Simple sentiment analysis with Keras and FastText embeddings.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Simple sentiment analysis with Keras and FastText embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import gensim\n",
    "from gensim.models.wrappers import FastText"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "g = FastText.load_fasttext_format('wiki.en.bin')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "g.most_similar(positive=[\"final_fantasy\"], topn=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import nltk\n",
    "stopwords = set(nltk.corpus.stopwords.words('english'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:2: DeprecationWarning: Call to deprecated `__contains__` (Method will be removed in 4.0.0, use self.wv.__contains__() instead).\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "with open(\"reviews.txt\") as f:\n",
    "    reviews = [ [y.lower() for y in line.strip().split() if y not in stopwords and len(y) > 2 and y in g ] for line in f ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"labels.txt\") as f:\n",
    "    labels = np.array([ 1 if line.strip() == 'positive' else 0 for line in f ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_words = 100\n",
    "embedding_dim = 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "tokenizer = Tokenizer(num_words=50000)\n",
    "texts = [\" \".join(review) for review in reviews]\n",
    "tokenizer.fit_on_texts(texts)\n",
    "sequences = tokenizer.texts_to_sequences(texts)\n",
    "data = pad_sequences(sequences, maxlen=max_words)\n",
    "vocab = tokenizer.word_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:3: DeprecationWarning: Call to deprecated `__contains__` (Method will be removed in 4.0.0, use self.wv.__contains__() instead).\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n",
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:4: DeprecationWarning: Call to deprecated `__getitem__` (Method will be removed in 4.0.0, use self.wv.__getitem__() instead).\n",
      "  after removing the cwd from sys.path.\n"
     ]
    }
   ],
   "source": [
    "glove_emb = np.zeros((len(vocab) + 1, embedding_dim))\n",
    "for word, i in vocab.items():\n",
    "    if word in g:\n",
    "        glove_emb[i] = g[word]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from keras.preprocessing import sequence\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Embedding\n",
    "from keras.layers import LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Embedding(len(vocab) + 1, embedding_dim, weights=[glove_emb], input_length=max_words, trainable=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.add(LSTM(128, dropout=0.3, recurrent_dropout=0.3))\n",
    "model.add(Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, y_train, X_val, y_val = data[:20000], labels[:20000], data[20000:], labels[20000:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/5\n",
      "20000/20000 [==============================] - 111s - loss: 0.5771 - acc: 0.6992 - val_loss: 0.6501 - val_acc: 0.6522\n",
      "Epoch 2/5\n",
      "20000/20000 [==============================] - 111s - loss: 0.5034 - acc: 0.7595 - val_loss: 0.4250 - val_acc: 0.8138\n",
      "Epoch 3/5\n",
      "20000/20000 [==============================] - 112s - loss: 0.4592 - acc: 0.7859 - val_loss: 0.3992 - val_acc: 0.8326\n",
      "Epoch 4/5\n",
      "20000/20000 [==============================] - 112s - loss: 0.4239 - acc: 0.8055 - val_loss: 0.3658 - val_acc: 0.8392\n",
      "Epoch 5/5\n",
      "20000/20000 [==============================] - 113s - loss: 0.3980 - acc: 0.8193 - val_loss: 0.3906 - val_acc: 0.8194\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fb81a509eb8>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train, validation_data=(X_val, y_val), epochs=5, batch_size=128)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.6096567]], dtype=float32)"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s = \"cool movie !\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_on_texts(s):\n",
    "    print(model.predict(pad_sequences(tokenizer.texts_to_sequences(s), maxlen=max_words)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.84083194]\n",
      " [0.7083693 ]\n",
      " [0.18270014]\n",
      " [0.52240455]\n",
      " [0.8464341 ]\n",
      " [0.7111294 ]]\n"
     ]
    }
   ],
   "source": [
    "test_on_texts([\n",
    "    'this was a great movie',\n",
    "    'i really liked it',\n",
    "    'terrible film, hated it',\n",
    "    'the movie was so-so',\n",
    "    'oh yeah, great movie, a new godfather',\n",
    "    'i would watch it again and again and again'\n",
    "    ])"
   ]
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}