CoNLL_Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\gensim\\utils.py:1212: UserWarning: detected Windows; aliasing chunkize to chunkize_serial\n",
      "  warnings.warn(\"detected Windows; aliasing chunkize to chunkize_serial\")\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import pickle\n",
    "import networkx as nx\n",
    "import numpy as np\n",
    "from tqdm import tqdm_notebook\n",
    "from sklearn.feature_extraction import DictVectorizer\n",
    "from sklearn.metrics import silhouette_score\n",
    "from sklearn.cluster import KMeans\n",
    "from scipy.cluster.hierarchy import dendrogram, linkage  \n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "from collections import defaultdict, Counter\n",
    "from itertools import combinations\n",
    "%matplotlib inline\n",
    "from sklearn.preprocessing import normalize\n",
    "import os\n",
    "from gensim.models.doc2vec import Doc2Vec, TaggedDocument\n",
    "import matplotlib.pyplot as plt\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import spacy\n",
    "from spacy import displacy\n",
    "\n",
    "\n",
    "PATH_TO_RUS_LANG_MODEL = \"C:/Users/User/Documents/spacy-ru-master/ru2/\"\n",
    "ru_nlp = spacy.load(PATH_TO_RUS_LANG_MODEL) # Загрузим языковую модель\n",
    "#ru_nlp = spacy.load('en_core_web_sm')\n",
    "\n",
    "#current_language = 'en'\n",
    "current_language = 'ru'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "graphs_and_sentences = pickle.load(open('conll_parsed_ru_language', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We have 61889 sentences (graphs)\n"
     ]
    }
   ],
   "source": [
    "print('We have {} sentences (graphs)'.format(len(graphs_and_sentences)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# WL Kernel with dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<dep_tree_embedding.WLExtractor object at 0x000001BB9497EF60>, <dep_tree_embedding.DepTreeEmbedding object at 0x000001BBFFFE9588>)\n",
      "graph2vec shape: (61889, 51)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Type</th>\n",
       "      <th>x_0</th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>x_5</th>\n",
       "      <th>x_6</th>\n",
       "      <th>x_7</th>\n",
       "      <th>x_8</th>\n",
       "      <th>...</th>\n",
       "      <th>x_40</th>\n",
       "      <th>x_41</th>\n",
       "      <th>x_42</th>\n",
       "      <th>x_43</th>\n",
       "      <th>x_44</th>\n",
       "      <th>x_45</th>\n",
       "      <th>x_46</th>\n",
       "      <th>x_47</th>\n",
       "      <th>x_48</th>\n",
       "      <th>x_49</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.522512</td>\n",
       "      <td>0.257050</td>\n",
       "      <td>-0.145618</td>\n",
       "      <td>-0.398585</td>\n",
       "      <td>-0.265484</td>\n",
       "      <td>-0.143339</td>\n",
       "      <td>-0.218729</td>\n",
       "      <td>0.085656</td>\n",
       "      <td>0.369437</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.002601</td>\n",
       "      <td>0.076505</td>\n",
       "      <td>-0.656494</td>\n",
       "      <td>-0.241970</td>\n",
       "      <td>0.073937</td>\n",
       "      <td>0.180409</td>\n",
       "      <td>-0.234990</td>\n",
       "      <td>0.168578</td>\n",
       "      <td>0.168087</td>\n",
       "      <td>-0.028329</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.125956</td>\n",
       "      <td>-0.113120</td>\n",
       "      <td>0.127866</td>\n",
       "      <td>-0.243473</td>\n",
       "      <td>-0.246138</td>\n",
       "      <td>-0.521300</td>\n",
       "      <td>-0.748871</td>\n",
       "      <td>0.055020</td>\n",
       "      <td>0.617417</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.063075</td>\n",
       "      <td>-0.099647</td>\n",
       "      <td>-0.240150</td>\n",
       "      <td>0.063227</td>\n",
       "      <td>-0.303290</td>\n",
       "      <td>-0.113590</td>\n",
       "      <td>-0.244465</td>\n",
       "      <td>0.427193</td>\n",
       "      <td>-0.419203</td>\n",
       "      <td>0.152318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>-0.232250</td>\n",
       "      <td>0.061611</td>\n",
       "      <td>0.147097</td>\n",
       "      <td>-0.102518</td>\n",
       "      <td>-0.194391</td>\n",
       "      <td>0.187656</td>\n",
       "      <td>0.050501</td>\n",
       "      <td>-0.135576</td>\n",
       "      <td>-0.077290</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.073048</td>\n",
       "      <td>-0.367363</td>\n",
       "      <td>0.149369</td>\n",
       "      <td>-0.026017</td>\n",
       "      <td>0.453523</td>\n",
       "      <td>-0.281772</td>\n",
       "      <td>-0.091005</td>\n",
       "      <td>0.264164</td>\n",
       "      <td>-0.156693</td>\n",
       "      <td>0.045124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>-0.973664</td>\n",
       "      <td>-0.103102</td>\n",
       "      <td>-0.763641</td>\n",
       "      <td>-0.235664</td>\n",
       "      <td>-0.247679</td>\n",
       "      <td>-0.155536</td>\n",
       "      <td>-0.031103</td>\n",
       "      <td>0.375137</td>\n",
       "      <td>0.245736</td>\n",
       "      <td>...</td>\n",
       "      <td>0.255985</td>\n",
       "      <td>-0.509347</td>\n",
       "      <td>0.002029</td>\n",
       "      <td>0.039524</td>\n",
       "      <td>0.026820</td>\n",
       "      <td>-0.065614</td>\n",
       "      <td>0.203438</td>\n",
       "      <td>-0.094295</td>\n",
       "      <td>-0.093370</td>\n",
       "      <td>-0.432650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>-0.479133</td>\n",
       "      <td>0.019136</td>\n",
       "      <td>-0.397532</td>\n",
       "      <td>0.206036</td>\n",
       "      <td>-0.124943</td>\n",
       "      <td>0.123089</td>\n",
       "      <td>-0.273900</td>\n",
       "      <td>-0.393271</td>\n",
       "      <td>0.112122</td>\n",
       "      <td>...</td>\n",
       "      <td>0.244408</td>\n",
       "      <td>0.191309</td>\n",
       "      <td>0.118795</td>\n",
       "      <td>-0.093861</td>\n",
       "      <td>-0.215216</td>\n",
       "      <td>0.057014</td>\n",
       "      <td>-0.366756</td>\n",
       "      <td>-0.027513</td>\n",
       "      <td>-0.155265</td>\n",
       "      <td>-0.513476</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Type       x_0       x_1       x_2       x_3       x_4       x_5       x_6  \\\n",
       "0     0  0.522512  0.257050 -0.145618 -0.398585 -0.265484 -0.143339 -0.218729   \n",
       "1     1  0.125956 -0.113120  0.127866 -0.243473 -0.246138 -0.521300 -0.748871   \n",
       "2     2 -0.232250  0.061611  0.147097 -0.102518 -0.194391  0.187656  0.050501   \n",
       "3     3 -0.973664 -0.103102 -0.763641 -0.235664 -0.247679 -0.155536 -0.031103   \n",
       "4     4 -0.479133  0.019136 -0.397532  0.206036 -0.124943  0.123089 -0.273900   \n",
       "\n",
       "        x_7       x_8  ...      x_40      x_41      x_42      x_43      x_44  \\\n",
       "0  0.085656  0.369437  ... -0.002601  0.076505 -0.656494 -0.241970  0.073937   \n",
       "1  0.055020  0.617417  ... -0.063075 -0.099647 -0.240150  0.063227 -0.303290   \n",
       "2 -0.135576 -0.077290  ... -0.073048 -0.367363  0.149369 -0.026017  0.453523   \n",
       "3  0.375137  0.245736  ...  0.255985 -0.509347  0.002029  0.039524  0.026820   \n",
       "4 -0.393271  0.112122  ...  0.244408  0.191309  0.118795 -0.093861 -0.215216   \n",
       "\n",
       "       x_45      x_46      x_47      x_48      x_49  \n",
       "0  0.180409 -0.234990  0.168578  0.168087 -0.028329  \n",
       "1 -0.113590 -0.244465  0.427193 -0.419203  0.152318  \n",
       "2 -0.281772 -0.091005  0.264164 -0.156693  0.045124  \n",
       "3 -0.065614  0.203438 -0.094295 -0.093370 -0.432650  \n",
       "4  0.057014 -0.366756 -0.027513 -0.155265 -0.513476  \n",
       "\n",
