chat.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First Goal is to make the computer understand the language<br>\n",
    "\n",
    "Here are the steps\n",
    "\n",
    "1. Sentence segmentation\n",
    "2. Word tokenization\n",
    "3. Stemming\n",
    "4. Dependency parsing\n",
    "5. Part-of-speech (POS) tagging"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from nltk.stem.snowball import SnowballStemmer\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "import matplotlib.pyplot as plt\n",
    "import re\n",
    "import spacy\n",
    "from keras.models import Sequential\n",
    "from sklearn.utils.class_weight import compute_class_weight\n",
    "from keras.layers import Dense, Embedding, LSTM\n",
    "from spacy import displacy\n",
    "import nltk\n",
    "import tensorflow as tf\n",
    "from nltk.parse.dependencygraph import DependencyGraph\n",
    "from prettytable import PrettyTable\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "TAGS = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Prepare a proper storage**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Tweet:\n",
    "    def __init__(self, count, hate_speech_count, offensive_language_count, neither_count, classs, tweet):\n",
    "        \"\"\"\n",
    "        - count (int): The total count of the tweet.\n",
    "        - hate (int): The count of hate speech in the tweet.\n",
    "        - offensive (int): The count of offensive language in the tweet.\n",
    "        - neither (int): The count of content classified as neither hate speech nor offensive.\n",
    "        - classs (str): The classification of the tweet.\n",
    "        - tweet (str): The text content of the tweet.\n",
    "        \"\"\"\n",
    "        self.count = count\n",
    "        self.hate = hate_speech_count\n",
    "        self.offensive = offensive_language_count\n",
    "        self.neither = neither_count\n",
    "        self.classs = classs\n",
    "        self.tweet = tweet\n",
    "        self.stems = []\n",
    "        self.tokens = []\n",
    "        self.code = []\n",
    "        self.tags = []\n",
    "        self.t_code = []\n",
    "        self.presence_word = []\n",
    "\n",
    "    def __str__(self):\n",
    "        return f\"{self.count} ; {self.hate} ; {self.offensive} ; {self.neither} ; {self.classs} ;; {self.tweet}\"\n",
    "    \n",
    "    def peacefullness(self):\n",
    "        return self.neither_count / self.count\n",
    "    \n",
    "    def offensiveness(self):\n",
    "        return self.offensive / self.count\n",
    "    \n",
    "    def hateness(self):\n",
    "        return self.hate / self.count"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Load the data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "Tweets = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('./archive/train.csv', 'r+') as file:\n",
    "    previous_line = ''\n",
    "\n",
    "    # Initialize a list to accumulate the modified content\n",
    "    final_content_lines = []\n",
    "\n",
    "    # Read and accumulate non-empty lines\n",
    "    for line in file:\n",
    "        stripped_line = line.strip()\n",
    "\n",
    "        if stripped_line and stripped_line[0].isdigit():\n",
    "            # If the current line is not empty and starts with an integer, accumulate it\n",
    "            final_content_lines.append(stripped_line)\n",
    "            previous_line = stripped_line\n",
    "        else:\n",
    "            # If the current line doesn't start with an integer, append it to the previous line\n",
    "            previous_line += stripped_line"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "for line in final_content_lines:\n",
    "    \n",
    "    comma_indices = [index for index, char in enumerate(line) if char == ',']\n",
    "\n",
    "    # Extracting substrings between commas\n",
    "    count_str = line[0:comma_indices[0]].strip()\n",
    "    hate_str = line[comma_indices[0]+1:comma_indices[1]].strip()\n",
    "    offensive_str = line[comma_indices[1]+1:comma_indices[2]].strip()\n",
    "    neither_str = line[comma_indices[2]+1:comma_indices[3]].strip()\n",
    "    classs_str = line[comma_indices[3]+1:comma_indices[4]].strip()\n",
    "    tweet_str = line[comma_indices[4]+1:].strip()\n",
    "\n",
    "    # Converting to integers\n",
    "    count = int(count_str) if count_str.isdigit() else None\n",
    "    hate = int(hate_str) if hate_str.isdigit() else None\n",
    "    offensive = int(offensive_str) if offensive_str.isdigit() else None\n",
    "    neither = int(neither_str) if neither_str.isdigit() else None\n",
    "    classs = int(classs_str) if classs_str.isdigit() else None\n",
    "\n",
    "    # Creating an instance of the Tweet class\n",
    "    tweet_instance = Tweet(count, hate, offensive, neither, classs_str, tweet_str)\n",
    "\n",
    "    # Append the tweet instance to a list or do whatever you need to do with it\n",
