Chatbot.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import nltk\n",
    "# nltk.download('punkt')\n",
    "# nltk.download('wordnet')\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.corpus import stopwords\n",
    "lemmatizer = WordNetLemmatizer()\n",
    "import json\n",
    "import pickle\n",
    "\n",
    "import numpy as np\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Activation, Dropout\n",
    "from keras.optimizers import SGD\n",
    "import random\n",
    "\n",
    "words=[]\n",
    "classes = []\n",
    "documents = []\n",
    "ignore_words = ['?', '!', '.',',']\n",
    "stop_words = list(stopwords.words('english'))\n",
    "data_file = open('intents.json').read()\n",
    "intents = json.loads(data_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "for intent in intents['intents']:\n",
    "    for pattern in intent['patterns']:\n",
    "\n",
    "        # take each word and tokenize it\n",
    "        w = nltk.word_tokenize(pattern)\n",
    "        #filter out stop words\n",
    "        words.extend(word for word in w if word not in stop_words)\n",
    "            # adding documents\n",
    "        documents.append((w, intent['tag']))\n",
    "        # adding classes to our class list\n",
    "        if intent['tag'] not in classes:\n",
    "            classes.append(intent['tag'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42 documents\n",
      "18 classes ['breast_cancer', 'calories', 'childhood_earaches', 'cyberbullying', 'day_care', 'eating_plan', 'goodbye', 'greeting', 'hearing_aids', 'kidney_stones', 'options', 'shutup', 'stroke', 'thanks', 'type2_diabetes', 'vegan_diet', 'vegetarian', 'yogurt']\n",
      "99 unique lemmatized words [\"'s\", '2', 'advice', 'aid', 'allergy', 'anyone', 'asthma', 'awesome', 'breast', 'bye', 'calorie', 'can', 'cancer', 'care', 'cause', 'chatting', 'child', 'childhood', 'could', 'cyberbullying', 'day', 'developing', 'diabetes', 'diet', 'do', 'doe', 'drinking', 'earache', 'eat', 'eating', 'effect', 'factor', 'genetic', 'get', 'go', 'good', 'goodbye', 'greek', 'hard', 'healthier', 'hearing', 'hello', 'help', 'helpful', 'helping', 'hey', 'hi', 'hola', 'how', 'i', 'inherited', 'is', 'kidney', 'later', 'lemon', 'life', 'liquor', 'loss', 'lupus', 'making', 'motivated', 'need', 'next', 'nice', 'offered', 'often', 'patient', 'plan', 'preschool', 'prevent', 'protein', 'provide', 'recommend', 'regular', 'remission', 'replace', 'see', 'sense', 'shutup', 'something', 'stay', 'stone', 'stroke', 'support', 'tell', 'thank', 'thanks', 'that', 'till', 'time', 'tool', 'trying', 'type', 'vegan', 'vegetarian', 'water', 'what', 'yogurt', 'you']\n"
     ]
    }
   ],
   "source": [
    "words = [lemmatizer.lemmatize(w.lower()) for w in words if w not in ignore_words]\n",
    "words = sorted(list(set(words)))\n",
    "\n",
    "classes = sorted(list(set(classes)))\n",
    "\n",
    "print (len(documents), \"documents\")\n",
    "\n",
    "print (len(classes), \"classes\", classes)\n",
    "\n",
    "print (len(words), \"unique lemmatized words\", words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "pickle.dump(words,open('words.pkl','wb'))\n",
    "pickle.dump(classes,open('classes.pkl','wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(['Hi', 'there'], 'greeting'),\n",
       " (['How', 'are', 'you'], 'greeting'),\n",
       " (['Is', 'anyone', 'there', '?'], 'greeting'),\n",
       " (['Hey'], 'greeting'),\n",
       " (['Hola'], 'greeting'),\n",
       " (['Hello'], 'greeting'),\n",
       " (['Good', 'day'], 'greeting'),\n",
       " (['Bye'], 'goodbye'),\n",
       " (['See', 'you', 'later'], 'goodbye'),\n",
       " (['Goodbye'], 'goodbye'),\n",
       " (['Nice', 'chatting', 'to', 'you', ',', 'bye'], 'goodbye'),\n",
       " (['Till', 'next', 'time'], 'goodbye'),\n",
       " (['Shutup'], 'shutup'),\n",
       " (['get', 'out'], 'shutup'),\n",
       " (['You', 'are', 'not', 'making', 'sense'], 'shutup'),\n",
       " (['Thanks'], 'thanks'),\n",
       " (['Thank', 'you'], 'thanks'),\n",
       " (['That', \"'s\", 'helpful'], 'thanks'),\n",
       " (['Awesome', ',', 'thanks'], 'thanks'),\n",
       " (['Thanks', 'for', 'helping', 'me'], 'thanks'),\n",
       " (['How', 'you', 'could', 'help', 'me', '?'], 'options'),\n",
       " (['What', 'you', 'can', 'do', '?'], 'options'),\n",
       " (['What', 'help', 'you', 'provide', '?'], 'options'),\n",
       " (['How', 'you', 'can', 'be', 'helpful', '?'], 'options'),\n",
       " (['What', 'support', 'is', 'offered'], 'options'),\n",
       " (['Tell', 'me', 'something', 'about', 'type', '2', 'diabetes'],\n",
       "  'type2_diabetes'),\n",
       " (['Tell', 'me', 'something', 'about', 'calories', 'from', 'hard', 'liquor'],\n",
       "  'calories'),\n",
       " (['Can',\n",
       "   'a',\n",
       "   'vegan',\n",
       "   'diet',\n",
       "   'cause',\n",
       "   'lupus',\n",
       "   'patient',\n",
       "   'to',\n",
       "   'go',\n",
       "   'into',\n",
       "   'remission',\n",
       "   '?'],\n",
       "  'vegan_diet'),\n",
       " (['Is', 'Greek', 'yogurt', 'healthier', 'than', 'regular', 'yogurt', '?'],\n",
       "  'yogurt'),\n",
       " (['Greek', 'yogurt', 'and', 'regular', 'yogurt', '?'], 'yogurt'),\n",
       " (['I',\n",
       "   'am',\n",
       "   'trying',\n",
       "   'to',\n",
       "   'eat',\n",
       "   'a',\n",
       "   'vegetarian',\n",
       "   'diet',\n",