       "[5 rows x 51 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html_name = 'WL_Kernel_with_dependencies_{}_clusters_{}.html'\n",
    "name = 'WL Kernel with dependencies'\n",
    "\n",
    "path_to_d2v_model = 'FULL_wl_kernel_2_with_dependency_ru.model'\n",
    "path_to_csv_with_embeddings = 'C:\\\\Users\\\\User\\\\AuthorStyle\\\\CoNLL\\\\EXPERIMENTAL_EMBEDDING\\\\wl_kernel_2_with_dependency_vectors_dim_50_ru.csv'\n",
    "\n",
    "model = pickle.load(open(path_to_d2v_model, 'rb'))\n",
    "print(str(model))\n",
    "df = pd.read_csv(path_to_csv_with_embeddings)\n",
    "print('graph2vec shape: {}'.format(df.shape))\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X : (61889, 50)\n"
     ]
    }
   ],
   "source": [
    "numbers_type = df.Type.values\n",
    "X = df.drop('Type', axis=1).values\n",
    "\n",
    "print('Shape of X : {}'.format(X.shape))\n",
    "X2 = normalize(X)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Silhouette Score\n",
      "Silhouette Coefficient for 2 clusters: 0.06626236603558787\n",
      "Silhouette Coefficient for 3 clusters: 0.05621421636719863\n",
      "Silhouette Coefficient for 4 clusters: 0.052220893213846646\n",
      "Silhouette Coefficient for 5 clusters: 0.052428196024281226\n",
      "Silhouette Coefficient for 6 clusters: 0.04684816618689872\n",
      "Silhouette Coefficient for 7 clusters: 0.04208517060879941\n",
      "Silhouette Coefficient for 8 clusters: 0.03866158253551506\n",
      "Silhouette Coefficient for 9 clusters: 0.0351843692432015\n",
      "Silhouette Coefficient for 10 clusters: 0.035133030356791224\n",
      "Silhouette Coefficient for 11 clusters: 0.03309440143624586\n",
      "Silhouette Coefficient for 12 clusters: 0.03275982442548702\n",
      "Silhouette Coefficient for 13 clusters: 0.03118080784007762\n",
      "Silhouette Coefficient for 14 clusters: 0.031487807317844466\n",
      "Silhouette Coefficient for 15 clusters: 0.031299282870292267\n",
      "Silhouette Coefficient for 16 clusters: 0.029670215663010386\n",
      "Silhouette Coefficient for 17 clusters: 0.029256908776393788\n",
      "Silhouette Coefficient for 18 clusters: 0.02902878824718979\n",
      "Silhouette Coefficient for 19 clusters: 0.02697324347801037\n",
      "Silhouette Coefficient for 20 clusters: 0.027538687165943133\n",
      "Silhouette Coefficient for 21 clusters: 0.02707772719363644\n",
      "Silhouette Coefficient for 22 clusters: 0.024673354351848864\n",
      "Silhouette Coefficient for 23 clusters: 0.02527162275897493\n",
      "Silhouette Coefficient for 24 clusters: 0.024808274690748557\n",
      "Silhouette Coefficient for 25 clusters: 0.02500868096175038\n",
      "Silhouette Coefficient for 26 clusters: 0.025008858211920115\n",
      "Silhouette Coefficient for 27 clusters: 0.023703083289349895\n",
      "Silhouette Coefficient for 28 clusters: 0.02357821742463121\n",
      "Silhouette Coefficient for 29 clusters: 0.022563255524388346\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Silhouette Score\")\n",
    "silhouette_scores_list = []\n",
    "\n",
    "for cluster_number in range(2,30):\n",
    "    kMInstance = KMeans(n_clusters=cluster_number, init='k-means++', n_init=5)\n",
    "    kMInstance.fit(X2)\n",
    "    current_silhouette_score = silhouette_score(X2, kMInstance.labels_, metric='euclidean', sample_size=10000)\n",
    "    silhouette_scores_list.append((cluster_number, current_silhouette_score))\n",
    "    print(\"Silhouette Coefficient for {} clusters: {}\".format(cluster_number, current_silhouette_score))\n",
    "\n",
    "plt.plot(list(map(lambda x : x[0], silhouette_scores_list)),\n",
    "       list(map(lambda x : x[1], silhouette_scores_list)))\n",
    "\n",
    "plt.xlabel('Cluster Number')\n",
    "plt.ylabel('Silhouette Score')\n",
    "plt.title('Silhouette Score Analysis')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "OPTIMAL_CLUSTER_NUMBER = 18"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "kMInstance = KMeans(n_clusters=OPTIMAL_CLUSTER_NUMBER,\n",
    "                    init='k-means++', \n",
    "                    n_init=10)\n",
    "clusters = kMInstance.fit_predict(X2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels, counts = np.unique(clusters, return_counts=True)\n",
    "plt.bar(labels, counts, align='center')\n",
    "plt.gca().set_xticks(labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import cross_validate, cross_val_score, StratifiedKFold\n",
    "from collections import defaultdict\n",
    "from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.dummy import DummyClassifier\n",
    "import seaborn as sns; sns.set()\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_of_classes = defaultdict(set)\n",
    "doc2vec_instance = model[1].model\n",
    "htmlReport = '''<!DOCTYPE html>\n",
    "<html lang=\"en\">\n",
    "<head>\n",
    "<meta charset=\"UTF-8\" />\n",
    "<title>Embedded SVG</title>\n",
    "</head>\n",
    "<body>\n",
    "'''\n",
    "htmlReport += '''<h1>{}</h1>'''.format(name)\n",
    "for cluster_number in np.unique(clusters):\n",
    "    \n",
    "    htmlReport += '\\n<h2>Cluster #{}</h2>'.format(cluster_number)\n",
    "    centroid_of_current_cluster = kMInstance.cluster_centers_[cluster_number]\n",
    "    list_of_the_most_similar_vectors = doc2vec_instance.docvecs.most_similar([centroid_of_current_cluster], topn=12)\n",
    "    \n",
    "    for one_of_similar_vector in list_of_the_most_similar_vectors:\n",
    "        name = one_of_similar_vector[0]\n",
    "        position = int(name.split('_')[-1])\n",
    "        if clusters[position] != cluster_number:\n",
    "            continue\n",
    "        sent = graphs_and_sentences[position]['sentence']\n",
    "        htmlReport += '<p>Sentence (position = {}): {}</p>'.format(position, sent)\n",
    "        doc = ru_nlp(sent)\n",
    "        svg = displacy.render(doc, style='dep', options={'collapse_punct' : False, \n",
    "                                                         #'compact' : True\n",
    "                                                        })\n",
    "\n",
    "        htmlReport += svg\n",
    "        htmlReport += \"<hr>\"\n",
    "\n",
    "htmlReport += \"\"\"</body>\n",
    "</html>\"\"\"\n",
    "\n",
    "htmlReportFile = open(\n",
    "    os.path.join('Reports', html_name.format(OPTIMAL_CLUSTER_NUMBER, current_language)),\n",
    "    'w', encoding='utf-8')\n",
    "htmlReportFile.write(htmlReport)\n",
    "htmlReportFile.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# WL Kernel without dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<dep_tree_embedding.WLExtractor object at 0x000001BC1D4F80F0>, <dep_tree_embedding.DepTreeEmbedding object at 0x000001BC1D4F8160>)\n",
      "graph2vec shape: (61889, 51)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Type</th>\n",
       "      <th>x_0</th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>x_5</th>\n",
       "      <th>x_6</th>\n",
       "      <th>x_7</th>\n",
       "      <th>x_8</th>\n",
       "      <th>...</th>\n",
       "      <th>x_40</th>\n",
       "      <th>x_41</th>\n",
       "      <th>x_42</th>\n",
       "      <th>x_43</th>\n",
       "      <th>x_44</th>\n",
       "      <th>x_45</th>\n",
       "      <th>x_46</th>\n",
       "      <th>x_47</th>\n",
       "      <th>x_48</th>\n",
       "      <th>x_49</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.037602</td>\n",
       "      <td>0.806286</td>\n",
       "      <td>0.436710</td>\n",
       "      <td>-0.176122</td>\n",
       "      <td>-0.381618</td>\n",
       "      <td>0.516263</td>\n",
       "      <td>-0.672581</td>\n",
       "      <td>-0.001295</td>\n",
       "      <td>0.240555</td>\n",
       "      <td>...</td>\n",
       "      <td>0.142585</td>\n",
       "      <td>-0.512157</td>\n",
       "      <td>0.035070</td>\n",
       "      <td>-0.109651</td>\n",
       "      <td>-0.074435</td>\n",
       "      <td>-0.179311</td>\n",
       "      <td>0.080432</td>\n",