    "    Tweets.append(tweet_instance)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3 ; 0 ; 0 ; 3 ; 2 ;; !!! RT @mayasolovely: As a woman you shouldn't complain about cleaning up your house. &amp; as a man you should always take the trash out...\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; !!!!! RT @mleew17: boy dats cold...tyga dwn bad for cuffin dat hoe in the 1st place!!\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; !!!!!!! RT @UrKindOfBrand Dawg!!!! RT @80sbaby4life: You ever fuck a bitch and she start to cry? You be confused as shit\n",
      "3 ; 0 ; 2 ; 1 ; 1 ;; !!!!!!!!! RT @C_G_Anderson: @viva_based she look like a tranny\n",
      "6 ; 0 ; 6 ; 0 ; 1 ;; !!!!!!!!!!!!! RT @ShenikaRoberts: The shit you hear about me might be true or it might be faker than the bitch who told it to ya &#57361;\n",
      "3 ; 1 ; 2 ; 0 ; 1 ;; \"!!!!!!!!!!!!!!!!!!\"\"@T_Madison_x: The shit just blows me..claim you so faithful and down for somebody but still fucking with hoes! &#128514;&#128514;&#128514;\"\"\"\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; \"!!!!!!\"\"@__BrighterDays: I can not just sit up and HATE on another bitch .. I got too much shit going on!\"\"\"\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; !!!!&#8220;@selfiequeenbri: cause I'm tired of you big bitches coming for us skinny girls!!&#8221;\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; \"\"\" &amp; you might not get ya bitch back &amp; thats that \"\"\"\n",
      "3 ; 1 ; 2 ; 0 ; 1 ;; \"\"\" @rhythmixx_ :hobbies include: fighting Mariam\"\"\n"
     ]
    }
   ],
   "source": [
    "for t in Tweets[:10]:\n",
    "    print(t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tweet formatting**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "remove all points when it's not the end of a sentence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_points(line):\n",
    "    l = list(line)\n",
    "\n",
    "    for i in range(len(l)):\n",
    "        if i < len(l) - 2 and l[i] == '.' and 'A' <= l[i+2] <= 'Z':\n",
    "            l[i] = ' '\n",
    "\n",
    "    line = ' '.join(l)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "apply first format"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's change:\n",
    "- all `@<username>` to `username`\n",
    "- all URLs to `weblink`\n",
    "- all `&amp` to `&`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    t.tweet = re.sub(r'@([a-zA-Z0-9_]+)', 'username', t.tweet).replace('username:', '') # replace first username\n",
    "    t.tweet = re.sub(r'http?://\\S+', 'weblink', t.tweet)\n",
    "    t.tweet = re.sub(r'&amp', '&', t.tweet)\n",
    "\n",
    "    if t.tweet and t.tweet[0] == '.':\n",
    "        t.tweet = t.tweet[1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3 ; 0 ; 0 ; 3 ; 2 ;; As a woman you shouldnt complain about cleaning up your house. & as a man you should always take the trash out.\n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; boy dats cold.tyga dwn bad for cuffin dat hoe in the 1st place \n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; username Dawg You ever fuck a bitch & she start to cry. You be confused as shit\n",
      "3 ; 0 ; 2 ; 1 ; 1 ;; username she look like a tranny\n",
      "6 ; 0 ; 6 ; 0 ; 1 ;; The shit you hear about me might be true or it might be faker than the bitch who told it to ya 57361 \n",
      "3 ; 1 ; 2 ; 0 ; 1 ;;  The shit just blows me.claim you so faithful & down for somebody but still fucking with hoes 128514 128514 128514 \n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; I can not just sit up & HATE on another bitch. I got too much shit going on \n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; 8220 cause Im tired of you big bitches coming for us skinny girls 8221 \n",
      "3 ; 0 ; 3 ; 0 ; 1 ;; & you might not get ya bitch back & thats that \n",
      "3 ; 1 ; 2 ; 0 ; 1 ;; username :hobbies include: fighting Mariam\n"
     ]
    }
   ],
   "source": [
    "for t in Tweets:\n",
    "    remove_points(t.tweet)\n",
    "    t.tweet = t.tweet   .replace('RT', '').replace('!', ' ').replace('\"', '').replace(\"\\n\", ' ')\\\n",
    "                        .replace(';', ' ').replace('-', ' ').replace(' and ', ' & ').replace('\\'', '')\\\n",
    "                        .replace('?', '.').replace(',', '').replace('~', ' ').replace('|', ' ').replace('°', ' ')\\\n",
    "                        .replace('`', ' ').replace('~', ' ').replace('*', ' ').replace('+', ' ').replace('/', ' ')\\\n",
    "                        .replace(' # ', ' ').replace('http', ' ').replace('t.co', ' ').replace('\\\\', ' ').replace('&#', ' ')\n",
    "    for _ in range(4):\n",
    "        t.tweet = t.tweet.replace('  ', ' ').replace('..', '.').replace(' .', '.') # remove multiple points & space\n",
    "    \n",
    "    if t.tweet and t.tweet[0] == ' ':\n",
    "        t.tweet = t.tweet[1:]\n",
    "\n",
    "\n",
    "for t in Tweets[:10]:\n",
    "    print(t)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Change everything in **lowercase**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    t.tweet = t.tweet.lower()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Change some **abbreviations**<br>\n",
