|
| 1 | +{ |
| 2 | + "cells": [ |
| 3 | + { |
| 4 | + "cell_type": "code", |
| 5 | + "execution_count": 40, |
| 6 | + "metadata": { |
| 7 | + "collapsed": false |
| 8 | + }, |
| 9 | + "outputs": [], |
| 10 | + "source": [ |
| 11 | + "# things we need for NLP\n", |
| 12 | + "import nltk\n", |
| 13 | + "from nltk.stem.lancaster import LancasterStemmer\n", |
| 14 | + "stemmer = LancasterStemmer()\n", |
| 15 | + "\n", |
| 16 | + "# things we need for Tensorflow\n", |
| 17 | + "import numpy as np\n", |
| 18 | + "import tflearn\n", |
| 19 | + "import tensorflow as tf\n", |
| 20 | + "import random" |
| 21 | + ] |
| 22 | + }, |
| 23 | + { |
| 24 | + "cell_type": "code", |
| 25 | + "execution_count": 41, |
| 26 | + "metadata": { |
| 27 | + "collapsed": false |
| 28 | + }, |
| 29 | + "outputs": [], |
| 30 | + "source": [ |
| 31 | + "# import our chat-bot intents file\n", |
| 32 | + "import json\n", |
| 33 | + "with open('intents_Rx.json') as json_data:\n", |
| 34 | + " intents = json.load(json_data)" |
| 35 | + ] |
| 36 | + }, |
| 37 | + { |
| 38 | + "cell_type": "code", |
| 39 | + "execution_count": 42, |
| 40 | + "metadata": { |
| 41 | + "collapsed": false |
| 42 | + }, |
| 43 | + "outputs": [ |
| 44 | + { |
| 45 | + "name": "stdout", |
| 46 | + "output_type": "stream", |
| 47 | + "text": [ |
| 48 | + "17 documents\n", |
| 49 | + "5 classes ['coupon', 'goodbye', 'greeting', 'med', 'thanks']\n", |
| 50 | + "40 unique stemmed words [\"'m\", \"'s\", 'a', 'anyon', 'ar', 'buy', 'bye', 'can', 'cheap', 'cheapest', 'coupon', 'day', 'deal', 'find', 'for', 'good', 'goodby', 'hello', 'help', 'hi', 'how', 'i', 'is', 'lat', 'less', 'look', 'me', 'med', 'money', 'see', 'send', 'thank', 'that', 'the', 'ther', 'to', 'want', 'what', 'wher', 'you']\n" |
| 51 | + ] |
| 52 | + } |
| 53 | + ], |
| 54 | + "source": [ |
| 55 | + "words = []\n", |
| 56 | + "classes = []\n", |
| 57 | + "documents = []\n", |
| 58 | + "ignore_words = ['?']\n", |
| 59 | + "# loop through each sentence in our intents patterns\n", |
| 60 | + "for intent in intents['intents']:\n", |
| 61 | + " for pattern in intent['patterns']:\n", |
| 62 | + " # tokenize each word in the sentence\n", |
| 63 | + " w = nltk.word_tokenize(pattern)\n", |
| 64 | + " # add to our words list\n", |
| 65 | + " words.extend(w)\n", |
| 66 | + " # add to documents in our corpus\n", |
| 67 | + " documents.append((w, intent['tag']))\n", |
| 68 | + " # add to our classes list\n", |
| 69 | + " if intent['tag'] not in classes:\n", |
| 70 | + " classes.append(intent['tag'])\n", |
| 71 | + "\n", |
| 72 | + "# stem and lower each word and remove duplicates\n", |
| 73 | + "words = [stemmer.stem(w.lower()) for w in words if w not in ignore_words]\n", |
| 74 | + "words = sorted(list(set(words)))\n", |
| 75 | + "\n", |
| 76 | + "# remove duplicates\n", |
| 77 | + "classes = sorted(list(set(classes)))\n", |
| 78 | + "\n", |
| 79 | + "print (len(documents), \"documents\")\n", |
| 80 | + "print (len(classes), \"classes\", classes)\n", |
| 81 | + "print (len(words), \"unique stemmed words\", words)" |
| 82 | + ] |
| 83 | + }, |
| 84 | + { |
| 85 | + "cell_type": "code", |
| 86 | + "execution_count": 43, |
| 87 | + "metadata": { |
| 88 | + "collapsed": false |
| 89 | + }, |
| 90 | + "outputs": [], |
| 91 | + "source": [ |
| 92 | + "# create our training data\n", |