       "   '.',\n",
       "   'How',\n",
       "   'can',\n",
       "   'I',\n",
       "   'get',\n",
       "   'the',\n",
       "   'protein',\n",
       "   'that',\n",
       "   'I',\n",
       "   'need',\n",
       "   '?'],\n",
       "  'vegetarian'),\n",
       " (['how', 'vegetarian', 'get', 'protein', '?'], 'vegetarian'),\n",
       " (['Is', 'breast', 'cancer', 'inherited', '?'], 'breast_cancer'),\n",
       " (['Is', 'there', 'a', 'genetic', 'factor', 'to', 'strokes', '?'], 'stroke'),\n",
       " (['Can',\n",
       "   'you',\n",
       "   'recommend',\n",
       "   'any',\n",
       "   'tools',\n",
       "   'that',\n",
       "   'will',\n",
       "   'help',\n",
       "   'me',\n",
       "   'stay',\n",
       "   'motivated',\n",
       "   'to',\n",
       "   'stay',\n",
       "   'on',\n",
       "   'my',\n",
       "   'eating',\n",
       "   'plan',\n",
       "   '?'],\n",
       "  'eating_plan'),\n",
       " (['What', 'advice', 'for', 'cyberbullying'], 'cyberbullying'),\n",
       " (['Tell', 'me', 'something', 'more', 'about', 'cyberbullying'],\n",
       "  'cyberbullying'),\n",
       " (['How', 'often', 'to', 'replace', 'hearing', 'aids'], 'hearing_aids'),\n",
       " (['Can',\n",
       "   'drinking',\n",
       "   'water',\n",
       "   'with',\n",
       "   'lemon',\n",
       "   'prevent',\n",
       "   'kidney',\n",
       "   'stones',\n",
       "   '?'],\n",
       "  'kidney_stones'),\n",
       " (['Does',\n",
       "   'day',\n",
       "   'care',\n",
       "   'have',\n",
       "   'an',\n",
       "   'effect',\n",
       "   'on',\n",
       "   'children',\n",
       "   'developing',\n",
       "   'allergies',\n",
       "   'or',\n",
       "   'asthma',\n",
       "   '?'],\n",
       "  'day_care'),\n",
       " (['Does',\n",
       "   'preschool',\n",
       "   'have',\n",
       "   'an',\n",
       "   'effect',\n",
       "   'on',\n",
       "   'children',\n",
       "   'developing',\n",
       "   'allergies',\n",
       "   'or',\n",
       "   'asthma',\n",
       "   '?'],\n",
       "  'day_care'),\n",
       " (['Do',\n",
       "   'childhood',\n",
       "   'earaches',\n",
       "   'cause',\n",
       "   'hearing',\n",
       "   'loss',\n",
       "   'later',\n",
       "   'in',\n",
       "   'life',\n",
       "   '?'],\n",
       "  'childhood_earaches')]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "documents"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Hi', 'there']\n",
      "['How', 'are', 'you']\n",
      "['Is', 'anyone', 'there', '?']\n",
      "['Hey']\n",
      "['Hola']\n",
      "['Hello']\n",
      "['Good', 'day']\n",
      "['Bye']\n",
      "['See', 'you', 'later']\n",
      "['Goodbye']\n",
      "['Nice', 'chatting', 'to', 'you', ',', 'bye']\n",
      "['Till', 'next', 'time']\n",
      "['Shutup']\n",
      "['get', 'out']\n",
      "['You', 'are', 'not', 'making', 'sense']\n",
      "['Thanks']\n",
      "['Thank', 'you']\n",
      "['That', \"'s\", 'helpful']\n",
      "['Awesome', ',', 'thanks']\n",
      "['Thanks', 'for', 'helping', 'me']\n",
      "['How', 'you', 'could', 'help', 'me', '?']\n",
      "['What', 'you', 'can', 'do', '?']\n",
      "['What', 'help', 'you', 'provide', '?']\n",
      "['How', 'you', 'can', 'be', 'helpful', '?']\n",
      "['What', 'support', 'is', 'offered']\n",
      "['Tell', 'me', 'something', 'about', 'type', '2', 'diabetes']\n",
      "['Tell', 'me', 'something', 'about', 'calories', 'from', 'hard', 'liquor']\n",
      "['Can', 'a', 'vegan', 'diet', 'cause', 'lupus', 'patient', 'to', 'go', 'into', 'remission', '?']\n",
      "['Is', 'Greek', 'yogurt', 'healthier', 'than', 'regular', 'yogurt', '?']\n",
      "['Greek', 'yogurt', 'and', 'regular', 'yogurt', '?']\n",
      "['I', 'am', 'trying', 'to', 'eat', 'a', 'vegetarian', 'diet', '.', 'How', 'can', 'I', 'get', 'the', 'protein', 'that', 'I', 'need', '?']\n",
      "['how', 'vegetarian', 'get', 'protein', '?']\n",
      "['Is', 'breast', 'cancer', 'inherited', '?']\n",
      "['Is', 'there', 'a', 'genetic', 'factor', 'to', 'strokes', '?']\n",
      "['Can', 'you', 'recommend', 'any', 'tools', 'that', 'will', 'help', 'me', 'stay', 'motivated', 'to', 'stay', 'on', 'my', 'eating', 'plan', '?']\n",
      "['What', 'advice', 'for', 'cyberbullying']\n",
      "['Tell', 'me', 'something', 'more', 'about', 'cyberbullying']\n",
      "['How', 'often', 'to', 'replace', 'hearing', 'aids']\n",
      "['Can', 'drinking', 'water', 'with', 'lemon', 'prevent', 'kidney', 'stones', '?']\n",
      "['Does', 'day', 'care', 'have', 'an', 'effect', 'on', 'children', 'developing', 'allergies', 'or', 'asthma', '?']\n",
      "['Does', 'preschool', 'have', 'an', 'effect', 'on', 'children', 'developing', 'allergies', 'or', 'asthma', '?']\n",
      "['Do', 'childhood', 'earaches', 'cause', 'hearing', 'loss', 'later', 'in', 'life', '?']\n"
     ]
    }
   ],
   "source": [
    "for doc in documents:\n",
    "    print(doc[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['breast_cancer',\n",
       " 'calories',\n",
       " 'childhood_earaches',\n",
       " 'cyberbullying',\n",
       " 'day_care',\n",
       " 'eating_plan',\n",
       " 'goodbye',\n",
       " 'greeting',\n",
       " 'hearing_aids',\n",
       " 'kidney_stones',\n",
       " 'options',\n",
       " 'shutup',\n",
       " 'stroke',\n",
       " 'thanks',\n",
       " 'type2_diabetes',\n",