    "Here it is important not to change all abbreviations. some abbreviations are used by 50% of population. our model will think they are different if we don't change those 50% - 50% to a 100% unique expressions. But on the other hand, some are used by all the population (eg. `lol`) so we can keep them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "abbr = {\n",
    "    'ninstagram'        : 'instagram',\n",
    "    'instagramgram'     : 'instagram',\n",
    "    'ig'                : 'instagram',\n",
    "    'strainstagramht'   : 'instagram',\n",
    "    'insta'             : 'instagram',\n",
    "    'rinstagramht'      : 'instagram',\n",
    "    'ninstagramguh'     : 'instagram',\n",
    "    'instagramz'        : 'instagram',\n",
    "    'sinstagramn'       : 'instagram',\n",
    "    'binstagramgest'    : 'instagram',\n",
    "    'pinstagram'        : 'instagram',\n",
    "    'linstagramht'      : 'instagram',\n",
    "    'ninstagramg'       : 'instagram',\n",
    "    'instagramh'        : 'instagram',\n",
    "    'instagramnor'      : 'instagram',\n",
    "    'ninstagramht'      : 'instagram',\n",
    "    'ninstagramgramga'  : 'instagram',\n",
    "    'finstagramht'      : 'instagram',\n",
    "    'binstagram'        : 'instagram',\n",
    "    'hinstagramh'       : 'instagram',\n",
    "    'ninstagramga'      : 'instagram',\n",
    "    'toninstagramht'    : 'instagram',\n",
    "    'minstagramht'      : 'instagram',\n",
    "    'minstagramt'       : 'instagram',\n",
    "    'dwn'               : 'down',\n",
    "    'dawn'              : 'down',\n",
    "    'ta'                : 'that',\n",
    "    'dat'               : 'that',\n",
    "    'dawg'              : 'dude',\n",
    "    'smh'               : 'head',\n",
    "    'fr'                : 'real',\n",
    "    'plz'               : 'please',\n",
    "    'tf'                : 'wtf',\n",
    "    'theyr'             : 'are',\n",
    "    'bc'                : 'because',\n",
    "    'af'                : 'lot',\n",
    "    'u'                 : 'you',\n",
    "    'ppl'               : 'people',\n",
    "    'dm'                : 'message',\n",
    "    'bf'                : 'friend',\n",
    "    'gt'                : 'getting',\n",
    "    'ya'                : 'yes',\n",
    "    'na'                : 'no',\n",
    "    'ur'                : 'your',\n",
    "    'tryna'             : 'to',\n",
    "    'lmfao'             : 'lmao',\n",
    "    'ive'               : 'have'\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply the modification "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for old in abbr:\n",
    "        new = abbr[old]\n",
    "        t.tweet = t.tweet.replace( ' ' + old + ' ', ' ' + new + ' ' ) # add some space arround the world to avoid matching a part of a word"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a function that reduce the **repetitions** (people use to add a lot of letters of type a lot of emojis)<br>This reduce the data to handle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def reduce_repetition(s):\n",
    "    # Use regular expression to find repeated substrings\n",
    "    pattern = re.compile(r'(.+?)\\1{%d,}' % 2)\n",
    "    match = pattern.search(s)\n",
    "\n",
    "    # Reduce repetition to two occurrences\n",
    "    while match:\n",
    "        repeated_substring = match.group(1)\n",
    "        s = s.replace(match.group(), repeated_substring, 1)\n",
    "        match = pattern.search(s)\n",
    "\n",
    "    return s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    t.tweet = reduce_repetition(t.tweet)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Tokenisation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    t.tokens += nltk.word_tokenize(t.tweet)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for i in range(len(t.tokens) - 1, 0, -1):\n",
    "        if len(t.tokens[i].strip()) == 0 or (len(t.tokens[i].strip()) == 1 and t.tokens[i].strip() != 'a' and t.tokens[i].strip() != 'i' and t.tokens[i].strip() != '&'):\n",
    "            t.tokens.pop(i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Stemmatisation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "stemmer = SnowballStemmer('english')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "let's **compare** original string aigains its stemmed version to see the difference and say if the **stemmatisation** is **effective**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Split according to multiple charachters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for tok in t.tokens:\n",
    "        t.stems.append(stemmer.stem(tok))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for i in range(len(t.stems) - 1, 0, -1):\n",
    "        if len(t.stems[i].strip()) == 0:\n",
    "            t.stems.pop(i)\n",
    "        elif len(t.stems[i].strip()) == 1:\n",
    "            if 'a' < t.stems[i].strip() <= 'z'and t.stems[i].strip() != 'i':\n",
    "                t.stems.pop(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for i in range(len(t.stems) - 1, 0, -1):\n",
    "        arr = t.stems[i].split('.')\n",
    "        t.stems[i] = arr[0]\n",
