Gensim_Word2Vec_BM_LSTM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from gensim.models import Word2Vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Word2Vec.load(\"word2vecBM.model\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of word vectors: 24132\n"
     ]
    }
   ],
   "source": [
    "word_vectors = model.wv\n",
    "print(\"Number of word vectors: {}\".format(len(word_vectors.vocab)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "MAX_NB_WORDS = len(word_vectors.vocab)\n",
    "MAX_SEQUENCE_LENGTH = 1000\n",
    "EMBEDDING_DIM = 100\n",
    "VALIDATION_SPLIT = 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import pickle\n",
    "from collections import defaultdict\n",
    "import re\n",
    "from bs4 import BeautifulSoup\n",
    "import sys\n",
    "import os\n",
    "#os.environ['KERAS_BACKEND']='theano'\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from keras.utils.np_utils import to_categorical\n",
    "from keras.layers import Embedding\n",
    "from keras.layers import Dense, Input, Flatten\n",
    "from keras.layers import Conv1D, MaxPooling1D, Embedding, Dropout, LSTM, GRU, Bidirectional\n",
    "from keras.models import Model\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "import matplotlib.pyplot as plt\n",
    "plt.switch_backend('agg')\n",
    "from keras import backend as K\n",
    "from keras.engine.topology import Layer, InputSpec\n",
    "from keras import initializers\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_str(string):\n",
    "    string = re.sub(r\"\\\\\", \"\", string)\n",
    "    string = re.sub(r\"\\'\", \"\", string)\n",
    "    string = re.sub(r\"\\\"\", \"\", string)\n",
    "    return string.strip().lower()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of dataset  (236, 2)\n",
      "Index(['SENTENCE', 'CATEGORY'], dtype='object')\n",
      "No. of unique classes 6\n",
      "No. of tokens 4729\n"
     ]
    }
   ],
   "source": [
    "# reading data\n",
    "df = pd.read_excel('wsd_data.xlsx')\n",
    "df = df.dropna()\n",
    "df = df.reset_index(drop=True)\n",
    "print('Shape of dataset ',df.shape)\n",
    "print(df.columns)\n",
    "print('No. of unique classes',len(set(df['CATEGORY'])))\n",
    "print('No. of tokens',df['SENTENCE'].apply(lambda x: len(x.split(' '))).sum())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "macronum=sorted(set(df['CATEGORY']))\n",
    "macro_to_id = dict((note, number) for number, note in enumerate(macronum))\n",
    "\n",
    "def fun(i):\n",
    "    return macro_to_id[i]\n",
    "\n",
    "df['class']=df['CATEGORY'].apply(fun)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Home\\Anaconda3\\lib\\site-packages\\bs4\\__init__.py:181: UserWarning: No parser was explicitly specified, so I'm using the best available HTML parser for this system (\"lxml\"). This usually isn't a problem, but if you run this code on another system, or in a different virtual environment, it may use a different parser and behave differently.\n",
      "\n",
      "The code that caused this warning is on line 193 of the file C:\\Users\\Home\\Anaconda3\\lib\\runpy.py. To get rid of this warning, change code that looks like this:\n",
      "\n",
      " BeautifulSoup(YOUR_MARKUP})\n",
      "\n",
      "to this:\n",
      "\n",
      " BeautifulSoup(YOUR_MARKUP, \"lxml\")\n",
      "\n",
      "  markup_type=markup_type))\n"
     ]
    }
   ],
   "source": [
    "texts = []\n",
    "labels = []\n",
    "\n",
    "for idx in range(df.SENTENCE.shape[0]):\n",
    "    text = BeautifulSoup(df.SENTENCE[idx])\n",
    "    texts.append(clean_str(str(text.get_text().encode())))\n",
    "\n",
    "for idx in df['class']:\n",
    "    labels.append(idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of Unique Tokens 2061\n"
     ]
    }
   ],
   "source": [
    "tokenizer = Tokenizer(num_words=MAX_NB_WORDS)\n",
    "tokenizer.fit_on_texts(texts)\n",
    "sequences = tokenizer.texts_to_sequences(texts)\n",
    "\n",
    "word_index = tokenizer.word_index\n",
    "\n",