       "      <td>0.114086</td>\n",
       "      <td>-0.271878</td>\n",
       "      <td>0.230591</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.116713</td>\n",
       "      <td>0.282714</td>\n",
       "      <td>0.067951</td>\n",
       "      <td>0.216779</td>\n",
       "      <td>-0.418636</td>\n",
       "      <td>-0.155188</td>\n",
       "      <td>-0.351700</td>\n",
       "      <td>0.057558</td>\n",
       "      <td>0.346293</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.430395</td>\n",
       "      <td>-0.572989</td>\n",
       "      <td>0.229403</td>\n",
       "      <td>-0.207340</td>\n",
       "      <td>0.391145</td>\n",
       "      <td>0.214434</td>\n",
       "      <td>-0.193132</td>\n",
       "      <td>-0.520221</td>\n",
       "      <td>-0.307543</td>\n",
       "      <td>-0.117942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.404255</td>\n",
       "      <td>0.193987</td>\n",
       "      <td>0.673277</td>\n",
       "      <td>0.657761</td>\n",
       "      <td>-0.410019</td>\n",
       "      <td>0.307626</td>\n",
       "      <td>0.194749</td>\n",
       "      <td>0.013635</td>\n",
       "      <td>-0.327872</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.744466</td>\n",
       "      <td>-0.260160</td>\n",
       "      <td>-0.026546</td>\n",
       "      <td>-0.145218</td>\n",
       "      <td>-0.173361</td>\n",
       "      <td>0.391810</td>\n",
       "      <td>-0.310338</td>\n",
       "      <td>-0.143581</td>\n",
       "      <td>0.073716</td>\n",
       "      <td>0.091555</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.095250</td>\n",
       "      <td>0.349020</td>\n",
       "      <td>-0.175190</td>\n",
       "      <td>-0.056710</td>\n",
       "      <td>-0.256382</td>\n",
       "      <td>0.675808</td>\n",
       "      <td>0.095214</td>\n",
       "      <td>-0.031617</td>\n",
       "      <td>0.001755</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.111078</td>\n",
       "      <td>0.117545</td>\n",
       "      <td>0.113698</td>\n",
       "      <td>0.400416</td>\n",
       "      <td>0.047402</td>\n",
       "      <td>0.752158</td>\n",
       "      <td>-0.328100</td>\n",
       "      <td>0.143358</td>\n",
       "      <td>0.272722</td>\n",
       "      <td>0.457913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.402822</td>\n",
       "      <td>0.008184</td>\n",
       "      <td>0.262642</td>\n",
       "      <td>-0.142328</td>\n",
       "      <td>0.302339</td>\n",
       "      <td>0.061752</td>\n",
       "      <td>-0.635233</td>\n",
       "      <td>0.259765</td>\n",
       "      <td>0.263220</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.736899</td>\n",
       "      <td>-0.119125</td>\n",
       "      <td>-0.370502</td>\n",
       "      <td>0.131716</td>\n",
       "      <td>0.361468</td>\n",
       "      <td>0.357344</td>\n",
       "      <td>-1.030990</td>\n",
       "      <td>-0.075151</td>\n",
       "      <td>-0.222653</td>\n",
       "      <td>0.012978</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Type       x_0       x_1       x_2       x_3       x_4       x_5       x_6  \\\n",
       "0     0  0.037602  0.806286  0.436710 -0.176122 -0.381618  0.516263 -0.672581   \n",
       "1     1  0.116713  0.282714  0.067951  0.216779 -0.418636 -0.155188 -0.351700   \n",
       "2     2  0.404255  0.193987  0.673277  0.657761 -0.410019  0.307626  0.194749   \n",
       "3     3  0.095250  0.349020 -0.175190 -0.056710 -0.256382  0.675808  0.095214   \n",
       "4     4  0.402822  0.008184  0.262642 -0.142328  0.302339  0.061752 -0.635233   \n",
       "\n",
       "        x_7       x_8  ...      x_40      x_41      x_42      x_43      x_44  \\\n",
       "0 -0.001295  0.240555  ...  0.142585 -0.512157  0.035070 -0.109651 -0.074435   \n",
       "1  0.057558  0.346293  ... -0.430395 -0.572989  0.229403 -0.207340  0.391145   \n",
       "2  0.013635 -0.327872  ... -0.744466 -0.260160 -0.026546 -0.145218 -0.173361   \n",
       "3 -0.031617  0.001755  ... -0.111078  0.117545  0.113698  0.400416  0.047402   \n",
       "4  0.259765  0.263220  ... -0.736899 -0.119125 -0.370502  0.131716  0.361468   \n",
       "\n",
       "       x_45      x_46      x_47      x_48      x_49  \n",
       "0 -0.179311  0.080432  0.114086 -0.271878  0.230591  \n",
       "1  0.214434 -0.193132 -0.520221 -0.307543 -0.117942  \n",
       "2  0.391810 -0.310338 -0.143581  0.073716  0.091555  \n",
       "3  0.752158 -0.328100  0.143358  0.272722  0.457913  \n",
       "4  0.357344 -1.030990 -0.075151 -0.222653  0.012978  \n",
       "\n",
       "[5 rows x 51 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html_name = 'WL_Kernel_without_dependencies_{}_clusters_{}.html'\n",
    "name = 'WL Kernel without dependencies'\n",
    "\n",
    "path_to_d2v_model = 'FULL_wl_kernel_2_without_dependency_ru.model'\n",
    "path_to_csv_with_embeddings = 'C:\\\\Users\\\\User\\\\AuthorStyle\\\\CoNLL\\\\EXPERIMENTAL_EMBEDDING\\\\wl_kernel_2_without_dependency_vectors_dim_50_ru.csv'\n",
    "\n",
    "\n",
    "\n",
    "model = pickle.load(open(path_to_d2v_model, 'rb'))\n",
    "print(str(model))\n",
    "df = pd.read_csv(path_to_csv_with_embeddings)\n",
    "print('graph2vec shape: {}'.format(df.shape))\n",
    "df.head()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X : (61889, 50)\n",
      "Silhouette Score\n",
      "Silhouette Coefficient for 2 clusters: 0.06275651200881875\n",
      "Silhouette Coefficient for 3 clusters: 0.057198303661223034\n",
      "Silhouette Coefficient for 4 clusters: 0.04834744796184395\n",
      "Silhouette Coefficient for 5 clusters: 0.046920866552012434\n",
      "Silhouette Coefficient for 6 clusters: 0.04006494793832707\n",
      "Silhouette Coefficient for 7 clusters: 0.036172586580014834\n",
      "Silhouette Coefficient for 8 clusters: 0.03696136423212062\n",
      "Silhouette Coefficient for 9 clusters: 0.034697963038252566\n",
      "Silhouette Coefficient for 10 clusters: 0.03494978620677491\n",
      "Silhouette Coefficient for 11 clusters: 0.03439847620616444\n",
      "Silhouette Coefficient for 12 clusters: 0.032084609696888464\n",
      "Silhouette Coefficient for 13 clusters: 0.03269747727954126\n",
      "Silhouette Coefficient for 14 clusters: 0.02859657323918832\n",
      "Silhouette Coefficient for 15 clusters: 0.029387647246373463\n",
      "Silhouette Coefficient for 16 clusters: 0.026008338577191016\n",
      "Silhouette Coefficient for 17 clusters: 0.026665639574058885\n",
      "Silhouette Coefficient for 18 clusters: 0.02517775546311986\n",
      "Silhouette Coefficient for 19 clusters: 0.02443155068495554\n",
      "Silhouette Coefficient for 20 clusters: 0.023201660076608132\n",
      "Silhouette Coefficient for 21 clusters: 0.022642125619182336\n",
      "Silhouette Coefficient for 22 clusters: 0.023792400046742046\n",
      "Silhouette Coefficient for 23 clusters: 0.0213790157555833\n",
      "Silhouette Coefficient for 24 clusters: 0.02180777077658868\n",
      "Silhouette Coefficient for 25 clusters: 0.022304917398485107\n",
      "Silhouette Coefficient for 26 clusters: 0.022196785326749506\n",
      "Silhouette Coefficient for 27 clusters: 0.018899360551860613\n",
      "Silhouette Coefficient for 28 clusters: 0.02180030010115376\n",
      "Silhouette Coefficient for 29 clusters: 0.019765668833301307\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "numbers_type = df.Type.values\n",
    "X = df.drop('Type', axis=1).values\n",
    "\n",
    "print('Shape of X : {}'.format(X.shape))\n",
    "X2 = normalize(X)\n",
    "\n",
    "\n",
    "print(\"Silhouette Score\")\n",
    "silhouette_scores_list = []\n",
    "\n",
    "for cluster_number in range(2,30):\n",
    "    kMInstance = KMeans(n_clusters=cluster_number, init='k-means++', n_init=5)\n",
    "    kMInstance.fit(X2)\n",
    "    current_silhouette_score = silhouette_score(X2, kMInstance.labels_, metric='euclidean', sample_size=10000)\n",