       " 'vegan_diet',\n",
       " 'vegetarian',\n",
       " 'yogurt']"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "training = []\n",
    "output_empty = [0] * len(classes)\n",
    "for doc in documents:\n",
    "    # initializing bag of words\n",
    "    bag = []\n",
    "    # list of tokenized words for the pattern\n",
    "    pattern_words = doc[0]\n",
    "    # lemmatize each word - create base word, in attempt to represent related words\n",
    "    pattern_words = [lemmatizer.lemmatize(word.lower()) for word in pattern_words]\n",
    "    # create our bag of words array with 1, if word match found in current pattern\n",
    "    for w in words:\n",
    "        bag.append(1) if w in pattern_words else bag.append(0)\n",
    "\n",
    "    # record the corrseponding class for each pattern\n",
    "    output_row = list(output_empty)\n",
    "    output_row[classes.index(doc[1])] = 1\n",
    "\n",
    "    training.append([bag, output_row])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data created\n",
      "Epoch 1/200\n",
      "42/42 [==============================] - 0s 9ms/step - loss: 2.9352 - accuracy: 0.0238\n",
      "Epoch 2/200\n",
      "42/42 [==============================] - 0s 238us/step - loss: 2.7831 - accuracy: 0.1429\n",
      "Epoch 3/200\n",
      "42/42 [==============================] - 0s 252us/step - loss: 2.7737 - accuracy: 0.1667\n",
      "Epoch 4/200\n",
      "42/42 [==============================] - 0s 296us/step - loss: 2.6376 - accuracy: 0.3095\n",
      "Epoch 5/200\n",
      "42/42 [==============================] - 0s 324us/step - loss: 2.5456 - accuracy: 0.2857\n",
      "Epoch 6/200\n",
      "42/42 [==============================] - 0s 289us/step - loss: 2.4927 - accuracy: 0.3333\n",
      "Epoch 7/200\n",
      "42/42 [==============================] - 0s 274us/step - loss: 2.4171 - accuracy: 0.2143\n",
      "Epoch 8/200\n",
      "42/42 [==============================] - 0s 238us/step - loss: 2.2841 - accuracy: 0.3810\n",
      "Epoch 9/200\n",
      "42/42 [==============================] - 0s 251us/step - loss: 2.3092 - accuracy: 0.2619\n",
      "Epoch 10/200\n",
      "42/42 [==============================] - 0s 259us/step - loss: 1.9858 - accuracy: 0.5476\n",
      "Epoch 11/200\n",
      "42/42 [==============================] - 0s 243us/step - loss: 2.2240 - accuracy: 0.2857\n",
      "Epoch 12/200\n",
      "42/42 [==============================] - 0s 230us/step - loss: 2.0089 - accuracy: 0.4524\n",
      "Epoch 13/200\n",
      "42/42 [==============================] - 0s 213us/step - loss: 1.8973 - accuracy: 0.4524\n",
      "Epoch 14/200\n",
      "42/42 [==============================] - 0s 220us/step - loss: 1.7340 - accuracy: 0.5476\n",
      "Epoch 15/200\n",
      "42/42 [==============================] - 0s 269us/step - loss: 1.7558 - accuracy: 0.5238\n",
      "Epoch 16/200\n",
      "42/42 [==============================] - 0s 288us/step - loss: 1.6531 - accuracy: 0.5238\n",
      "Epoch 17/200\n",
      "42/42 [==============================] - 0s 292us/step - loss: 1.4957 - accuracy: 0.5476\n",
      "Epoch 18/200\n",
      "42/42 [==============================] - 0s 258us/step - loss: 1.3469 - accuracy: 0.6429\n",
      "Epoch 19/200\n",
      "42/42 [==============================] - 0s 237us/step - loss: 1.2625 - accuracy: 0.6667\n",
      "Epoch 20/200\n",
      "42/42 [==============================] - 0s 284us/step - loss: 1.3122 - accuracy: 0.5952\n",
      "Epoch 21/200\n",
      "42/42 [==============================] - 0s 303us/step - loss: 1.0736 - accuracy: 0.7143\n",
      "Epoch 22/200\n",
      "42/42 [==============================] - 0s 255us/step - loss: 1.2629 - accuracy: 0.6190\n",
      "Epoch 23/200\n",
      "42/42 [==============================] - 0s 233us/step - loss: 1.1741 - accuracy: 0.6429\n",
      "Epoch 24/200\n",
      "42/42 [==============================] - 0s 238us/step - loss: 0.7934 - accuracy: 0.8095\n",
      "Epoch 25/200\n",
      "42/42 [==============================] - 0s 231us/step - loss: 1.0637 - accuracy: 0.6190\n",
      "Epoch 26/200\n",
      "42/42 [==============================] - 0s 236us/step - loss: 1.0597 - accuracy: 0.6429\n",
      "Epoch 27/200\n",
      "42/42 [==============================] - 0s 234us/step - loss: 0.8440 - accuracy: 0.8095\n",
      "Epoch 28/200\n",
      "42/42 [==============================] - 0s 222us/step - loss: 0.9493 - accuracy: 0.7143\n",
      "Epoch 29/200\n",
      "42/42 [==============================] - 0s 235us/step - loss: 0.8063 - accuracy: 0.7857\n",
      "Epoch 30/200\n",
      "42/42 [==============================] - 0s 236us/step - loss: 0.6325 - accuracy: 0.7857\n",
      "Epoch 31/200\n",
      "42/42 [==============================] - 0s 301us/step - loss: 0.8402 - accuracy: 0.7857\n",
      "Epoch 32/200\n",
      "42/42 [==============================] - 0s 280us/step - loss: 0.6379 - accuracy: 0.8095\n",
      "Epoch 33/200\n",
      "42/42 [==============================] - 0s 272us/step - loss: 0.6159 - accuracy: 0.8095\n",
      "Epoch 34/200\n",
      "42/42 [==============================] - 0s 317us/step - loss: 0.5935 - accuracy: 0.8333\n",
      "Epoch 35/200\n",
      "42/42 [==============================] - 0s 297us/step - loss: 0.6091 - accuracy: 0.9286\n",
      "Epoch 36/200\n",