    "        for j in range(1, len(arr)):\n",
    "            t.stems.insert(i+j, arr[j])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Re-clean one time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for i in range(len(t.stems)):\n",
    "        if t.stems[i] in abbr.keys():\n",
    "            t.stems[i] = abbr[t.stems[i]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Dependency parsing**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    nlp = spacy.load(\"en_core_web_sm\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    print(Tweets[12].tweet)\n",
    "    print(Tweets[12].tokens)\n",
    "    print(Tweets[12].stems, \"\\n\")\n",
    "\n",
    "    doc = nlp(' '.join(Tweets[12].stems))\n",
    "    for token in doc:\n",
    "        print(f\"{token.text}: {token.dep_} -> {token.head.text}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# displacy.serve(doc, style=\"dep\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This allow us to know the role of a word in the sentence<br>\n",
    "let's store those information"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    for t in Tweets:\n",
    "        doc = nlp(' '.join(t.stems))\n",
    "        for token in doc:\n",
    "            t.tags.append(token.dep_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "List the most used words"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRESHOLD = 3\n",
    "code_table = []"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will record words if they are used more than `TRESHOLD`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "used_w = {}\n",
    "\n",
    "for t in Tweets:\n",
    "    for s in t.stems:\n",
    "        if not s in used_w.keys():\n",
    "            used_w[s] =  1\n",
    "        else:\n",
    "            used_w[s] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "for k, v in sorted(used_w.items(), key=lambda item: item[1], reverse=True):\n",
    "    if v < TRESHOLD:\n",
    "        break\n",
    "    code_table.append(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5707\n"
     ]
    }
   ],
   "source": [
    "print(len(code_table))\n",
    "MAX_CODE = len(code_table) + 1 # we book 1 index for the non-referenced words"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Encode-Decode** functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def enc(word):\n",
    "    index = -1\n",
    "    try:\n",
    "        index = code_table.index(word)\n",
    "    except ValueError:\n",
    "        index = -1\n",
    "    \n",
    "    return index + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dec(index):\n",
    "    if index-1 >= len(code_table) or index-1 < 0: return ''\n",
    "    return code_table[index-1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A quick verification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "hello\n"
     ]
    }
   ],
   "source": [
    "print(dec(enc('hello')))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "for t in Tweets:\n",
    "    for s in t.stems:\n",
    "        t.code.append(enc(s))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's **encode** also the **tags** for the words. It will be **easier** to **handle**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    d_tags = {}\n",
    "    \n",
    "    for t in Tweets:\n",
    "        for g in t.tags:\n",
    "            if not g in d_tags.keys():\n",
    "                d_tags[g]  = 1\n",
    "            else:\n",
    "                d_tags[g] += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: Here it's not necessary to **sort** the tags usage count but it's not slowing the learning process"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    tag_code_table = []\n",
    "    for k, v in sorted(d_tags.items(), key=lambda item: item[1], reverse=True):\n",
    "        tag_code_table.append(k)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    print(tag_code_table)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "def t_enc(tag):\n",
    "    index = -1\n",
    "    try:\n",
    "        index = tag_code_table.index(tag)\n",
    "    except ValueError:\n",
    "        index = -1\n",
    "    \n",
    "    return index + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def t_dec(index):\n",
    "    ind = index - 1\n",
    "    if ind >= len(tag_code_table) or ind < 0: return ''\n",
    "    return tag_code_table(ind)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Record that data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    for t in Tweets:\n",
    "        for g in t.tags:\n",
    "            t.t_code.append(t_enc(g))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's how our data looks so far!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "if TAGS:\n",
    "    table = PrettyTable([\"TAG\", \"tag CODE\", \"STEM\", \"stem CODE\"])\n",
    "\n",
    "    for t in Tweets[:5]:\n",
    "        for i in range(len(t.stems)):\n",
    "            table.add_row([t.tags[i], t.t_code[i], t.stems[i], t.code[i]])\n",
    "        table.add_row(['-', '-', '-', '-'])\n",
    "\n",