    "    silhouette_scores_list.append((cluster_number, current_silhouette_score))\n",
    "    print(\"Silhouette Coefficient for {} clusters: {}\".format(cluster_number, current_silhouette_score))\n",
    "\n",
    "plt.plot(list(map(lambda x : x[0], silhouette_scores_list)),\n",
    "       list(map(lambda x : x[1], silhouette_scores_list)))\n",
    "\n",
    "plt.xlabel('Cluster Number')\n",
    "plt.ylabel('Silhouette Score')\n",
    "plt.title('Silhouette Score Analysis')\n",
    "plt.grid(True)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "OPTIMAL_CLUSTER_NUMBER = 22\n",
    "\n",
    "kMInstance = KMeans(n_clusters=OPTIMAL_CLUSTER_NUMBER,\n",
    "                    init='k-means++', \n",
    "                    n_init=10)\n",
    "clusters = kMInstance.fit_predict(X2)\n",
    "\n",
    "### Distribution\n",
    "\n",
    "labels, counts = np.unique(clusters, return_counts=True)\n",
    "plt.bar(labels, counts, align='center')\n",
    "plt.gca().set_xticks(labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "doc2vec_instance = model[1].model\n",
    "dict_of_classes = defaultdict(set)\n",
    "\n",
    "htmlReport = '''<!DOCTYPE html>\n",
    "<html lang=\"en\">\n",
    "<head>\n",
    "<meta charset=\"UTF-8\" />\n",
    "<title>Embedded SVG</title>\n",
    "</head>\n",
    "<body>\n",
    "'''\n",
    "htmlReport += '''<h1>{}</h1>'''.format(name)\n",
    "for cluster_number in np.unique(clusters):\n",
    "    \n",
    "    htmlReport += '\\n<h2>Cluster #{}</h2>'.format(cluster_number)\n",
    "    centroid_of_current_cluster = kMInstance.cluster_centers_[cluster_number]\n",
    "    list_of_the_most_similar_vectors = doc2vec_instance.docvecs.most_similar([centroid_of_current_cluster], topn=12)\n",
    "    \n",
    "    for one_of_similar_vector in list_of_the_most_similar_vectors:\n",
    "        name = one_of_similar_vector[0]\n",
    "        position = int(name.split('_')[-1])\n",
    "        if clusters[position] != cluster_number:\n",
    "            continue\n",
    "        sent = graphs_and_sentences[position]['sentence']\n",
    "        htmlReport += '<p>Sentence (position = {}): {}</p>'.format(position, sent)\n",
    "        doc = ru_nlp(sent)\n",
    "        svg = displacy.render(doc, style='dep', options={'collapse_punct' : False, \n",
    "                                                         #'compact' : True\n",
    "                                                        })\n",
    "\n",
    "        htmlReport += svg\n",
    "        htmlReport += \"<hr>\"\n",
    "\n",
    "htmlReport += \"\"\"</body>\n",
    "</html>\"\"\"\n",
    "\n",
    "htmlReportFile = open(\n",
    "    os.path.join('Reports', html_name.format(OPTIMAL_CLUSTER_NUMBER, current_language)),\n",
    "    'w', encoding='utf-8')\n",
    "htmlReportFile.write(htmlReport)\n",
    "htmlReportFile.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Simple paths"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<dep_tree_embedding.SimplePathExtractor object at 0x000001BC1C67A898>, <dep_tree_embedding.DepTreeEmbedding object at 0x000001BC586C5978>)\n",
      "graph2vec shape: (61889, 51)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Type</th>\n",
       "      <th>x_0</th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>x_5</th>\n",
       "      <th>x_6</th>\n",
       "      <th>x_7</th>\n",
       "      <th>x_8</th>\n",
       "      <th>...</th>\n",
       "      <th>x_40</th>\n",
       "      <th>x_41</th>\n",
       "      <th>x_42</th>\n",
       "      <th>x_43</th>\n",
       "      <th>x_44</th>\n",
       "      <th>x_45</th>\n",
       "      <th>x_46</th>\n",
       "      <th>x_47</th>\n",
       "      <th>x_48</th>\n",
       "      <th>x_49</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.895372</td>\n",
       "      <td>0.249476</td>\n",
       "      <td>1.008298</td>\n",
       "      <td>-0.072381</td>\n",
       "      <td>-0.284903</td>\n",
       "      <td>0.938598</td>\n",
       "      <td>-0.392589</td>\n",
       "      <td>-0.272816</td>\n",
       "      <td>-0.225930</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.288795</td>\n",
       "      <td>0.480338</td>\n",
       "      <td>-0.028810</td>\n",
       "      <td>-0.963966</td>\n",
       "      <td>-0.911812</td>\n",
       "      <td>0.275412</td>\n",
       "      <td>-0.289885</td>\n",
       "      <td>0.164613</td>\n",
       "      <td>-0.830212</td>\n",
       "      <td>-0.027437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>-0.160846</td>\n",
       "      <td>-0.062222</td>\n",
       "      <td>0.327905</td>\n",
       "      <td>0.483652</td>\n",
       "      <td>-0.114403</td>\n",
       "      <td>0.417911</td>\n",
       "      <td>0.795983</td>\n",
       "      <td>-0.573350</td>\n",
       "      <td>0.699927</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.614480</td>\n",
       "      <td>-0.662076</td>\n",
       "      <td>-1.203910</td>\n",
       "      <td>1.111488</td>\n",
       "      <td>-0.105999</td>\n",
       "      <td>-0.418944</td>\n",
       "      <td>-0.695406</td>\n",
       "      <td>0.066845</td>\n",
       "      <td>-1.163377</td>\n",
       "      <td>1.081137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>1.186694</td>\n",
       "      <td>0.717619</td>\n",
       "      <td>0.761737</td>\n",
       "      <td>-0.215127</td>\n",
       "      <td>-0.604000</td>\n",
       "      <td>-0.520503</td>\n",
       "      <td>-1.285256</td>\n",
       "      <td>-0.176546</td>\n",
       "      <td>0.943991</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.568837</td>\n",
       "      <td>-0.058513</td>\n",
       "      <td>-0.405351</td>\n",
       "      <td>-0.491381</td>\n",
       "      <td>-1.724948</td>\n",
       "      <td>-0.090375</td>\n",
       "      <td>0.158659</td>\n",
       "      <td>0.302037</td>\n",
       "      <td>-0.317165</td>\n",
       "      <td>0.261320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>-0.124294</td>\n",
       "      <td>0.613017</td>\n",
       "      <td>0.830139</td>\n",
       "      <td>-0.536549</td>\n",
       "      <td>0.216531</td>\n",
       "      <td>0.517900</td>\n",
       "      <td>-0.735545</td>\n",
       "      <td>0.356909</td>\n",
       "      <td>0.832599</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.667444</td>\n",
       "      <td>0.430519</td>\n",
       "      <td>-0.874306</td>\n",
       "      <td>0.081627</td>\n",
       "      <td>-1.127759</td>\n",
       "      <td>0.008527</td>\n",
       "      <td>0.252356</td>\n",
       "      <td>0.765013</td>\n",
       "      <td>-1.084654</td>\n",
       "      <td>0.434188</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.056991</td>\n",
       "      <td>0.660093</td>\n",
       "      <td>0.556052</td>\n",
       "      <td>-0.614171</td>\n",
       "      <td>-0.299939</td>\n",
       "      <td>-0.464508</td>\n",
       "      <td>-0.536911</td>\n",
       "      <td>1.452551</td>\n",
       "      <td>1.347900</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.008559</td>\n",
       "      <td>-0.143607</td>\n",
       "      <td>0.017452</td>\n",
       "      <td>1.143364</td>\n",
       "      <td>-0.113435</td>\n",
       "      <td>-0.470418</td>\n",
       "      <td>-0.946289</td>\n",
       "      <td>-0.851114</td>\n",
       "      <td>-2.319543</td>\n",
       "      <td>0.072105</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Type       x_0       x_1       x_2       x_3       x_4       x_5       x_6  \\\n",
       "0     0 -0.895372  0.249476  1.008298 -0.072381 -0.284903  0.938598 -0.392589   \n",
       "1     1 -0.160846 -0.062222  0.327905  0.483652 -0.114403  0.417911  0.795983   \n",
       "2     2  1.186694  0.717619  0.761737 -0.215127 -0.604000 -0.520503 -1.285256   \n",
       "3     3 -0.124294  0.613017  0.830139 -0.536549  0.216531  0.517900 -0.735545   \n",