      "42/42 [==============================] - 0s 247us/step - loss: 0.5450 - accuracy: 0.9286\n",
      "Epoch 37/200\n",
      "42/42 [==============================] - 0s 247us/step - loss: 0.5525 - accuracy: 0.8571\n",
      "Epoch 38/200\n",
      "42/42 [==============================] - 0s 279us/step - loss: 0.5369 - accuracy: 0.8571\n",
      "Epoch 39/200\n",
      "42/42 [==============================] - 0s 274us/step - loss: 0.5969 - accuracy: 0.8095\n",
      "Epoch 40/200\n",
      "42/42 [==============================] - 0s 240us/step - loss: 0.3950 - accuracy: 0.9286\n",
      "Epoch 41/200\n",
      "42/42 [==============================] - 0s 210us/step - loss: 0.2954 - accuracy: 0.9286\n",
      "Epoch 42/200\n",
      "42/42 [==============================] - 0s 194us/step - loss: 0.2620 - accuracy: 0.9524\n",
      "Epoch 43/200\n",
      "42/42 [==============================] - 0s 224us/step - loss: 0.3184 - accuracy: 0.9048\n",
      "Epoch 44/200\n",
      "42/42 [==============================] - 0s 237us/step - loss: 0.3595 - accuracy: 0.9048\n",
      "Epoch 45/200\n",
      "42/42 [==============================] - 0s 214us/step - loss: 0.4525 - accuracy: 0.8810\n",
      "Epoch 46/200\n",
      "42/42 [==============================] - 0s 217us/step - loss: 0.2798 - accuracy: 1.0000\n",
      "Epoch 47/200\n",
      "42/42 [==============================] - 0s 220us/step - loss: 0.2507 - accuracy: 0.9524\n",
      "Epoch 48/200\n",
      "42/42 [==============================] - 0s 248us/step - loss: 0.2530 - accuracy: 0.9524\n",
      "Epoch 49/200\n",
      "42/42 [==============================] - 0s 234us/step - loss: 0.3133 - accuracy: 0.8810\n",
      "Epoch 50/200\n",
      "42/42 [==============================] - 0s 208us/step - loss: 0.3169 - accuracy: 0.9048\n",
      "Epoch 51/200\n",
      "42/42 [==============================] - 0s 195us/step - loss: 0.3237 - accuracy: 0.9524\n",
      "Epoch 52/200\n",
      "42/42 [==============================] - 0s 231us/step - loss: 0.2281 - accuracy: 0.9286\n",
      "Epoch 53/200\n",
      "42/42 [==============================] - 0s 231us/step - loss: 0.3415 - accuracy: 0.9286\n",
      "Epoch 54/200\n",
      "42/42 [==============================] - 0s 198us/step - loss: 0.3216 - accuracy: 0.9286\n",
      "Epoch 55/200\n",
      "42/42 [==============================] - 0s 202us/step - loss: 0.2430 - accuracy: 0.9762\n",
      "Epoch 56/200\n",
      "42/42 [==============================] - 0s 224us/step - loss: 0.2707 - accuracy: 0.9048\n",
      "Epoch 57/200\n",
      "42/42 [==============================] - 0s 243us/step - loss: 0.2938 - accuracy: 0.9286\n",
      "Epoch 58/200\n",
      "42/42 [==============================] - 0s 208us/step - loss: 0.3049 - accuracy: 0.8810\n",
      "Epoch 59/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.2677 - accuracy: 0.9286\n",
      "Epoch 60/200\n",
      "42/42 [==============================] - 0s 217us/step - loss: 0.3743 - accuracy: 0.9286\n",
      "Epoch 61/200\n",
      "42/42 [==============================] - 0s 233us/step - loss: 0.1305 - accuracy: 1.0000\n",
      "Epoch 62/200\n",
      "42/42 [==============================] - 0s 224us/step - loss: 0.1435 - accuracy: 0.9762\n",
      "Epoch 63/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.1842 - accuracy: 0.9524\n",
      "Epoch 64/200\n",
      "42/42 [==============================] - 0s 251us/step - loss: 0.2579 - accuracy: 0.9524\n",
      "Epoch 65/200\n",
      "42/42 [==============================] - 0s 270us/step - loss: 0.1091 - accuracy: 0.9762\n",
      "Epoch 66/200\n",
      "42/42 [==============================] - 0s 234us/step - loss: 0.2734 - accuracy: 0.9286\n",
      "Epoch 67/200\n",
      "42/42 [==============================] - 0s 228us/step - loss: 0.2156 - accuracy: 0.9524\n",
      "Epoch 68/200\n",
      "42/42 [==============================] - 0s 239us/step - loss: 0.1341 - accuracy: 0.9762\n",
      "Epoch 69/200\n",
      "42/42 [==============================] - 0s 240us/step - loss: 0.1725 - accuracy: 1.0000\n",
      "Epoch 70/200\n",
      "42/42 [==============================] - 0s 202us/step - loss: 0.2393 - accuracy: 0.9286\n",
      "Epoch 71/200\n",
      "42/42 [==============================] - 0s 213us/step - loss: 0.0920 - accuracy: 1.0000\n",
      "Epoch 72/200\n",
      "42/42 [==============================] - 0s 257us/step - loss: 0.3216 - accuracy: 0.9048\n",
      "Epoch 73/200\n",
      "42/42 [==============================] - 0s 250us/step - loss: 0.3538 - accuracy: 0.9048\n",
      "Epoch 74/200\n",
      "42/42 [==============================] - 0s 247us/step - loss: 0.2238 - accuracy: 0.9286\n",
      "Epoch 75/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.1465 - accuracy: 0.9524\n",
      "Epoch 76/200\n",
      "42/42 [==============================] - 0s 251us/step - loss: 0.1754 - accuracy: 0.9524\n",
      "Epoch 77/200\n",
      "42/42 [==============================] - 0s 264us/step - loss: 0.1643 - accuracy: 0.9762\n",
      "Epoch 78/200\n",
      "42/42 [==============================] - 0s 291us/step - loss: 0.2789 - accuracy: 0.9286\n",
      "Epoch 79/200\n",
      "42/42 [==============================] - 0s 246us/step - loss: 0.1655 - accuracy: 0.9524\n",
      "Epoch 80/200\n",
      "42/42 [==============================] - 0s 219us/step - loss: 0.2301 - accuracy: 0.9048\n",