    "    print(table)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Time for making a model !"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here are my ideas for the input of the model \n",
    "\n",
    "1. An array of this type:\n",
    "  - [$stem_0$, $tag_0$, $stem_1$, $tag_1$, $stem_2$, $tag_2$ ... $stem_n$, $tag_n$]\n",
    "\n",
    "2. Or an array of this type: \n",
    "  - [$stem_0$, $stem_1$, $stem_2$ ... $stem_n$, $tag_0$, $tag_1$, $tag_2$ ... $tag_n$]\n",
    "\n",
    "3. Or two input arrays:\n",
    "  - stem code\n",
    "  - tag code"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Keras** offers us the `Embedding` layer with those arguments:\n",
    "\n",
    "- input_dim: Integer. Size of the vocabulary, i.e. maximum integer index + 1.\n",
    "\n",
    "- output_dim: Integer. Dimension of the dense embedding.\n",
    "\n",
    "- embeddings_initializer: Initializer for the embeddings matrix (see keras.initializers).\n",
    "\n",
    "- embeddings_regularizer: Regularizer function applied to the embeddings matrix (see keras.regularizers).\n",
    "\n",
    "- embeddings_constraint: Constraint function applied to the embeddings matrix (see keras.constraints).\n",
    "\n",
    "- mask_zero: Boolean, whether or not the input value 0 is a special \"padding\" value that should be masked out.\n",
    "\n",
    "This is useful when using recurrent layers which may take variable length input. If this is True, then all subsequent layers in the model need to support masking or an exception will be raised. If mask_zero is set to True, as a consequence, index 0 cannot be used in the vocabulary (input_dim should equal size of vocabulary + 1).\n",
    "\n",
    "> Source: [Keras' documentation](https://keras.io/api/layers/core_layers/embedding/)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have to figure the maximum integer index. in the 1st and 2nd case, it is equals to two times the number of words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "max   = 67 \n",
      "2*max = 134\n"
     ]
    }
   ],
   "source": [
    "max = 0\n",
    "for t in Tweets:\n",
    "    if max < len(t.stems):\n",
    "        max = len(t.stems)\n",
    "\n",
    "print(\"max   =\", max, \"\\n2*max =\", 2*max)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A maximal input dimension of **500** seems enough and will allow the user to input a large text up to 500 words and **compute** it in the **model**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train without the word tags"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model wants 100 integers input. almost every tweet is about 50 words, let's add padding of -1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "INPUT= 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = []\n",
    "for t in Tweets:\n",
    "    row = []\n",
    "    for c in t.code:\n",
    "        row.append(c)\n",
    "    elements_to_add = INPUT - len(row)\n",
    "    \n",
    "    for _ in range(elements_to_add):\n",
    "        row.append(0)\n",
    "\n",
    "    x.append(row)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(24783, 100)\n"
     ]
    }
   ],
   "source": [
    "X = np.array(x)\n",
    "x = None # free memory\n",
    "\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "create the hateness vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_hate = []\n",
    "for t in Tweets:\n",
    "    y_hate.append(t.hateness())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_hate = np.array(y_hate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(24783,)\n"
     ]
    }
   ],
   "source": [
    "print(y_hate.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y_hate, test_size=.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(19826, 100)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding (Embedding)       (None, 100, 50)           285400    \n",
      "                                                                 \n",
      " lstm (LSTM)                 (None, 100)               60400     \n",
      "                                                                 \n",
      " dense (Dense)               (None, 1)                 101       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 345901 (1.32 MB)\n",
      "Trainable params: 345901 (1.32 MB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Assuming you have a vocabulary size and maximum sequence length\n",
    "vocab_size = MAX_CODE  # total Number of words\n",
    "max_seq_length = 100  # max number of words in a sentence\n",
    "\n",
    "# Define the model\n",
    "model = Sequential()\n",
    "\n",
    "# Add an Embedding layer to convert words into vectors\n",
    "model.add(Embedding(input_dim=vocab_size, output_dim=50, input_length=max_seq_length))\n",
    "\n",
    "# Add an LSTM layer to capture sequential dependencies\n",
    "model.add(LSTM(units=100, dropout=0.2, recurrent_dropout=0.2))\n",
    "\n",
    "# Add a Dense layer for binary classification (output 1 or 0)\n",
    "model.add(Dense(units=1, activation='sigmoid'))\n",
    "\n",
    "# Print the model summary\n",