       "4     4  0.056991  0.660093  0.556052 -0.614171 -0.299939 -0.464508 -0.536911   \n",
       "\n",
       "        x_7       x_8  ...      x_40      x_41      x_42      x_43      x_44  \\\n",
       "0 -0.272816 -0.225930  ... -1.288795  0.480338 -0.028810 -0.963966 -0.911812   \n",
       "1 -0.573350  0.699927  ... -1.614480 -0.662076 -1.203910  1.111488 -0.105999   \n",
       "2 -0.176546  0.943991  ... -0.568837 -0.058513 -0.405351 -0.491381 -1.724948   \n",
       "3  0.356909  0.832599  ... -0.667444  0.430519 -0.874306  0.081627 -1.127759   \n",
       "4  1.452551  1.347900  ... -1.008559 -0.143607  0.017452  1.143364 -0.113435   \n",
       "\n",
       "       x_45      x_46      x_47      x_48      x_49  \n",
       "0  0.275412 -0.289885  0.164613 -0.830212 -0.027437  \n",
       "1 -0.418944 -0.695406  0.066845 -1.163377  1.081137  \n",
       "2 -0.090375  0.158659  0.302037 -0.317165  0.261320  \n",
       "3  0.008527  0.252356  0.765013 -1.084654  0.434188  \n",
       "4 -0.470418 -0.946289 -0.851114 -2.319543  0.072105  \n",
       "\n",
       "[5 rows x 51 columns]"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html_name = 'Simple_Path_Kernel_{}_clusters_{}.html'\n",
    "name = 'Simple Path Kernel'\n",
    "\n",
    "path_to_d2v_model = 'FULL_path_extractor_ru.model'\n",
    "\n",
    "\n",
    "\n",
    "model = pickle.load(open(path_to_d2v_model, 'rb'))\n",
    "print(str(model))\n",
    "df = pd.read_csv(path_to_csv_with_embeddings)\n",
    "print('graph2vec shape: {}'.format(df.shape))\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X : (61889, 50)\n"
     ]
    }
   ],
   "source": [
    "numbers_type = df.Type.values\n",
    "X = df.drop('Type', axis=1).values\n",
    "\n",
    "print('Shape of X : {}'.format(X.shape))\n",
    "X2 = normalize(X)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Silhouette Score\n",
      "Silhouette Coefficient for 2 clusters: 0.08018827210503235\n",
      "Silhouette Coefficient for 3 clusters: 0.06904636909368604\n",
      "Silhouette Coefficient for 4 clusters: 0.058286077882283335\n",
      "Silhouette Coefficient for 5 clusters: 0.059138203835225146\n",
      "Silhouette Coefficient for 6 clusters: 0.06420548707438366\n",
      "Silhouette Coefficient for 7 clusters: 0.0632518581256645\n",
      "Silhouette Coefficient for 8 clusters: 0.064046376490147\n",
      "Silhouette Coefficient for 9 clusters: 0.06828183319425754\n",
      "Silhouette Coefficient for 10 clusters: 0.06513963213668496\n",
      "Silhouette Coefficient for 11 clusters: 0.06977546283300053\n",
      "Silhouette Coefficient for 12 clusters: 0.06250601013175677\n",
      "Silhouette Coefficient for 13 clusters: 0.06305456347730835\n",
      "Silhouette Coefficient for 14 clusters: 0.05039519669286394\n",
      "Silhouette Coefficient for 15 clusters: 0.06020714762948828\n",
      "Silhouette Coefficient for 16 clusters: 0.0599642710020114\n",
      "Silhouette Coefficient for 17 clusters: 0.05568243164780963\n",
      "Silhouette Coefficient for 18 clusters: 0.0579169465291513\n",
      "Silhouette Coefficient for 19 clusters: 0.054529822245564755\n",
      "Silhouette Coefficient for 20 clusters: 0.055597691798766266\n",
      "Silhouette Coefficient for 21 clusters: 0.05722446509595668\n",
      "Silhouette Coefficient for 22 clusters: 0.04886239699911771\n",
      "Silhouette Coefficient for 23 clusters: 0.047830801144045185\n",
      "Silhouette Coefficient for 24 clusters: 0.047470833667190966\n",
      "Silhouette Coefficient for 25 clusters: 0.04254049497792406\n",
      "Silhouette Coefficient for 26 clusters: 0.04409070800418517\n",
      "Silhouette Coefficient for 27 clusters: 0.028204550524311083\n",
      "Silhouette Coefficient for 28 clusters: 0.04788749071897244\n",
      "Silhouette Coefficient for 29 clusters: 0.04275256877754864\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Silhouette Score\")\n",
    "silhouette_scores_list = []\n",
    "\n",
    "for cluster_number in range(2,30):\n",
    "    kMInstance = KMeans(n_clusters=cluster_number, init='k-means++', n_init=5)\n",
    "    kMInstance.fit(X2)\n",
    "    current_silhouette_score = silhouette_score(X2, kMInstance.labels_, metric='euclidean', sample_size=10000)\n",
    "    silhouette_scores_list.append((cluster_number, current_silhouette_score))\n",
    "    print(\"Silhouette Coefficient for {} clusters: {}\".format(cluster_number, current_silhouette_score))\n",
    "\n",
    "plt.plot(list(map(lambda x : x[0], silhouette_scores_list)),\n",
    "       list(map(lambda x : x[1], silhouette_scores_list)))\n",
    "\n",
    "plt.xlabel('Cluster Number')\n",
    "plt.ylabel('Silhouette Score')\n",
    "plt.title('Silhouette Score Analysis')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "OPTIMAL_CLUSTER_NUMBER = 21"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "kMInstance = KMeans(n_clusters=OPTIMAL_CLUSTER_NUMBER,\n",
    "                    init='k-means++', \n",
    "                    n_init=10)\n",
    "clusters = kMInstance.fit_predict(X2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels, counts = np.unique(clusters, return_counts=True)\n",
    "plt.bar(labels, counts, align='center')\n",
    "plt.gca().set_xticks(labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "htmlReport = '''<!DOCTYPE html>\n",
    "<html lang=\"en\">\n",
    "<head>\n",
    "<meta charset=\"UTF-8\" />\n",
    "<title>Embedded SVG</title>\n",
    "</head>\n",
    "<body>\n",
    "'''\n",
    "htmlReport += '''<h1>{}</h1>'''.format(name)\n",
    "for cluster_number in np.unique(clusters):\n",
    "    \n",
    "    htmlReport += '\\n<h2>Cluster #{}</h2>'.format(cluster_number)\n",
    "    centroid_of_current_cluster = kMInstance.cluster_centers_[cluster_number]\n",
    "    list_of_the_most_similar_vectors = model[1].model.docvecs.most_similar([centroid_of_current_cluster], topn=10)\n",
    "    \n",
    "    for one_of_similar_vector in list_of_the_most_similar_vectors:\n",
    "        name = one_of_similar_vector[0]\n",
    "        position = int(name.split('_')[-1])\n",
    "        if clusters[position] != cluster_number:\n",
    "            continue\n",
    "        sent = graphs_and_sentences[position]['sentence']\n",
    "        htmlReport += '<p>Sentence (position = {}): {}</p>'.format(position, sent)\n",
    "        doc = ru_nlp(sent)\n",
    "        svg = displacy.render(doc, style='dep', options={'collapse_punct' : False, \n",
    "                                                         #'compact' : True\n",
    "                                                        })\n",
    "        htmlReport += svg\n",
    "        htmlReport += \"<hr>\"\n",
    "\n",
    "htmlReport += \"\"\"</body>\n",
    "</html>\"\"\"\n",
    "\n",
    "htmlReportFile = open(\n",
    "    os.path.join('Reports', html_name.format(OPTIMAL_CLUSTER_NUMBER, current_language)),\n",
    "    'w', encoding='utf-8')\n",
    "htmlReportFile.write(htmlReport)\n",
    "htmlReportFile.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Contracted Nodes Kernel with dependencies\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<dep_tree_embedding.ContractedFeatureExtractor object at 0x000001BC5F783CF8>, <dep_tree_embedding.DepTreeEmbedding object at 0x000001BC113C5630>)\n",
      "graph2vec shape: (61889, 51)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Type</th>\n",
       "      <th>x_0</th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>x_5</th>\n",
       "      <th>x_6</th>\n",
       "      <th>x_7</th>\n",
       "      <th>x_8</th>\n",
       "      <th>...</th>\n",
       "      <th>x_40</th>\n",
       "      <th>x_41</th>\n",
       "      <th>x_42</th>\n",
       "      <th>x_43</th>\n",
       "      <th>x_44</th>\n",
       "      <th>x_45</th>\n",
       "      <th>x_46</th>\n",
       "      <th>x_47</th>\n",
       "      <th>x_48</th>\n",
