Attempt - Antonym Relation Establishment.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Antonym Word Relation Establishment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from gensim.models import Word2Vec\n",
    "from gensim.models import KeyedVectors\n",
    "import gensim.downloader\n",
    "#Import numpy for array calculation|\n",
    "import numpy as np\n",
    "\n",
    "import re\n",
    "\n",
    "#import nltk for string operations\n",
    "from nltk.stem import WordNetLemmatizer \n",
    "from nltk.corpus import stopwords \n",
    "from nltk.tokenize import word_tokenize\n",
    "\n",
    "import relations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import word list\n",
    "imported_tuples = relations.read_relations(\"./relations/antonyms.txt\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the model\n",
    "model = gensim.downloader.load('glove-wiki-gigaword-50')\n",
    "dims = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "# preprocess data\n",
    "antonym_tuples = []\n",
    "for tup in imported_tuples:\n",
    "    if \" \" not in tup[0].strip() and \" \" not in tup[1].strip():\n",
    "        antonym_tuples.append((tup[0].strip().lower(), tup[1].strip().lower()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "# conversion to vectors\n",
    "antonym_vectors = []\n",
    "for tup in antonym_tuples:\n",
    "    if tup[0] in model.vocab and tup[1] in model.vocab:\n",
    "        antonym_vectors.append((model.get_vector(tup[0]), model.get_vector(tup[1])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multi-Variate Linear Regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Given the input vector, the aim is to create a model that outputs its antonym vector. \n",
    "The output variable has multiple dimensions, the strategy is to regress over each dimension and get coefficients for each dimension."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LinearRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model score:  0.308727771650265\n"
     ]
    }
   ],
   "source": [
    "# create a coefficients array\n",
    "coeff = []\n",
    "intercepts = []\n",
    "avg_score = 0\n",
    "# regressing a linear model on each independent dimension of response variable\n",
    "for i in range(dims):\n",
    "    y = [x[0][i] for x in antonym_vectors]\n",
    "    x = [tup[1] for tup in antonym_vectors]\n",
    "    regr = LinearRegression()\n",
    "    regr.fit(x, y)\n",
    "    coeff.append(regr.coef_)\n",
    "    intercepts.append(regr.intercept_)\n",
    "    avg_score += regr.score(x,y)\n",
    "print(\"Model score: \", avg_score/dims)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model has a very low accuracy, lower than it had when the data set was smaller. An explanation could be that words are used in more than one context unlike country-capital vectors and hence don't have a simple relationship."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Relation Establishment using Neural Nets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "# importing the required libs\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "import keras\n",
    "\n",
    "import math\n",
    "from scipy import spatial"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data preparation\n",
    "X = [x[0] for x in antonym_vectors]\n",
    "y = [x[1] for x in antonym_vectors]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "nn = Sequential()\n",
    "nn.add(Dense(12, input_dim=dims, activation='sigmoid'))\n",
    "nn.add(Dense(12, activation='relu'))\n",
    "nn.add(Dense(12, activation='relu'))\n",
    "nn.add(Dense(12, activation='sigmoid'))\n",
    "nn.add(Dense(dims, activation='linear'))\n",
    "nn.compile(loss='mean_squared_error', optimizer='sgd', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "1749/1749 [==============================] - 0s 108us/step - loss: 0.9157 - accuracy: 0.0154\n",
      "Epoch 2/15\n",
      "1749/1749 [==============================] - 0s 61us/step - loss: 0.6898 - accuracy: 0.2093\n",
      "Epoch 3/15\n",
      "1749/1749 [==============================] - 0s 58us/step - loss: 0.5663 - accuracy: 0.3928\n",
      "Epoch 4/15\n",
      "1749/1749 [==============================] - 0s 58us/step - loss: 0.4974 - accuracy: 0.3928\n",
      "Epoch 5/15\n",
      "1749/1749 [==============================] - 0s 59us/step - loss: 0.4592 - accuracy: 0.3928\n",
      "Epoch 6/15\n",
      "1749/1749 [==============================] - 0s 62us/step - loss: 0.4381 - accuracy: 0.3928\n",
      "Epoch 7/15\n",
      "1749/1749 [==============================] - 0s 66us/step - loss: 0.4266 - accuracy: 0.3928\n",
      "Epoch 8/15\n",
      "1749/1749 [==============================] - 0s 64us/step - loss: 0.4204 - accuracy: 0.3928\n",
      "Epoch 9/15\n",
      "1749/1749 [==============================] - 0s 61us/step - loss: 0.4171 - accuracy: 0.3928\n",
      "Epoch 10/15\n",
      "1749/1749 [==============================] - 0s 64us/step - loss: 0.4153 - accuracy: 0.3928\n",
      "Epoch 11/15\n",
      "1749/1749 [==============================] - 0s 70us/step - loss: 0.4144 - accuracy: 0.3928\n",
      "Epoch 12/15\n",
      "1749/1749 [==============================] - 0s 75us/step - loss: 0.4139 - accuracy: 0.3928\n",
      "Epoch 13/15\n",
      "1749/1749 [==============================] - 0s 63us/step - loss: 0.4136 - accuracy: 0.3928\n",
      "Epoch 14/15\n",
      "1749/1749 [==============================] - ETA: 0s - loss: 0.4161 - accuracy: 0.38 - 0s 67us/step - loss: 0.4134 - accuracy: 0.3928\n",
      "Epoch 15/15\n",
      "1749/1749 [==============================] - 0s 69us/step - loss: 0.4133 - accuracy: 0.3928\n"
     ]
    }
   ],
   "source": [
    "# fit the keras model on the dataset\n",
    "history = nn.fit(np.array(X), np.array(y), epochs=15, batch_size=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epoch')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_val = nn.predict(np.array([model.get_vector(\"lazy\")]))\n",
    "actual_val = model.get_vector(\"active\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "# cosine similarity functions\n",
    "def calculate_cosine_distance(a, b):\n",
    "    cosine_distance = float(spatial.distance.cosine(a, b))\n",
    "    return cosine_distance\n",
    "\n",
    "def calculate_cosine_similarity(a, b):\n",
    "    cosine_similarity = 1 - calculate_cosine_distance(a, b)\n",
    "    return cosine_similarity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7912224531173706"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calculate_cosine_similarity(pred_val, actual_val)"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}