    "model.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "KerasTensor(type_spec=TensorSpec(shape=(None, 100), dtype=tf.float32, name='embedding_input'), name='embedding_input', description=\"created by layer 'embedding_input'\")\n"
     ]
    }
   ],
   "source": [
    "print(model.layers[0].input)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As my data is not **balanced**, I will use this dictionnary to create an artificial balancing. \n",
    "Augmenting the data would have been too difficult for me."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.0, 0.1111111111111111, 0.14285714285714285, 0.16666666666666666, 0.2222222222222222, 0.25, 0.2857142857142857, 0.3333333333333333, 0.42857142857142855, 0.4444444444444444, 0.5, 0.5555555555555556, 0.6666666666666666, 0.75, 0.7777777777777778, 0.8333333333333334, 0.8571428571428571, 1.0]\n"
     ]
    }
   ],
   "source": [
    "dic = {}\n",
    "for i in y_train:\n",
    "    dic[i] = 0\n",
    "\n",
    "cc = []\n",
    "\n",
    "for i in (dic.keys()):\n",
    "    cc.append(i)\n",
    "\n",
    "cc.sort()\n",
    "\n",
    "print(cc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[6.96103422e-02 4.07942387e+01 2.20288889e+02 4.23632479e+00\n",
      " 6.88402778e+01 3.33771044e+01 1.10144444e+03 4.37950077e-01\n",
      " 3.67148148e+02 2.75361111e+02 2.20288889e+02 5.50722222e+02\n",
      " 1.21841200e+00 2.20288889e+02 3.67148148e+02 2.75361111e+02\n",
      " 1.10144444e+03 5.12299742e+00]\n"
     ]
    }
   ],
   "source": [
    "# Assuming y_train contains the binary labels (0 or 1)\n",
    "class_weights = compute_class_weight('balanced', classes=cc, y=y_train)\n",
    "print(class_weights)\n",
    "# Create a dictionary mapping class indices to their respective weights\n",
    "class_weight_dict = {}\n",
    "\n",
    "for i in range(len(cc)):\n",
    "    class_weight_dict[cc[i]] = class_weights[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "def custom_loss(y_true, y_pred):\n",
    "\n",
    "    mse = tf.keras.losses.mean_squared_error(y_true, y_pred)\n",
    "\n",
    "    # Increase importance for errors when true value is close to 1\n",
    "    weight_factor = 100.0\n",
    "    weights = tf.where(tf.math.abs(y_true**2 - 1) < 0.1, weight_factor, 1.0)\n",
    "    \n",
    "    # Apply weights to the MSE\n",
    "    weighted_mse = mse * weights\n",
    "\n",
    "    return tf.reduce_mean(weighted_mse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer='adam', loss=custom_loss, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "620/620 [==============================] - 67s 102ms/step - loss: 0.1057 - accuracy: 0.7981 - val_loss: 0.0977 - val_accuracy: 0.8003\n",
      "Epoch 2/10\n",
      "620/620 [==============================] - 62s 101ms/step - loss: 0.1059 - accuracy: 0.7981 - val_loss: 0.0976 - val_accuracy: 0.8003\n",
      "Epoch 3/10\n",
      "620/620 [==============================] - 63s 101ms/step - loss: 0.1117 - accuracy: 0.7981 - val_loss: 0.0977 - val_accuracy: 0.8003\n",
      "Epoch 4/10\n",
      "620/620 [==============================] - 64s 103ms/step - loss: 0.1050 - accuracy: 0.7981 - val_loss: 0.0977 - val_accuracy: 0.8003\n",
      "Epoch 5/10\n",
      "620/620 [==============================] - 64s 103ms/step - loss: 0.1078 - accuracy: 0.7981 - val_loss: 0.0977 - val_accuracy: 0.8003\n",
      "Epoch 6/10\n",
      "620/620 [==============================] - 64s 103ms/step - loss: 0.1044 - accuracy: 0.7981 - val_loss: 0.0976 - val_accuracy: 0.8003\n",
      "Epoch 7/10\n",
      "305/620 [=============>................] - ETA: 30s - loss: 0.1197 - accuracy: 0.7996"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[77], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m his \u001b[38;5;241m=\u001b[39m \u001b[43mmodel\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mfit\u001b[49m\u001b[43m(\u001b[49m\u001b[43mX_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my_train\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mepochs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m10\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mbatch_size\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m32\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mvalidation_data\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mX_test\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my_test\u001b[49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/keras/src/utils/traceback_utils.py:65\u001b[0m, in \u001b[0;36mfilter_traceback.<locals>.error_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m filtered_tb \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m     64\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m---> 65\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     66\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m e:\n\u001b[1;32m     67\u001b[0m     filtered_tb \u001b[38;5;241m=\u001b[39m _process_traceback_frames(e\u001b[38;5;241m.\u001b[39m__traceback__)\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/keras/src/engine/training.py:1783\u001b[0m, in \u001b[0;36mModel.fit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1775\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m tf\u001b[38;5;241m.