       "      <th>x_49</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.103616</td>\n",
       "      <td>0.454216</td>\n",
       "      <td>0.661439</td>\n",
       "      <td>0.080668</td>\n",
       "      <td>-0.487843</td>\n",
       "      <td>-0.043934</td>\n",
       "      <td>-0.135324</td>\n",
       "      <td>0.305138</td>\n",
       "      <td>-0.378445</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.009001</td>\n",
       "      <td>0.076024</td>\n",
       "      <td>-0.391137</td>\n",
       "      <td>0.316226</td>\n",
       "      <td>0.169799</td>\n",
       "      <td>0.454038</td>\n",
       "      <td>-0.366445</td>\n",
       "      <td>-0.065790</td>\n",
       "      <td>-0.441295</td>\n",
       "      <td>-0.076138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.020970</td>\n",
       "      <td>0.094200</td>\n",
       "      <td>0.447375</td>\n",
       "      <td>0.336880</td>\n",
       "      <td>-0.163178</td>\n",
       "      <td>-0.178024</td>\n",
       "      <td>-0.335563</td>\n",
       "      <td>0.061683</td>\n",
       "      <td>-0.201310</td>\n",
       "      <td>...</td>\n",
       "      <td>0.175824</td>\n",
       "      <td>0.031507</td>\n",
       "      <td>0.245940</td>\n",
       "      <td>0.440801</td>\n",
       "      <td>0.516355</td>\n",
       "      <td>0.385501</td>\n",
       "      <td>-0.341830</td>\n",
       "      <td>0.156132</td>\n",
       "      <td>-0.269613</td>\n",
       "      <td>-0.021296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.328347</td>\n",
       "      <td>0.130693</td>\n",
       "      <td>0.786791</td>\n",
       "      <td>0.235939</td>\n",
       "      <td>-0.005866</td>\n",
       "      <td>-0.139739</td>\n",
       "      <td>-0.235796</td>\n",
       "      <td>-0.017536</td>\n",
       "      <td>-0.101509</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.078379</td>\n",
       "      <td>-0.104432</td>\n",
       "      <td>0.183406</td>\n",
       "      <td>0.398201</td>\n",
       "      <td>-0.252036</td>\n",
       "      <td>0.673237</td>\n",
       "      <td>-0.660485</td>\n",
       "      <td>0.005587</td>\n",
       "      <td>-0.216375</td>\n",
       "      <td>0.474494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.095271</td>\n",
       "      <td>0.061687</td>\n",
       "      <td>0.079709</td>\n",
       "      <td>0.126707</td>\n",
       "      <td>-0.417104</td>\n",
       "      <td>-0.061243</td>\n",
       "      <td>-0.413393</td>\n",
       "      <td>0.266885</td>\n",
       "      <td>-0.156141</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.178429</td>\n",
       "      <td>-0.127240</td>\n",
       "      <td>0.352781</td>\n",
       "      <td>-0.249899</td>\n",
       "      <td>0.094378</td>\n",
       "      <td>0.002144</td>\n",
       "      <td>-0.755316</td>\n",
       "      <td>-0.042940</td>\n",
       "      <td>-0.135242</td>\n",
       "      <td>0.391758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.195974</td>\n",
       "      <td>0.507407</td>\n",
       "      <td>0.262294</td>\n",
       "      <td>0.085478</td>\n",
       "      <td>0.361244</td>\n",
       "      <td>-0.253245</td>\n",
       "      <td>0.069682</td>\n",
       "      <td>0.362219</td>\n",
       "      <td>0.177376</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.165069</td>\n",
       "      <td>-0.205077</td>\n",
       "      <td>0.322228</td>\n",
       "      <td>-0.017558</td>\n",
       "      <td>0.252894</td>\n",
       "      <td>0.204401</td>\n",
       "      <td>-0.880028</td>\n",
       "      <td>-0.228738</td>\n",
       "      <td>0.229081</td>\n",
       "      <td>0.115928</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Type       x_0       x_1       x_2       x_3       x_4       x_5       x_6  \\\n",
       "0     0 -0.103616  0.454216  0.661439  0.080668 -0.487843 -0.043934 -0.135324   \n",
       "1     1  0.020970  0.094200  0.447375  0.336880 -0.163178 -0.178024 -0.335563   \n",
       "2     2  0.328347  0.130693  0.786791  0.235939 -0.005866 -0.139739 -0.235796   \n",
       "3     3  0.095271  0.061687  0.079709  0.126707 -0.417104 -0.061243 -0.413393   \n",
       "4     4  0.195974  0.507407  0.262294  0.085478  0.361244 -0.253245  0.069682   \n",
       "\n",
       "        x_7       x_8  ...      x_40      x_41      x_42      x_43      x_44  \\\n",
       "0  0.305138 -0.378445  ... -0.009001  0.076024 -0.391137  0.316226  0.169799   \n",
       "1  0.061683 -0.201310  ...  0.175824  0.031507  0.245940  0.440801  0.516355   \n",
       "2 -0.017536 -0.101509  ... -0.078379 -0.104432  0.183406  0.398201 -0.252036   \n",
       "3  0.266885 -0.156141  ... -0.178429 -0.127240  0.352781 -0.249899  0.094378   \n",
       "4  0.362219  0.177376  ... -0.165069 -0.205077  0.322228 -0.017558  0.252894   \n",
       "\n",
       "       x_45      x_46      x_47      x_48      x_49  \n",
       "0  0.454038 -0.366445 -0.065790 -0.441295 -0.076138  \n",
       "1  0.385501 -0.341830  0.156132 -0.269613 -0.021296  \n",
       "2  0.673237 -0.660485  0.005587 -0.216375  0.474494  \n",
       "3  0.002144 -0.755316 -0.042940 -0.135242  0.391758  \n",
       "4  0.204401 -0.880028 -0.228738  0.229081  0.115928  \n",
       "\n",
       "[5 rows x 51 columns]"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "html_name = 'Contracted_Kernel_with_dependencies_{}_clusters_{}.html'\n",
    "name = 'Contracted Kernel with dependencies'\n",
    "\n",
    "path_to_d2v_model = 'FULL_contracted_with_dependency_ru.model'\n",
    "path_to_csv_with_embeddings = 'C:\\\\Users\\\\User\\\\AuthorStyle\\\\CoNLL\\\\EXPERIMENTAL_EMBEDDING\\\\contracted_with_dependency_vectors_dim_50_ru.csv'\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "model = pickle.load(open(path_to_d2v_model, 'rb'))\n",
    "print(str(model))\n",
    "df = pd.read_csv(path_to_csv_with_embeddings)\n",
    "print('graph2vec shape: {}'.format(df.shape))\n",
    "df.head()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X : (61889, 50)\n",
      "Silhouette Score\n",
      "Silhouette Coefficient for 2 clusters: 0.05573778165828407\n",
      "Silhouette Coefficient for 3 clusters: 0.04860999899717607\n",
      "Silhouette Coefficient for 4 clusters: 0.04208575075198232\n",
      "Silhouette Coefficient for 5 clusters: 0.04106008963908093\n",
      "Silhouette Coefficient for 6 clusters: 0.034009456579197996\n",
      "Silhouette Coefficient for 7 clusters: 0.028279652557831582\n",
      "Silhouette Coefficient for 8 clusters: 0.02653349921429062\n",
      "Silhouette Coefficient for 9 clusters: 0.02353020209622561\n",
      "Silhouette Coefficient for 10 clusters: 0.023610633673880634\n",
      "Silhouette Coefficient for 11 clusters: 0.02544175009241476\n",
      "Silhouette Coefficient for 12 clusters: 0.018728002595509753\n",
      "Silhouette Coefficient for 13 clusters: 0.02333289386838351\n",
      "Silhouette Coefficient for 14 clusters: 0.024088098392952825\n",
      "Silhouette Coefficient for 15 clusters: 0.02342401398266627\n",
      "Silhouette Coefficient for 16 clusters: 0.020026595218170497\n",
      "Silhouette Coefficient for 17 clusters: 0.01966489370737512\n",
      "Silhouette Coefficient for 18 clusters: 0.020235832335101754\n",
      "Silhouette Coefficient for 19 clusters: 0.017142311516237425\n",
      "Silhouette Coefficient for 20 clusters: 0.02019884750075935\n",
      "Silhouette Coefficient for 21 clusters: 0.020404160582868175\n",
      "Silhouette Coefficient for 22 clusters: 0.021197919712492103\n",
      "Silhouette Coefficient for 23 clusters: 0.020588516344494646\n",
      "Silhouette Coefficient for 24 clusters: 0.018401998858892998\n",
      "Silhouette Coefficient for 25 clusters: 0.01929559777104867\n",
      "Silhouette Coefficient for 26 clusters: 0.02059072639783754\n",
      "Silhouette Coefficient for 27 clusters: 0.0190895614144254\n",