      "Epoch 81/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42/42 [==============================] - 0s 214us/step - loss: 0.3880 - accuracy: 0.9048\n",
      "Epoch 82/200\n",
      "42/42 [==============================] - 0s 201us/step - loss: 0.1419 - accuracy: 0.9762\n",
      "Epoch 83/200\n",
      "42/42 [==============================] - 0s 227us/step - loss: 0.1669 - accuracy: 0.9524\n",
      "Epoch 84/200\n",
      "42/42 [==============================] - 0s 249us/step - loss: 0.1329 - accuracy: 0.9762\n",
      "Epoch 85/200\n",
      "42/42 [==============================] - 0s 231us/step - loss: 0.1390 - accuracy: 0.9762\n",
      "Epoch 86/200\n",
      "42/42 [==============================] - 0s 226us/step - loss: 0.1085 - accuracy: 1.0000\n",
      "Epoch 87/200\n",
      "42/42 [==============================] - 0s 222us/step - loss: 0.1098 - accuracy: 0.9762\n",
      "Epoch 88/200\n",
      "42/42 [==============================] - 0s 188us/step - loss: 0.1041 - accuracy: 0.9762\n",
      "Epoch 89/200\n",
      "42/42 [==============================] - 0s 215us/step - loss: 0.0710 - accuracy: 1.0000\n",
      "Epoch 90/200\n",
      "42/42 [==============================] - 0s 221us/step - loss: 0.2905 - accuracy: 0.9524\n",
      "Epoch 91/200\n",
      "42/42 [==============================] - 0s 214us/step - loss: 0.2014 - accuracy: 0.9524\n",
      "Epoch 92/200\n",
      "42/42 [==============================] - 0s 189us/step - loss: 0.1037 - accuracy: 0.9762\n",
      "Epoch 93/200\n",
      "42/42 [==============================] - 0s 220us/step - loss: 0.0560 - accuracy: 1.0000\n",
      "Epoch 94/200\n",
      "42/42 [==============================] - 0s 255us/step - loss: 0.1729 - accuracy: 0.9524\n",
      "Epoch 95/200\n",
      "42/42 [==============================] - 0s 249us/step - loss: 0.1197 - accuracy: 0.9762\n",
      "Epoch 96/200\n",
      "42/42 [==============================] - 0s 211us/step - loss: 0.1712 - accuracy: 0.9286\n",
      "Epoch 97/200\n",
      "42/42 [==============================] - 0s 193us/step - loss: 0.0847 - accuracy: 0.9762\n",
      "Epoch 98/200\n",
      "42/42 [==============================] - 0s 214us/step - loss: 0.1091 - accuracy: 0.9762\n",
      "Epoch 99/200\n",
      "42/42 [==============================] - 0s 216us/step - loss: 0.0865 - accuracy: 0.9762\n",
      "Epoch 100/200\n",
      "42/42 [==============================] - 0s 184us/step - loss: 0.0391 - accuracy: 1.0000\n",
      "Epoch 101/200\n",
      "42/42 [==============================] - 0s 216us/step - loss: 0.1162 - accuracy: 0.9524\n",
      "Epoch 102/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.2511 - accuracy: 0.9048\n",
      "Epoch 103/200\n",
      "42/42 [==============================] - 0s 221us/step - loss: 0.1228 - accuracy: 0.9762\n",
      "Epoch 104/200\n",
      "42/42 [==============================] - 0s 192us/step - loss: 0.0425 - accuracy: 1.0000\n",
      "Epoch 105/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.0544 - accuracy: 1.0000\n",
      "Epoch 106/200\n",
      "42/42 [==============================] - 0s 221us/step - loss: 0.1511 - accuracy: 0.9762\n",
      "Epoch 107/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.1383 - accuracy: 0.9524\n",
      "Epoch 108/200\n",
      "42/42 [==============================] - 0s 195us/step - loss: 0.1291 - accuracy: 0.9762\n",
      "Epoch 109/200\n",
      "42/42 [==============================] - 0s 205us/step - loss: 0.0826 - accuracy: 0.9762\n",
      "Epoch 110/200\n",
      "42/42 [==============================] - 0s 218us/step - loss: 0.1184 - accuracy: 0.9762\n",
      "Epoch 111/200\n",
      "42/42 [==============================] - 0s 233us/step - loss: 0.1098 - accuracy: 0.9762\n",
      "Epoch 112/200\n",
      "42/42 [==============================] - 0s 263us/step - loss: 0.0446 - accuracy: 1.0000\n",
      "Epoch 113/200\n",
      "42/42 [==============================] - 0s 191us/step - loss: 0.0593 - accuracy: 1.0000\n",
      "Epoch 114/200\n",
      "42/42 [==============================] - 0s 212us/step - loss: 0.1184 - accuracy: 0.9762\n",
      "Epoch 115/200\n",
      "42/42 [==============================] - 0s 219us/step - loss: 0.0278 - accuracy: 1.0000\n",
      "Epoch 116/200\n",
      "42/42 [==============================] - 0s 210us/step - loss: 0.1576 - accuracy: 0.9286\n",
      "Epoch 117/200\n",
      "42/42 [==============================] - 0s 223us/step - loss: 0.0173 - accuracy: 1.0000\n",
      "Epoch 118/200\n",
      "42/42 [==============================] - 0s 231us/step - loss: 0.0858 - accuracy: 0.9762\n",
      "Epoch 119/200\n",
      "42/42 [==============================] - 0s 288us/step - loss: 0.1101 - accuracy: 0.9762\n",
      "Epoch 120/200\n",
      "42/42 [==============================] - 0s 291us/step - loss: 0.1516 - accuracy: 0.9762\n",
      "Epoch 121/200\n",
      "42/42 [==============================] - 0s 287us/step - loss: 0.1557 - accuracy: 0.9286\n",
      "Epoch 122/200\n",
      "42/42 [==============================] - 0s 245us/step - loss: 0.1669 - accuracy: 0.9762\n",
      "Epoch 123/200\n",
      "42/42 [==============================] - 0s 233us/step - loss: 0.0552 - accuracy: 0.9762\n",
      "Epoch 124/200\n",
      "42/42 [==============================] - 0s 261us/step - loss: 0.1310 - accuracy: 0.9524\n",
      "Epoch 125/200\n",