\u001b[39mprofiler\u001b[38;5;241m.\u001b[39mexperimental\u001b[38;5;241m.\u001b[39mTrace(\n\u001b[1;32m   1776\u001b[0m     \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mtrain\u001b[39m\u001b[38;5;124m\"\u001b[39m,\n\u001b[1;32m   1777\u001b[0m     epoch_num\u001b[38;5;241m=\u001b[39mepoch,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m   1780\u001b[0m     _r\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m1\u001b[39m,\n\u001b[1;32m   1781\u001b[0m ):\n\u001b[1;32m   1782\u001b[0m     callbacks\u001b[38;5;241m.\u001b[39mon_train_batch_begin(step)\n\u001b[0;32m-> 1783\u001b[0m     tmp_logs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mtrain_function\u001b[49m\u001b[43m(\u001b[49m\u001b[43miterator\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1784\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m data_handler\u001b[38;5;241m.\u001b[39mshould_sync:\n\u001b[1;32m   1785\u001b[0m         context\u001b[38;5;241m.\u001b[39masync_wait()\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/util/traceback_utils.py:150\u001b[0m, in \u001b[0;36mfilter_traceback.<locals>.error_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    148\u001b[0m filtered_tb \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m    149\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m--> 150\u001b[0m   \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfn\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    151\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m \u001b[38;5;28;01mas\u001b[39;00m e:\n\u001b[1;32m    152\u001b[0m   filtered_tb \u001b[38;5;241m=\u001b[39m _process_traceback_frames(e\u001b[38;5;241m.\u001b[39m__traceback__)\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:831\u001b[0m, in \u001b[0;36mFunction.__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    828\u001b[0m compiler \u001b[38;5;241m=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mxla\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_jit_compile \u001b[38;5;28;01melse\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mnonXla\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m    830\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m OptionalXlaContext(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_jit_compile):\n\u001b[0;32m--> 831\u001b[0m   result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwds\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    833\u001b[0m new_tracing_count \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexperimental_get_tracing_count()\n\u001b[1;32m    834\u001b[0m without_tracing \u001b[38;5;241m=\u001b[39m (tracing_count \u001b[38;5;241m==\u001b[39m new_tracing_count)\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:867\u001b[0m, in \u001b[0;36mFunction._call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    864\u001b[0m   \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_lock\u001b[38;5;241m.\u001b[39mrelease()\n\u001b[1;32m    865\u001b[0m   \u001b[38;5;66;03m# In this case we have created variables on the first call, so we run the\u001b[39;00m\n\u001b[1;32m    866\u001b[0m   \u001b[38;5;66;03m# defunned version which is guaranteed to never create variables.\u001b[39;00m\n\u001b[0;32m--> 867\u001b[0m   \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mtracing_compilation\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcall_function\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    868\u001b[0m \u001b[43m      \u001b[49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mkwds\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_no_variable_creation_config\u001b[49m\n\u001b[1;32m    869\u001b[0m \u001b[43m  \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    870\u001b[0m \u001b[38;5;28;01melif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_variable_creation_config \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[1;32m    871\u001b[0m   \u001b[38;5;66;03m# Release the lock early so that multiple threads can perform the call\u001b[39;00m\n\u001b[1;32m    872\u001b[0m   \u001b[38;5;66;03m# in parallel.\u001b[39;00m\n\u001b[1;32m    873\u001b[0m   \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_lock\u001b[38;5;241m.\u001b[39mrelease()\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:139\u001b[0m, in \u001b[0;36mcall_function\u001b[0;34m(args, kwargs, tracing_options)\u001b[0m\n\u001b[1;32m    137\u001b[0m bound_args \u001b[38;5;241m=\u001b[39m function\u001b[38;5;241m.\u001b[39mfunction_type\u001b[38;5;241m.\u001b[39mbind(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs)\n\u001b[1;32m    138\u001b[0m flat_inputs \u001b[38;5;241m=\u001b[39m function\u001b[38;5;241m.\u001b[39mfunction_type\u001b[38;5;241m.