      "Silhouette Coefficient for 28 clusters: 0.019902219104212956\n",
      "Silhouette Coefficient for 29 clusters: 0.01686612214529574\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "numbers_type = df.Type.values\n",
    "X = df.drop('Type', axis=1).values\n",
    "\n",
    "print('Shape of X : {}'.format(X.shape))\n",
    "X2 = normalize(X)\n",
    "\n",
    "\n",
    "print(\"Silhouette Score\")\n",
    "silhouette_scores_list = []\n",
    "\n",
    "for cluster_number in range(2,30):\n",
    "    kMInstance = KMeans(n_clusters=cluster_number, init='k-means++', n_init=5)\n",
    "    kMInstance.fit(X2)\n",
    "    current_silhouette_score = silhouette_score(X2, kMInstance.labels_, metric='euclidean', sample_size=10000)\n",
    "    silhouette_scores_list.append((cluster_number, current_silhouette_score))\n",
    "    print(\"Silhouette Coefficient for {} clusters: {}\".format(cluster_number, current_silhouette_score))\n",
    "\n",
    "plt.plot(list(map(lambda x : x[0], silhouette_scores_list)),\n",
    "       list(map(lambda x : x[1], silhouette_scores_list)))\n",
    "\n",
    "plt.xlabel('Cluster Number')\n",
    "plt.ylabel('Silhouette Score')\n",
    "plt.title('Silhouette Score Analysis')\n",
    "plt.grid(True)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "OPTIMAL_CLUSTER_NUMBER = 22\n",
    "\n",
    "kMInstance = KMeans(n_clusters=OPTIMAL_CLUSTER_NUMBER,\n",
    "                    init='k-means++', \n",
    "                    n_init=10)\n",
    "clusters = kMInstance.fit_predict(X2)\n",
    "\n",
    "### Distribution\n",
    "\n",
    "labels, counts = np.unique(clusters, return_counts=True)\n",
    "plt.bar(labels, counts, align='center')\n",
    "plt.gca().set_xticks(labels)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "htmlReport = '''<!DOCTYPE html>\n",
    "<html lang=\"en\">\n",
    "<head>\n",
    "<meta charset=\"UTF-8\" />\n",
    "<title>Embedded SVG</title>\n",
    "</head>\n",
    "<body>\n",
    "'''\n",
    "htmlReport += '''<h1>{}</h1>'''.format(name)\n",
    "for cluster_number in np.unique(clusters):\n",
    "    \n",
    "    htmlReport += '\\n<h2>Cluster #{}</h2>'.format(cluster_number)\n",
    "    centroid_of_current_cluster = kMInstance.cluster_centers_[cluster_number]\n",
    "    list_of_the_most_similar_vectors = model[1].model.docvecs.most_similar([centroid_of_current_cluster], topn=10)\n",
    "    \n",
    "    for one_of_similar_vector in list_of_the_most_similar_vectors:\n",
    "        name = one_of_similar_vector[0]\n",
    "        position = int(name.split('_')[-1])\n",
    "        if clusters[position] != cluster_number:\n",
    "            continue\n",
    "        sent = graphs_and_sentences[position]['sentence']\n",
    "        htmlReport += '<p>Sentence (position = {}): {}</p>'.format(position, sent)\n",
    "        doc = ru_nlp(sent)\n",
    "        svg = displacy.render(doc, style='dep', options={'collapse_punct' : False, \n",
    "                                                         #'compact' : True\n",
    "                                                        })\n",
    "\n",
    "        htmlReport += svg\n",
    "        htmlReport += \"<hr>\"\n",
    "\n",
    "htmlReport += \"\"\"</body>\n",
    "</html>\"\"\"\n",
    "\n",
    "htmlReportFile = open(\n",
    "    os.path.join('Reports', html_name.format(OPTIMAL_CLUSTER_NUMBER, current_language)),\n",
    "    'w', encoding='utf-8')\n",
    "htmlReportFile.write(htmlReport)\n",
    "htmlReportFile.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Contracted Nodes Kernel without dependencies\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(<dep_tree_embedding.ContractedFeatureExtractor object at 0x000001BC34E67208>, <dep_tree_embedding.DepTreeEmbedding object at 0x000001BC136FC128>)\n",
      "graph2vec shape: (61889, 51)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Type</th>\n",
       "      <th>x_0</th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>x_5</th>\n",
       "      <th>x_6</th>\n",
       "      <th>x_7</th>\n",
       "      <th>x_8</th>\n",
       "      <th>...</th>\n",
       "      <th>x_40</th>\n",
       "      <th>x_41</th>\n",
       "      <th>x_42</th>\n",
       "      <th>x_43</th>\n",
       "      <th>x_44</th>\n",
       "      <th>x_45</th>\n",
       "      <th>x_46</th>\n",
       "      <th>x_47</th>\n",
       "      <th>x_48</th>\n",
       "      <th>x_49</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.080374</td>\n",
       "      <td>0.520760</td>\n",
       "      <td>0.309990</td>\n",
       "      <td>0.039027</td>\n",
       "      <td>-0.527354</td>\n",
       "      <td>0.159694</td>\n",
       "      <td>-0.266302</td>\n",
       "      <td>0.250471</td>\n",
       "      <td>-0.305119</td>\n",
       "      <td>...</td>\n",
       "      <td>0.082623</td>\n",
       "      <td>0.224690</td>\n",
       "      <td>-0.270106</td>\n",
       "      <td>0.071185</td>\n",
       "      <td>-0.008845</td>\n",
       "      <td>0.278201</td>\n",
       "      <td>-0.458851</td>\n",
       "      <td>-0.325947</td>\n",
       "      <td>-0.579898</td>\n",
       "      <td>-0.056903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.242846</td>\n",
       "      <td>-0.088717</td>\n",
       "      <td>0.197954</td>\n",
       "      <td>-0.044248</td>\n",
       "      <td>-0.234970</td>\n",
       "      <td>0.022015</td>\n",
       "      <td>-0.100390</td>\n",
       "      <td>0.072054</td>\n",
       "      <td>0.022311</td>\n",
       "      <td>...</td>\n",
       "      <td>0.005621</td>\n",
       "      <td>-0.111998</td>\n",
       "      <td>0.073503</td>\n",
       "      <td>-0.280996</td>\n",
       "      <td>0.453760</td>\n",
       "      <td>0.151140</td>\n",
       "      <td>-0.370162</td>\n",
       "      <td>-0.120951</td>\n",
       "      <td>-0.488342</td>\n",
       "      <td>0.229263</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.558147</td>\n",
       "      <td>-0.052629</td>\n",
       "      <td>0.694048</td>\n",
       "      <td>0.139801</td>\n",
       "      <td>-0.031083</td>\n",
       "      <td>-0.244531</td>\n",
       "      <td>-0.360659</td>\n",
       "      <td>-0.096737</td>\n",
       "      <td>0.161229</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.488476</td>\n",
       "      <td>-0.137342</td>\n",
       "      <td>-0.110738</td>\n",
       "      <td>-0.008219</td>\n",
       "      <td>-0.155078</td>\n",
       "      <td>0.273542</td>\n",
       "      <td>-0.833630</td>\n",
       "      <td>0.544772</td>\n",
       "      <td>0.161784</td>\n",
       "      <td>0.145499</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.085733</td>\n",
       "      <td>0.043934</td>\n",
       "      <td>-0.138394</td>\n",
       "      <td>0.317128</td>\n",
       "      <td>-0.305726</td>\n",
       "      <td>-0.199068</td>\n",
       "      <td>-0.537115</td>\n",
       "      <td>0.433412</td>\n",
       "      <td>0.282479</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.345268</td>\n",
       "      <td>-0.008013</td>\n",
       "      <td>0.187837</td>\n",
       "      <td>-0.426704</td>\n",
       "      <td>0.301055</td>\n",
       "      <td>0.207015</td>\n",
       "      <td>-0.480886</td>\n",
       "      <td>-0.224751</td>\n",
       "      <td>0.120544</td>\n",
       "      <td>0.642232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.037600</td>\n",
       "      <td>0.597004</td>\n",
       "      <td>0.147141</td>\n",
       "      <td>0.090706</td>\n",
       "      <td>-0.056560</td>\n",
       "      <td>-0.082539</td>\n",
       "      <td>0.178335</td>\n",
       "      <td>0.306295</td>\n",
       "      <td>0.321238</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.258225</td>\n",
       "      <td>-0.049790</td>\n",
       "      <td>0.308568</td>\n",
       "      <td>0.154916</td>\n",
       "      <td>0.634599</td>\n",
       "      <td>0.150529</td>\n",