| 93 | + "training = []\n", |
| 94 | + "output = []\n", |
| 95 | + "# create an empty array for our output\n", |
| 96 | + "output_empty = [0] * len(classes)\n", |
| 97 | + "\n", |
| 98 | + "# training set, bag of words for each sentence\n", |
| 99 | + "for doc in documents:\n", |
| 100 | + " # initialize our bag of words\n", |
| 101 | + " bag = []\n", |
| 102 | + " # list of tokenized words for the pattern\n", |
| 103 | + " pattern_words = doc[0]\n", |
| 104 | + " # stem each word\n", |
| 105 | + " pattern_words = [stemmer.stem(word.lower()) for word in pattern_words]\n", |
| 106 | + " # create our bag of words array\n", |
| 107 | + " for w in words:\n", |
| 108 | + " bag.append(1) if w in pattern_words else bag.append(0)\n", |
| 109 | + "\n", |
| 110 | + " # output is a '0' for each tag and '1' for current tag\n", |
| 111 | + " output_row = list(output_empty)\n", |
| 112 | + " output_row[classes.index(doc[1])] = 1\n", |
| 113 | + "\n", |
| 114 | + " training.append([bag, output_row])\n", |
| 115 | + "\n", |
| 116 | + "# shuffle our features and turn into np.array\n", |
| 117 | + "random.shuffle(training)\n", |
| 118 | + "training = np.array(training)\n", |
| 119 | + "\n", |
| 120 | + "# create train and test lists\n", |
| 121 | + "train_x = list(training[:,0])\n", |
| 122 | + "train_y = list(training[:,1])" |
| 123 | + ] |
| 124 | + }, |
| 125 | + { |
| 126 | + "cell_type": "code", |
| 127 | + "execution_count": 44, |
| 128 | + "metadata": { |
| 129 | + "collapsed": false, |
| 130 | + "scrolled": true |
| 131 | + }, |
| 132 | + "outputs": [ |
| 133 | + { |
| 134 | + "name": "stdout", |
| 135 | + "output_type": "stream", |
| 136 | + "text": [ |
| 137 | + "Training Step: 2999 | total loss: \u001b[1m\u001b[32m0.15651\u001b[0m\u001b[0m | time: 0.006s\n", |
| 138 | + "| Adam | epoch: 1000 | loss: 0.15651 - acc: 0.9794 -- iter: 16/17\n", |
| 139 | + "Training Step: 3000 | total loss: \u001b[1m\u001b[32m0.14101\u001b[0m\u001b[0m | time: 0.008s\n", |
| 140 | + "| Adam | epoch: 1000 | loss: 0.14101 - acc: 0.9815 -- iter: 17/17\n", |
| 141 | + "--\n", |
| 142 | + "INFO:tensorflow:/home/gk/gensim/notebooks/Rxmodel.tflearn is not in all_model_checkpoint_paths. Manually adding it.\n" |
| 143 | + ] |
| 144 | + } |
| 145 | + ], |
| 146 | + "source": [ |
| 147 | + "# reset underlying graph data\n", |
| 148 | + "tf.reset_default_graph()\n", |
| 149 | + "# Build neural network\n", |
| 150 | + "net = tflearn.input_data(shape=[None, len(train_x[0])])\n", |
| 151 | + "net = tflearn.fully_connected(net, 8)\n", |
| 152 | + "net = tflearn.fully_connected(net, 8)\n", |
| 153 | + "net = tflearn.fully_connected(net, len(train_y[0]), activation='softmax')\n", |
| 154 | + "net = tflearn.regression(net)\n", |
| 155 | + "\n", |
| 156 | + "# Define model and setup tensorboard\n", |
| 157 | + "model = tflearn.DNN(net, tensorboard_dir='tflearn_logs')\n", |
| 158 | + "# Start training (apply gradient descent algorithm)\n", |
| 159 | + "model.fit(train_x, train_y, n_epoch=1000, batch_size=8, show_metric=True)\n", |
| 160 | + "model.save('Rxmodel.tflearn')" |
| 161 | + ] |
| 162 | + }, |
| 163 | + { |
| 164 | + "cell_type": "code", |
| 165 | + "execution_count": 45, |
| 166 | + "metadata": { |
| 167 | + "collapsed": true |
| 168 | + }, |
| 169 | + "outputs": [], |
| 170 | + "source": [ |
