(s1, s2):\n",
    "    \n",
    "    # sentences to list\n",
    "    allsentences = [s1 , s2]\n",
    "    \n",
    "    # packages\n",
    "    from sklearn.feature_extraction.text import CountVectorizer\n",
    "    from scipy.spatial import distance\n",
    "    \n",
    "    # text to vector\n",
    "    vectorizer = CountVectorizer()\n",
    "    all_sentences_to_vector = vectorizer.fit_transform(allsentences)\n",
    "    text_to_vector_v1 = all_sentences_to_vector.toarray()[0].tolist()\n",
    "    text_to_vector_v2 = all_sentences_to_vector.toarray()[1].tolist()\n",
    "    \n",
    "    # distance of similarity\n",
    "    cosine = distance.cosine(text_to_vector_v1, text_to_vector_v2)\n",
    "    print('Similarity of two sentences are equal to ',round((1-cosine)*100,2),'%')\n",
    "    return cosine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Similarity of two sentences are equal to  37.8 %\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6220355269907728"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cosine_distance_countvectorizer_method(ss1 , ss2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Glove Embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "gloveFile = \"data\\\\glove.6B.50d.txt\"\n",
    "import numpy as np\n",
    "def loadGloveModel(gloveFile):\n",
    "    print (\"Loading Glove Model\")\n",
    "    with open(gloveFile, encoding=\"utf8\" ) as f:\n",
    "        content = f.readlines()\n",
    "    model = {}\n",
    "    for line in content:\n",
    "        splitLine = line.split()\n",
    "        word = splitLine[0]\n",
    "        embedding = np.array([float(val) for val in splitLine[1:]])\n",
    "        model[word] = embedding\n",
    "    print (\"Done.\",len(model),\" words loaded!\")\n",
    "    return model\n",
    "\n",
    "import re\n",
    "from nltk.corpus import stopwords\n",
    "import pandas as pd\n",
    "\n",
    "def preprocess(raw_text):\n",
    "\n",
    "    # keep only words\n",
    "    letters_only_text = re.sub(\"[^a-zA-Z]\", \" \", raw_text)\n",
    "\n",
    "    # convert to lower case and split \n",
    "    words = letters_only_text.lower().split()\n",
    "\n",
    "    # remove stopwords\n",
    "    stopword_set = set(stopwords.words(\"english\"))\n",
    "    cleaned_words = list(set([w for w in words if w not in stopword_set]))\n",
    "\n",
    "    return cleaned_words\n",
    "\n",
    "def cosine_distance_between_two_words(word1, word2):\n",
    "    import scipy\n",
    "    return (1- scipy.spatial.distance.cosine(model[word1], model[word2]))\n",
    "\n",
    "def calculate_heat_matrix_for_two_sentences(s1,s2):\n",
    "    s1 = preprocess(s1)\n",
    "    s2 = preprocess(s2)\n",
    "    result_list = [[cosine_distance_between_two_words(word1, word2) for word2 in s2] for word1 in s1]\n",
    "    result_df = pd.DataFrame(result_list)\n",
    "    result_df.columns = s2\n",
    "    result_df.index = s1\n",
    "    return result_df\n",
    "\n",
    "def cosine_distance_wordembedding_method(s1, s2):\n",
    "    import scipy\n",
    "    vector_1 = np.mean([model[word] for word in preprocess(s1)],axis=0)\n",
    "    vector_2 = np.mean([model[word] for word in preprocess(s2)],axis=0)\n",
    "    cosine = scipy.spatial.distance.cosine(vector_1, vector_2)\n",
    "    print('Word Embedding method with a cosine distance asses that our two sentences are similar to',round((1-cosine)*100,2),'%')\n",
    "\n",
    "def heat_map_matrix_between_two_sentences(s1,s2):\n",
    "    df = calculate_heat_matrix_for_two_sentences(s1,s2)\n",
    "    import seaborn as sns\n",
    "    import matplotlib.pyplot as plt\n",
    "    fig, ax = plt.subplots(figsize=(5,5)) \n",
    "    ax_blue = sns.heatmap(df, cmap=\"YlGnBu\")\n",
    "    # ax_red = sns.heatmap(df)\n",
    "    print(cosine_distance_wordembedding_method(s1, s2))\n",
    "    return ax_blue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading Glove Model\n",
      "Done. 400000  words loaded!\n",
      "Word Embedding method with a cosine distance asses that our two sentences are similar to 81.25 %\n",
      "None\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a66016b5c0>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ss1 = 'The president greets the press in Chicago'\n",
    "ss2 = 'Obama speaks to the media in Illinois'\n",
    "\n",
    "model = loadGloveModel(gloveFile)\n",
    "heat_map_matrix_between_two_sentences(ss1,ss2)"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}