       "      <td>-1.033028</td>\n",
       "      <td>-0.297857</td>\n",
       "      <td>0.186070</td>\n",
       "      <td>0.281345</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 51 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Type       x_0       x_1       x_2       x_3       x_4       x_5       x_6  \\\n",
       "0     0 -0.080374  0.520760  0.309990  0.039027 -0.527354  0.159694 -0.266302   \n",
       "1     1  0.242846 -0.088717  0.197954 -0.044248 -0.234970  0.022015 -0.100390   \n",
       "2     2  0.558147 -0.052629  0.694048  0.139801 -0.031083 -0.244531 -0.360659   \n",
       "3     3  0.085733  0.043934 -0.138394  0.317128 -0.305726 -0.199068 -0.537115   \n",
       "4     4  0.037600  0.597004  0.147141  0.090706 -0.056560 -0.082539  0.178335   \n",
       "\n",
       "        x_7       x_8  ...      x_40      x_41      x_42      x_43      x_44  \\\n",
       "0  0.250471 -0.305119  ...  0.082623  0.224690 -0.270106  0.071185 -0.008845   \n",
       "1  0.072054  0.022311  ...  0.005621 -0.111998  0.073503 -0.280996  0.453760   \n",
       "2 -0.096737  0.161229  ... -0.488476 -0.137342 -0.110738 -0.008219 -0.155078   \n",
       "3  0.433412  0.282479  ... -0.345268 -0.008013  0.187837 -0.426704  0.301055   \n",
       "4  0.306295  0.321238  ... -0.258225 -0.049790  0.308568  0.154916  0.634599   \n",
       "\n",
       "       x_45      x_46      x_47      x_48      x_49  \n",
       "0  0.278201 -0.458851 -0.325947 -0.579898 -0.056903  \n",
       "1  0.151140 -0.370162 -0.120951 -0.488342  0.229263  \n",
       "2  0.273542 -0.833630  0.544772  0.161784  0.145499  \n",
       "3  0.207015 -0.480886 -0.224751  0.120544  0.642232  \n",
       "4  0.150529 -1.033028 -0.297857  0.186070  0.281345  \n",
       "\n",
       "[5 rows x 51 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of X : (61889, 50)\n"
     ]
    }
   ],
   "source": [
    "# WL Kernel without dependencies\n",
    "\n",
    "html_name = 'Contracted_Kernel_without_dependencies_{}_clusters_{}.html'\n",
    "name = 'Contracted Kernel without dependencies'\n",
    "\n",
    "path_to_d2v_model = 'FULL_contracted_without_dependency_ru.model'\n",
    "path_to_csv_with_embeddings = 'C://Users//User//AuthorStyle//CoNLL//EXPERIMENTAL_EMBEDDING/contracted_without_dependency_vectors_dim_50_ru.csv'\n",
    "\n",
    "\n",
    "model = pickle.load(open(path_to_d2v_model, 'rb'))\n",
    "print(str(model))\n",
    "df = pd.read_csv(path_to_csv_with_embeddings)\n",
    "print('graph2vec shape: {}'.format(df.shape))\n",
    "display(df.head())\n",
    "\n",
    "\n",
    "\n",
    "numbers_type = df.Type.values\n",
    "X = df.drop('Type', axis=1).values\n",
    "\n",
    "print('Shape of X : {}'.format(X.shape))\n",
    "X2 = normalize(X)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Silhouette Score\n",
      "Silhouette Coefficient for 2 clusters: 0.0547618455675718\n",
      "Silhouette Coefficient for 3 clusters: 0.04623299106340653\n",
      "Silhouette Coefficient for 4 clusters: 0.04518981243117862\n",
      "Silhouette Coefficient for 5 clusters: 0.04097981854382721\n",
      "Silhouette Coefficient for 6 clusters: 0.03435878630386645\n",
      "Silhouette Coefficient for 7 clusters: 0.02864221193868285\n",
      "Silhouette Coefficient for 8 clusters: 0.027440168969618043\n",
      "Silhouette Coefficient for 9 clusters: 0.025621676358642695\n",
      "Silhouette Coefficient for 10 clusters: 0.024671333433304774\n",
      "Silhouette Coefficient for 11 clusters: 0.02541947965578208\n",
      "Silhouette Coefficient for 12 clusters: 0.026763437980144245\n",
      "Silhouette Coefficient for 13 clusters: 0.024368436139619438\n",
      "Silhouette Coefficient for 14 clusters: 0.02376983211744339\n",
      "Silhouette Coefficient for 15 clusters: 0.023353437589384473\n",
      "Silhouette Coefficient for 16 clusters: 0.022493728423715775\n",
      "Silhouette Coefficient for 17 clusters: 0.0188517285842084\n",
      "Silhouette Coefficient for 18 clusters: 0.021103028737100453\n",
      "Silhouette Coefficient for 19 clusters: 0.020956190512704285\n",
      "Silhouette Coefficient for 20 clusters: 0.01938113478323767\n",
      "Silhouette Coefficient for 21 clusters: 0.019997987429012692\n",
      "Silhouette Coefficient for 22 clusters: 0.01908815377545879\n",
      "Silhouette Coefficient for 23 clusters: 0.021146549176268046\n",
      "Silhouette Coefficient for 24 clusters: 0.01719029346701622\n",
      "Silhouette Coefficient for 25 clusters: 0.018135651951469545\n",
      "Silhouette Coefficient for 26 clusters: 0.019068540196781114\n",
      "Silhouette Coefficient for 27 clusters: 0.01848478099818752\n",
      "Silhouette Coefficient for 28 clusters: 0.018559894745030402\n",
      "Silhouette Coefficient for 29 clusters: 0.01868951874509658\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Silhouette Score\")\n",
    "silhouette_scores_list = []\n",
    "\n",
    "for cluster_number in range(2,30):\n",
    "    kMInstance = KMeans(n_clusters=cluster_number, init='k-means++', n_init=5)\n",
    "    kMInstance.fit(X2)\n",
    "    current_silhouette_score = silhouette_score(X2, kMInstance.labels_, metric='euclidean', sample_size=10000)\n",
    "    silhouette_scores_list.append((cluster_number, current_silhouette_score))\n",
    "    print(\"Silhouette Coefficient for {} clusters: {}\".format(cluster_number, current_silhouette_score))\n",
    "\n",
    "plt.plot(list(map(lambda x : x[0], silhouette_scores_list)),\n",
    "       list(map(lambda x : x[1], silhouette_scores_list)))\n",
    "\n",
    "plt.xlabel('Cluster Number')\n",
    "plt.ylabel('Silhouette Score')\n",
    "plt.title('Silhouette Score Analysis')\n",
    "plt.grid(True)\n",
    "plt.show()\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "OPTIMAL_CLUSTER_NUMBER = 23\n",
    "\n",
    "kMInstance = KMeans(n_clusters=OPTIMAL_CLUSTER_NUMBER,\n",
    "                    init='k-means++', \n",
    "                    n_init=10)\n",
    "clusters = kMInstance.fit_predict(X2)\n",
    "\n",
    "### Distribution\n",
    "\n",
    "labels, counts = np.unique(clusters, return_counts=True)\n",
    "plt.bar(labels, counts, align='center')\n",
    "plt.gca().set_xticks(labels)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [],
   "source": [
    "htmlReport = '''<!DOCTYPE html>\n",
    "<html lang=\"en\">\n",
    "<head>\n",
    "<meta charset=\"UTF-8\" />\n",
    "<title>Embedded SVG</title>\n",
    "</head>\n",
    "<body>\n",
    "'''\n",
    "htmlReport += '''<h1>{}</h1>'''.format(name)\n",
    "for cluster_number in np.unique(clusters):\n",
    "    \n",
    "    htmlReport += '\\n<h2>Cluster #{}</h2>'.format(cluster_number)\n",
    "    centroid_of_current_cluster = kMInstance.cluster_centers_[cluster_number]\n",
    "    list_of_the_most_similar_vectors = model[1].model.docvecs.most_similar([centroid_of_current_cluster], topn=10)\n",
    "    \n",
    "    for one_of_similar_vector in list_of_the_most_similar_vectors:\n",
    "        name = one_of_similar_vector[0]\n",
    "        position = int(name.split('_')[-1])\n",
    "        if clusters[position] != cluster_number:\n",
    "            continue\n",
    "        sent = graphs_and_sentences[position]['sentence']\n",
    "        htmlReport += '<p>Sentence (position = {}): {}</p>'.format(position, sent)\n",
    "        doc = ru_nlp(sent)\n",
    "        svg = displacy.render(doc, style='dep')\n",
    "        htmlReport += svg\n",
    "        htmlReport += \"<hr>\"\n",
    "\n",
    "htmlReport += \"\"\"</body>\n",
    "</html>\"\"\"\n",
    "\n",
    "htmlReportFile = open(\n",
    "    os.path.join('Reports', html_name.format(OPTIMAL_CLUSTER_NUMBER, current_language)),\n",
    "    'w', encoding='utf-8')\n",
    "htmlReportFile.write(htmlReport)\n",
    "htmlReportFile.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.vector_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}