    "print('Number of Unique Tokens',len(word_index))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of Data Tensor: (236, 1000)\n",
      "Shape of Label Tensor: (236, 6)\n"
     ]
    }
   ],
   "source": [
    "data = pad_sequences(sequences, maxlen=MAX_SEQUENCE_LENGTH)\n",
    "\n",
    "labels = to_categorical(np.asarray(labels))\n",
    "print('Shape of Data Tensor:', data.shape)\n",
    "print('Shape of Label Tensor:', labels.shape)\n",
    "\n",
    "indices = np.arange(data.shape[0])\n",
    "np.random.shuffle(indices)\n",
    "data = data[indices]\n",
    "labels = labels[indices]\n",
    "nb_validation_samples = int(VALIDATION_SPLIT * data.shape[0])\n",
    "\n",
    "x_train = data[:-nb_validation_samples]\n",
    "y_train = labels[:-nb_validation_samples]\n",
    "x_val = data[-nb_validation_samples:]\n",
    "y_val = labels[-nb_validation_samples:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "nb_words = min(MAX_NB_WORDS, len(word_vectors.vocab))\n",
    "# we initialize the matrix with random numbers\n",
    "wv_matrix = (np.random.rand(nb_words, EMBEDDING_DIM) - 0.5) / 5.0\n",
    "for word, i in word_index.items():\n",
    "    if i >= MAX_NB_WORDS:\n",
    "        continue\n",
    "    try:\n",
    "        embedding_vector = word_vectors[word]\n",
    "        # words not found in embedding index will be all-zeros.\n",
    "        wv_matrix[i] = embedding_vector\n",
    "    except:\n",
    "        pass        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_layer = Embedding(nb_words,\n",
    "                            EMBEDDING_DIM,\n",
    "                            weights=[wv_matrix],\n",
    "                            input_length=MAX_SEQUENCE_LENGTH,\n",
    "                            trainable=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bidirectional LSTM\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 1000)              0         \n",
      "_________________________________________________________________\n",
      "embedding_1 (Embedding)      (None, 1000, 100)         2413200   \n",
      "_________________________________________________________________\n",
      "bidirectional_1 (Bidirection (None, 200)               160800    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 6)                 1206      \n",
      "=================================================================\n",
      "Total params: 2,575,206\n",
      "Trainable params: 2,575,206\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
    "embedded_sequences = embedding_layer(sequence_input)\n",
    "l_lstm = Bidirectional(LSTM(100))(embedded_sequences)\n",
    "preds = Dense(len(macronum), activation='softmax')(l_lstm)\n",
    "model = Model(sequence_input, preds)\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='rmsprop',\n",
    "              metrics=['acc'])\n",
    "\n",
    "print(\"Bidirectional LSTM\")\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 189 samples, validate on 47 samples\n",
      "Epoch 1/15\n",
      "189/189 [==============================] - 120s 637ms/step - loss: 1.6389 - acc: 0.3439 - val_loss: 1.6100 - val_acc: 0.3404\n",
      "\n",
      "Epoch 00001: val_acc improved from -inf to 0.34043, saving model to model_lstm.hdf5\n",
      "Epoch 2/15\n",
      "189/189 [==============================] - 134s 707ms/step - loss: 1.3095 - acc: 0.5714 - val_loss: 1.3484 - val_acc: 0.4255\n",
      "\n",
      "Epoch 00002: val_acc improved from 0.34043 to 0.42553, saving model to model_lstm.hdf5\n",
      "Epoch 3/15\n",
      "189/189 [==============================] - 127s 672ms/step - loss: 0.8346 - acc: 0.7884 - val_loss: 1.1805 - val_acc: 0.6383\n",
      "\n",
      "Epoch 00003: val_acc improved from 0.42553 to 0.63830, saving model to model_lstm.hdf5\n",
      "Epoch 4/15\n",
      "189/189 [==============================] - 130s 689ms/step - loss: 0.5680 - acc: 0.8413 - val_loss: 1.1979 - val_acc: 0.6170\n",
      "\n",
      "Epoch 00004: val_acc did not improve from 0.63830\n",
      "Epoch 5/15\n",