      "42/42 [==============================] - 0s 294us/step - loss: 0.1309 - accuracy: 0.9524\n",
      "Epoch 126/200\n",
      "42/42 [==============================] - 0s 260us/step - loss: 0.1181 - accuracy: 0.9762\n",
      "Epoch 127/200\n",
      "42/42 [==============================] - 0s 218us/step - loss: 0.1279 - accuracy: 0.9524\n",
      "Epoch 128/200\n",
      "42/42 [==============================] - 0s 284us/step - loss: 0.1567 - accuracy: 0.9524\n",
      "Epoch 129/200\n",
      "42/42 [==============================] - 0s 244us/step - loss: 0.0525 - accuracy: 0.9762\n",
      "Epoch 130/200\n",
      "42/42 [==============================] - 0s 251us/step - loss: 0.0543 - accuracy: 0.9762\n",
      "Epoch 131/200\n",
      "42/42 [==============================] - 0s 326us/step - loss: 0.0506 - accuracy: 1.0000\n",
      "Epoch 132/200\n",
      "42/42 [==============================] - 0s 291us/step - loss: 0.1850 - accuracy: 0.9762\n",
      "Epoch 133/200\n",
      "42/42 [==============================] - 0s 255us/step - loss: 0.1423 - accuracy: 0.9524\n",
      "Epoch 134/200\n",
      "42/42 [==============================] - 0s 295us/step - loss: 0.0937 - accuracy: 1.0000\n",
      "Epoch 135/200\n",
      "42/42 [==============================] - 0s 276us/step - loss: 0.0353 - accuracy: 1.0000\n",
      "Epoch 136/200\n",
      "42/42 [==============================] - 0s 272us/step - loss: 0.1378 - accuracy: 0.9524\n",
      "Epoch 137/200\n",
      "42/42 [==============================] - 0s 290us/step - loss: 0.0468 - accuracy: 1.0000\n",
      "Epoch 138/200\n",
      "42/42 [==============================] - 0s 281us/step - loss: 0.0566 - accuracy: 0.9762\n",
      "Epoch 139/200\n",
      "42/42 [==============================] - 0s 291us/step - loss: 0.0467 - accuracy: 1.0000\n",
      "Epoch 140/200\n",
      "42/42 [==============================] - 0s 310us/step - loss: 0.0448 - accuracy: 1.0000\n",
      "Epoch 141/200\n",
      "42/42 [==============================] - 0s 254us/step - loss: 0.2090 - accuracy: 0.9524\n",
      "Epoch 142/200\n",
      "42/42 [==============================] - 0s 245us/step - loss: 0.1330 - accuracy: 0.9762\n",
      "Epoch 143/200\n",
      "42/42 [==============================] - 0s 213us/step - loss: 0.0377 - accuracy: 1.0000\n",
      "Epoch 144/200\n",
      "42/42 [==============================] - 0s 217us/step - loss: 0.0196 - accuracy: 1.0000\n",
      "Epoch 145/200\n",
      "42/42 [==============================] - 0s 220us/step - loss: 0.0808 - accuracy: 0.9762\n",
      "Epoch 146/200\n",
      "42/42 [==============================] - 0s 235us/step - loss: 0.0511 - accuracy: 0.9762\n",
      "Epoch 147/200\n",
      "42/42 [==============================] - 0s 239us/step - loss: 0.0527 - accuracy: 1.0000\n",
      "Epoch 148/200\n",
      "42/42 [==============================] - 0s 241us/step - loss: 0.0604 - accuracy: 1.0000\n",
      "Epoch 149/200\n",
      "42/42 [==============================] - 0s 269us/step - loss: 0.0546 - accuracy: 0.9762\n",
      "Epoch 150/200\n",
      "42/42 [==============================] - 0s 257us/step - loss: 0.0312 - accuracy: 1.0000\n",
      "Epoch 151/200\n",
      "42/42 [==============================] - 0s 230us/step - loss: 0.0497 - accuracy: 1.0000\n",
      "Epoch 152/200\n",
      "42/42 [==============================] - 0s 224us/step - loss: 0.1935 - accuracy: 0.9048\n",
      "Epoch 153/200\n",
      "42/42 [==============================] - 0s 241us/step - loss: 0.0462 - accuracy: 1.0000\n",
      "Epoch 154/200\n",
      "42/42 [==============================] - 0s 221us/step - loss: 0.0419 - accuracy: 1.0000\n",
      "Epoch 155/200\n",
      "42/42 [==============================] - 0s 226us/step - loss: 0.1754 - accuracy: 0.9524\n",
      "Epoch 156/200\n",
      "42/42 [==============================] - 0s 230us/step - loss: 0.0819 - accuracy: 0.9762\n",
      "Epoch 157/200\n",
      "42/42 [==============================] - 0s 220us/step - loss: 0.1076 - accuracy: 0.9524\n",
      "Epoch 158/200\n",
      "42/42 [==============================] - 0s 213us/step - loss: 0.0373 - accuracy: 1.0000\n",
      "Epoch 159/200\n",
      "42/42 [==============================] - 0s 218us/step - loss: 0.0887 - accuracy: 0.9762\n",
      "Epoch 160/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "42/42 [==============================] - 0s 245us/step - loss: 0.1118 - accuracy: 0.9762\n",
      "Epoch 161/200\n",
      "42/42 [==============================] - 0s 268us/step - loss: 0.0484 - accuracy: 1.0000\n",
      "Epoch 162/200\n",
      "42/42 [==============================] - 0s 264us/step - loss: 0.0513 - accuracy: 1.0000\n",
      "Epoch 163/200\n",
      "42/42 [==============================] - 0s 242us/step - loss: 0.0375 - accuracy: 1.0000\n",
      "Epoch 164/200\n",
      "42/42 [==============================] - 0s 262us/step - loss: 0.0364 - accuracy: 1.0000\n",
      "Epoch 165/200\n",
      "42/42 [==============================] - 0s 236us/step - loss: 0.0516 - accuracy: 1.0000\n",
      "Epoch 166/200\n",
      "42/42 [==============================] - 0s 302us/step - loss: 0.0266 - accuracy: 1.0000\n",
      "Epoch 167/200\n",
      "42/42 [==============================] - 0s 303us/step - loss: 0.2078 - accuracy: 0.9286\n",
      "Epoch 168/200\n",