| 171 | + "def clean_up_sentence(sentence):\n", |
| 172 | + " # tokenize the pattern\n", |
| 173 | + " sentence_words = nltk.word_tokenize(sentence)\n", |
| 174 | + " # stem each word\n", |
| 175 | + " sentence_words = [stemmer.stem(word.lower()) for word in sentence_words]\n", |
| 176 | + " return sentence_words\n", |
| 177 | + "\n", |
| 178 | + "# return bag of words array: 0 or 1 for each word in the bag that exists in the sentence\n", |
| 179 | + "def bow(sentence, words, show_details=False):\n", |
| 180 | + " # tokenize the pattern\n", |
| 181 | + " sentence_words = clean_up_sentence(sentence)\n", |
| 182 | + " # bag of words\n", |
| 183 | + " bag = [0]*len(words) \n", |
| 184 | + " for s in sentence_words:\n", |
| 185 | + " for i,w in enumerate(words):\n", |
| 186 | + " if w == s: \n", |
| 187 | + " bag[i] = 1\n", |
| 188 | + " if show_details:\n", |
| 189 | + " print (\"found in bag: %s\" % w)\n", |
| 190 | + "\n", |
| 191 | + " return(np.array(bag))" |
| 192 | + ] |
| 193 | + }, |
| 194 | + { |
| 195 | + "cell_type": "code", |
| 196 | + "execution_count": 46, |
| 197 | + "metadata": { |
| 198 | + "collapsed": false |
| 199 | + }, |
| 200 | + "outputs": [ |
| 201 | + { |
| 202 | + "name": "stdout", |
| 203 | + "output_type": "stream", |
| 204 | + "text": [ |
| 205 | + "[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n", |
| 206 | + " 0 0 0]\n", |
| 207 | + "['coupon', 'goodbye', 'greeting', 'med', 'thanks']\n" |
| 208 | + ] |
| 209 | + } |
| 210 | + ], |
| 211 | + "source": [ |
| 212 | + "p = bow(\"hello\", words)\n", |
| 213 | + "print (p)\n", |
| 214 | + "print (classes)" |
| 215 | + ] |
| 216 | + }, |
| 217 | + { |
| 218 | + "cell_type": "code", |
| 219 | + "execution_count": 47, |
| 220 | + "metadata": { |
| 221 | + "collapsed": false |
| 222 | + }, |
| 223 | + "outputs": [ |
| 224 | + { |
| 225 | + "name": "stdout", |
| 226 | + "output_type": "stream", |
| 227 | + "text": [ |
| 228 | + "[[4.3407872851730644e-09, 0.009914605878293514, 0.9880092740058899, 0.0020757599268108606, 3.3042027780538774e-07]]\n" |
| 229 | + ] |
| 230 | + } |
| 231 | + ], |
| 232 | + "source": [ |
| 233 | + "print(model.predict([p]))" |
| 234 | + ] |
| 235 | + }, |
| 236 | + { |
| 237 | + "cell_type": "code", |
| 238 | + "execution_count": 48, |
| 239 | + "metadata": { |
| 240 | + "collapsed": false |
| 241 | + }, |
| 242 | + "outputs": [], |
| 243 | + "source": [ |
| 244 | + "# save all of our data structures\n", |
| 245 | + "import pickle\n", |
| 246 | + "pickle.dump( {'words':words, 'classes':classes, 'train_x':train_x, 'train_y':train_y}, open( \"Rx_training_data\", \"wb\" ) )" |
| 247 | + ] |
| 248 | + } |
| 249 | + ], |
| 250 | + "metadata": { |
| 251 | + "kernelspec": { |
| 252 | + "display_name": "Python 3", |
| 253 | + "language": "python", |
| 254 | + "name": "python3" |
| 255 | + }, |
| 256 | + "language_info": { |
| 257 | + "codemirror_mode": { |
| 258 | + "name": "ipython", |
| 259 | + "version": 3 |
| 260 | + }, |
| 261 | + "file_extension": ".py", |
| 262 | + "mimetype": "text/x-python", |
| 263 | + "name": "python", |
| 264 | + "nbconvert_exporter": "python", |
| 265 | + "pygments_lexer": "ipython3", |
| 266 | + "version": "3.5.2" |
| 267 | + } |
| 268 | + }, |
| 269 | + "nbformat": 4, |
| 270 | + "nbformat_minor": 1 |
| 271 | +} |
0 commit comments