      "189/189 [==============================] - 128s 676ms/step - loss: 0.3730 - acc: 0.8889 - val_loss: 1.3677 - val_acc: 0.6596\n",
      "\n",
      "Epoch 00005: val_acc improved from 0.63830 to 0.65957, saving model to model_lstm.hdf5\n",
      "Epoch 6/15\n",
      "189/189 [==============================] - 129s 681ms/step - loss: 0.2794 - acc: 0.9259 - val_loss: 1.2328 - val_acc: 0.6809\n",
      "\n",
      "Epoch 00006: val_acc improved from 0.65957 to 0.68085, saving model to model_lstm.hdf5\n",
      "Epoch 7/15\n",
      "189/189 [==============================] - 130s 688ms/step - loss: 0.1467 - acc: 0.9630 - val_loss: 1.0482 - val_acc: 0.6809\n",
      "\n",
      "Epoch 00007: val_acc did not improve from 0.68085\n",
      "Epoch 8/15\n",
      "189/189 [==============================] - 130s 690ms/step - loss: 0.1224 - acc: 0.9683 - val_loss: 1.3096 - val_acc: 0.6809\n",
      "\n",
      "Epoch 00008: val_acc did not improve from 0.68085\n",
      "Epoch 9/15\n",
      "189/189 [==============================] - 132s 697ms/step - loss: 0.1402 - acc: 0.9630 - val_loss: 1.1379 - val_acc: 0.7660\n",
      "\n",
      "Epoch 00009: val_acc improved from 0.68085 to 0.76596, saving model to model_lstm.hdf5\n",
      "Epoch 10/15\n",
      "189/189 [==============================] - 144s 764ms/step - loss: 0.0382 - acc: 0.9841 - val_loss: 1.1620 - val_acc: 0.7021\n",
      "\n",
      "Epoch 00010: val_acc did not improve from 0.76596\n",
      "Epoch 11/15\n",
      "189/189 [==============================] - 134s 710ms/step - loss: 0.0733 - acc: 0.9735 - val_loss: 1.0598 - val_acc: 0.6809\n",
      "\n",
      "Epoch 00011: val_acc did not improve from 0.76596\n",
      "Epoch 12/15\n",
      "189/189 [==============================] - 139s 736ms/step - loss: 0.0701 - acc: 0.9894 - val_loss: 1.3927 - val_acc: 0.7447\n",
      "\n",
      "Epoch 00012: val_acc did not improve from 0.76596\n",
      "Epoch 13/15\n",
      "189/189 [==============================] - 133s 704ms/step - loss: 0.0102 - acc: 1.0000 - val_loss: 2.6547 - val_acc: 0.6383\n",
      "\n",
      "Epoch 00013: val_acc did not improve from 0.76596\n",
      "Epoch 14/15\n",
      "189/189 [==============================] - 133s 702ms/step - loss: 0.0050 - acc: 1.0000 - val_loss: 1.5707 - val_acc: 0.7021\n",
      "\n",
      "Epoch 00014: val_acc did not improve from 0.76596\n",
      "Epoch 15/15\n",
      "189/189 [==============================] - 135s 713ms/step - loss: 0.0322 - acc: 0.9894 - val_loss: 1.4650 - val_acc: 0.7660\n",
      "\n",
      "Epoch 00015: val_acc did not improve from 0.76596\n"
     ]
    }
   ],
   "source": [
    "cp=ModelCheckpoint('model_lstm.hdf5',monitor='val_acc',verbose=1,save_best_only=True)\n",
    "history=model.fit(x_train, y_train, validation_data=(x_val, y_val),epochs=15, batch_size=2, callbacks=[cp])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig1 = plt.figure()\n",
    "plt.plot(history.history['loss'],'r',linewidth=3.0)\n",
    "plt.plot(history.history['val_loss'],'b',linewidth=3.0)\n",
    "plt.legend(['Training loss', 'Validation Loss'],fontsize=18)\n",
    "plt.xlabel('Epochs ',fontsize=16)\n",
    "plt.ylabel('Loss',fontsize=16)\n",
    "plt.title('Loss Curves :RNN',fontsize=16)\n",
    "fig1.savefig('loss_rnn.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig2=plt.figure()\n",
    "plt.plot(history.history['acc'],'r',linewidth=3.0)\n",
    "plt.plot(history.history['val_acc'],'b',linewidth=3.0)\n",
    "plt.legend(['Training Accuracy', 'Validation Accuracy'],fontsize=18)\n",
    "plt.xlabel('Epochs ',fontsize=16)\n",
    "plt.ylabel('Accuracy',fontsize=16)\n",
    "plt.title('Accuracy Curves : RNN',fontsize=16)\n",
    "fig2.savefig('accuracy_rnn.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "47/47 [==============================] - 4s 85ms/step\n",
      "Test accuracy: 0.7659574468085106\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(x_val, y_val,\n",
    "                       batch_size=2, verbose=1)\n",
    "print('Test accuracy:', score[1])"
   ]
  },
  {
   "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
}