      "42/42 [==============================] - 0s 267us/step - loss: 0.0549 - accuracy: 0.9762\n",
      "Epoch 169/200\n",
      "42/42 [==============================] - 0s 276us/step - loss: 0.0821 - accuracy: 1.0000\n",
      "Epoch 170/200\n",
      "42/42 [==============================] - 0s 229us/step - loss: 0.0287 - accuracy: 1.0000\n",
      "Epoch 171/200\n",
      "42/42 [==============================] - 0s 227us/step - loss: 0.0200 - accuracy: 1.0000\n",
      "Epoch 172/200\n",
      "42/42 [==============================] - 0s 208us/step - loss: 0.0194 - accuracy: 1.0000\n",
      "Epoch 173/200\n",
      "42/42 [==============================] - 0s 255us/step - loss: 0.0796 - accuracy: 1.0000\n",
      "Epoch 174/200\n",
      "42/42 [==============================] - 0s 312us/step - loss: 0.0764 - accuracy: 0.9762\n",
      "Epoch 175/200\n",
      "42/42 [==============================] - 0s 281us/step - loss: 0.0237 - accuracy: 1.0000\n",
      "Epoch 176/200\n",
      "42/42 [==============================] - 0s 249us/step - loss: 0.1830 - accuracy: 0.9524\n",
      "Epoch 177/200\n",
      "42/42 [==============================] - 0s 270us/step - loss: 0.1300 - accuracy: 0.9524\n",
      "Epoch 178/200\n",
      "42/42 [==============================] - 0s 280us/step - loss: 0.0462 - accuracy: 1.0000\n",
      "Epoch 179/200\n",
      "42/42 [==============================] - 0s 263us/step - loss: 0.0547 - accuracy: 0.9762\n",
      "Epoch 180/200\n",
      "42/42 [==============================] - 0s 252us/step - loss: 0.0415 - accuracy: 1.0000\n",
      "Epoch 181/200\n",
      "42/42 [==============================] - 0s 241us/step - loss: 0.0512 - accuracy: 1.0000\n",
      "Epoch 182/200\n",
      "42/42 [==============================] - 0s 272us/step - loss: 0.1083 - accuracy: 1.0000\n",
      "Epoch 183/200\n",
      "42/42 [==============================] - 0s 300us/step - loss: 0.0924 - accuracy: 0.9762\n",
      "Epoch 184/200\n",
      "42/42 [==============================] - 0s 281us/step - loss: 0.0562 - accuracy: 1.0000\n",
      "Epoch 185/200\n",
      "42/42 [==============================] - 0s 273us/step - loss: 0.0315 - accuracy: 1.0000\n",
      "Epoch 186/200\n",
      "42/42 [==============================] - 0s 279us/step - loss: 0.1160 - accuracy: 0.9762\n",
      "Epoch 187/200\n",
      "42/42 [==============================] - 0s 285us/step - loss: 0.2084 - accuracy: 0.9762\n",
      "Epoch 188/200\n",
      "42/42 [==============================] - 0s 272us/step - loss: 0.0238 - accuracy: 1.0000\n",
      "Epoch 189/200\n",
      "42/42 [==============================] - 0s 217us/step - loss: 0.0287 - accuracy: 1.0000\n",
      "Epoch 190/200\n",
      "42/42 [==============================] - 0s 227us/step - loss: 0.0315 - accuracy: 1.0000\n",
      "Epoch 191/200\n",
      "42/42 [==============================] - 0s 228us/step - loss: 0.0244 - accuracy: 1.0000\n",
      "Epoch 192/200\n",
      "42/42 [==============================] - 0s 208us/step - loss: 0.0671 - accuracy: 0.9762\n",
      "Epoch 193/200\n",
      "42/42 [==============================] - 0s 213us/step - loss: 0.0926 - accuracy: 0.9762\n",
      "Epoch 194/200\n",
      "42/42 [==============================] - 0s 221us/step - loss: 0.0950 - accuracy: 0.9524\n",
      "Epoch 195/200\n",
      "42/42 [==============================] - 0s 249us/step - loss: 0.0492 - accuracy: 0.9762\n",
      "Epoch 196/200\n",
      "42/42 [==============================] - 0s 289us/step - loss: 0.0786 - accuracy: 0.9762\n",
      "Epoch 197/200\n",
      "42/42 [==============================] - 0s 259us/step - loss: 0.0610 - accuracy: 0.9762\n",
      "Epoch 198/200\n",
      "42/42 [==============================] - 0s 236us/step - loss: 0.0624 - accuracy: 0.9762\n",
      "Epoch 199/200\n",
      "42/42 [==============================] - 0s 226us/step - loss: 0.0207 - accuracy: 1.0000\n",
      "Epoch 200/200\n",
      "42/42 [==============================] - 0s 216us/step - loss: 0.0504 - accuracy: 0.9762\n",
      "model created\n"
     ]
    }
   ],
   "source": [
    "random.shuffle(training)\n",
    "training = np.array(training)\n",
    "# create train and test lists. X - patterns, Y - intents\n",
    "train_x = list(training[:,0])\n",
    "train_y = list(training[:,1])\n",
    "print(\"Training data created\")\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Dense(128, input_shape=(len(train_x[0]),), activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(len(train_y[0]), activation='softmax'))\n",
    "\n",
    "# Compile model. Stochastic gradient descent with Nesterov accelerated gradient gives good results for this model\n",
    "sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)\n",
    "model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])\n",
    "\n",
    "#fitting and saving the model\n",
    "hist = model.fit(np.array(train_x), np.array(train_y), epochs=200, batch_size=5, verbose=1)\n",
    "model.save('chatbot_model.h5', hist)\n",
    "\n",
    "print(\"model created\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import nltk\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "lemmatizer = WordNetLemmatizer()\n",
    "import pickle\n",
    "import numpy as np\n",
    "\n",
    "from keras.models import load_model\n",
    "model = load_model('chatbot_model.h5')\n",
    "import json\n",
    "import random\n",
    "intents = json.loads(open('intents.json').read())\n",
    "words = pickle.load(open('words.pkl','rb'))\n",
    "classes = pickle.load(open('classes.pkl','rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[\"'s\",\n",
       " '2',\n",
       " 'advice',\n",
       " 'aid',\n",
       " 'allergy',\n",
       " 'anyone',\n",
       " 'asthma',\n",
       " 'awesome',\n",
       " 'breast',\n",
       " 'bye',\n",
       " 'calorie',\n",
       " 'can',\n",
       " 'cancer',\n",
       " 'care',\n",
       " 'cause',\n",
       " 'chatting',\n",
       " 'child',\n",
       " 'childhood',\n",
       " 'could',\n",
       " 'cyberbullying',\n",
       " 'day',\n",
       " 'developing',\n",
       " 'diabetes',\n",
       " 'diet',\n",
       " 'do',\n",
       " 'doe',\n",
       " 'drinking',\n",
       " 'earache',\n",
       " 'eat',\n",
       " 'eating',\n",
       " 'effect',\n",
       " 'factor',\n",
       " 'genetic',\n",
       " 'get',\n",
       " 'go',\n",
       " 'good',\n",
       " 'goodbye',\n",
       " 'greek',\n",
       " 'hard',\n",
       " 'healthier',\n",
       " 'hearing',\n",
       " 'hello',\n",
       " 'help',\n",
       " 'helpful',\n",
       " 'helping',\n",
       " 'hey',\n",
       " 'hi',\n",
       " 'hola',\n",
       " 'how',\n",
       " 'i',\n",
       " 'inherited',\n",
       " 'is',\n",
       " 'kidney',\n",
       " 'later',\n",
       " 'lemon',\n",
       " 'life',\n",
       " 'liquor',\n",
       " 'loss',\n",
       " 'lupus',\n",
       " 'making',\n",
       " 'motivated',\n",
       " 'need',\n",
       " 'next',\n",
       " 'nice',\n",
       " 'offered',\n",
       " 'often',\n",
       " 'patient',\n",
       " 'plan',\n",
       " 'preschool',\n",
       " 'prevent',\n",
       " 'protein',\n",
       " 'provide',\n",
       " 'recommend',\n",
       " 'regular',\n",
       " 'remission',\n",
       " 'replace',\n",
       " 'see',\n",
       " 'sense',\n",
       " 'shutup',\n",
       " 'something',\n",
       " 'stay',\n",
       " 'stone',\n",
       " 'stroke',\n",
       " 'support',\n",
       " 'tell',\n",
       " 'thank',\n",
       " 'thanks',\n",
       " 'that',\n",
       " 'till',\n",
       " 'time',\n",
       " 'tool',\n",
       " 'trying',\n",
       " 'type',\n",
       " 'vegan',\n",
       " 'vegetarian',\n",
       " 'water',\n",
       " 'what',\n",
       " 'yogurt',\n",
       " 'you']"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['breast_cancer',\n",
       " 'calories',\n",
       " 'childhood_earaches',\n",
       " 'cyberbullying',\n",
       " 'day_care',\n",
       " 'eating_plan',\n",
       " 'goodbye',\n",
       " 'greeting',\n",
       " 'hearing_aids',\n",
       " 'kidney_stones',\n",
       " 'options',\n",
       " 'shutup',\n",
       " 'stroke',\n",
       " 'thanks',\n",
       " 'type2_diabetes',\n",
       " 'vegan_diet',\n",
       " 'vegetarian',\n",
       " 'yogurt']"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_up_sentence(sentence):\n",
    "    sentence_words = nltk.word_tokenize(sentence)\n",
    "    sentence_words = [lemmatizer.lemmatize(word.lower()) for word in sentence_words]\n",
    "    return sentence_words\n",
    "\n",
    "# return bag of words array: 0 or 1 for each word in the bag that exists in the sentence\n",
    "\n",
    "def bow(sentence, words, show_details=True):\n",
    "    # tokenize the pattern\n",
    "    sentence_words = clean_up_sentence(sentence)\n",
    "    # bag of words - matrix of N words, vocabulary matrix\n",
    "    bag = [0]*len(words)\n",
    "    for s in sentence_words:\n",
    "        for i,w in enumerate(words):\n",
    "            if w == s:\n",
    "                # assign 1 if current word is in the vocabulary position\n",
    "                bag[i] = 1\n",
    "                if show_details:\n",
    "                    print (\"found in bag: %s\" % w)\n",
    "    return(np.array(bag))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "p = bow('How to love Eva Wang', words, show_details = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0])"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]])"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array([p])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "res = model.predict(np.array([p]))[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.9567994e-04, 9.8773882e-05, 5.2265310e-05, 3.1873654e-04,\n",
       "       7.9671678e-05, 1.3645980e-05, 2.8414867e-04, 9.9508154e-01,\n",
       "       1.1016459e-03, 4.8743255e-05, 3.9591364e-04, 8.4698841e-04,\n",
       "       7.1388495e-05, 5.0170198e-05, 1.8295288e-05, 3.8299466e-05,\n",
       "       1.1244571e-03, 1.7953161e-04], dtype=float32)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "ERROR_THRESHOLD = 0.25\n",
    "results = [[i,r] for i,r in enumerate(res) if r>ERROR_THRESHOLD]\n",
    "    # sort by strength of probability\n",
    "results.sort(key=lambda x: x[1], reverse=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[7, 0.99508154]]"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "return_list = []\n",
    "for r in results:\n",
    "        return_list.append({\"intent\": classes[r[0]], \"probability\": str(r[1])})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'intent': 'greeting', 'probability': '0.99508154'}]"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "return_list\n"
   ]
  },
  {
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}