\u001b[39munpack_inputs(bound_args)\n\u001b[0;32m--> 139\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mfunction\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_flat\u001b[49m\u001b[43m(\u001b[49m\u001b[43m  \u001b[49m\u001b[38;5;66;43;03m# pylint: disable=protected-access\u001b[39;49;00m\n\u001b[1;32m    140\u001b[0m \u001b[43m    \u001b[49m\u001b[43mflat_inputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcaptured_inputs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mfunction\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcaptured_inputs\u001b[49m\n\u001b[1;32m    141\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/concrete_function.py:1264\u001b[0m, in \u001b[0;36mConcreteFunction._call_flat\u001b[0;34m(self, tensor_inputs, captured_inputs)\u001b[0m\n\u001b[1;32m   1260\u001b[0m possible_gradient_type \u001b[38;5;241m=\u001b[39m gradients_util\u001b[38;5;241m.\u001b[39mPossibleTapeGradientTypes(args)\n\u001b[1;32m   1261\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m (possible_gradient_type \u001b[38;5;241m==\u001b[39m gradients_util\u001b[38;5;241m.\u001b[39mPOSSIBLE_GRADIENT_TYPES_NONE\n\u001b[1;32m   1262\u001b[0m     \u001b[38;5;129;01mand\u001b[39;00m executing_eagerly):\n\u001b[1;32m   1263\u001b[0m   \u001b[38;5;66;03m# No tape is watching; skip to running the function.\u001b[39;00m\n\u001b[0;32m-> 1264\u001b[0m   \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_inference_function\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mflat_call\u001b[49m\u001b[43m(\u001b[49m\u001b[43margs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1265\u001b[0m forward_backward \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_select_forward_and_backward_functions(\n\u001b[1;32m   1266\u001b[0m     args,\n\u001b[1;32m   1267\u001b[0m     possible_gradient_type,\n\u001b[1;32m   1268\u001b[0m     executing_eagerly)\n\u001b[1;32m   1269\u001b[0m forward_function, args_with_tangents \u001b[38;5;241m=\u001b[39m forward_backward\u001b[38;5;241m.\u001b[39mforward()\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/atomic_function.py:217\u001b[0m, in \u001b[0;36mAtomicFunction.flat_call\u001b[0;34m(self, args)\u001b[0m\n\u001b[1;32m    215\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mflat_call\u001b[39m(\u001b[38;5;28mself\u001b[39m, args: Sequence[core\u001b[38;5;241m.\u001b[39mTensor]) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Any:\n\u001b[1;32m    216\u001b[0m   \u001b[38;5;124;03m\"\"\"Calls with tensor inputs and returns the structured output.\"\"\"\u001b[39;00m\n\u001b[0;32m--> 217\u001b[0m   flat_outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    218\u001b[0m   \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mfunction_type\u001b[38;5;241m.\u001b[39mpack_output(flat_outputs)\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/polymorphic_function/atomic_function.py:252\u001b[0m, in \u001b[0;36mAtomicFunction.__call__\u001b[0;34m(self, *args)\u001b[0m\n\u001b[1;32m    250\u001b[0m \u001b[38;5;28;01mwith\u001b[39;00m record\u001b[38;5;241m.\u001b[39mstop_recording():\n\u001b[1;32m    251\u001b[0m   \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_bound_context\u001b[38;5;241m.\u001b[39mexecuting_eagerly():\n\u001b[0;32m--> 252\u001b[0m     outputs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_bound_context\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcall_function\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m    253\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mname\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    254\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;28;43mlist\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43margs\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    255\u001b[0m \u001b[43m        \u001b[49m\u001b[38;5;28;43mlen\u001b[39;49m\u001b[43m(\u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mfunction_type\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mflat_outputs\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    256\u001b[0m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    257\u001b[0m   \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    258\u001b[0m     outputs \u001b[38;5;241m=\u001b[39m make_call_op_in_graph(\n\u001b[1;32m    259\u001b[0m         \u001b[38;5;28mself\u001b[39m,\n\u001b[1;32m    260\u001b[0m         \u001b[38;5;28mlist\u001b[39m(args),\n\u001b[1;32m    261\u001b[0m         \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_bound_context\u001b[38;5;241m.\u001b[39mfunction_call_options\u001b[38;5;241m.\u001b[39mas_attrs(),\n\u001b[1;32m    262\u001b[0m     )\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/context.py:1479\u001b[0m, in \u001b[0;36mContext.call_function\u001b[0;34m(self, name, tensor_inputs, num_outputs)\u001b[0m\n\u001b[1;32m   1477\u001b[0m cancellation_context \u001b[38;5;241m=\u001b[39m cancellation\u001b[38;5;241m.\u001b[39mcontext()\n\u001b[1;32m   1478\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m cancellation_context \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[0;32m-> 1479\u001b[0m   outputs \u001b[38;5;241m=\u001b[39m \u001b[43mexecute\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mexecute\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m   1480\u001b[0m \u001b[43m      \u001b[49m\u001b[43mname\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdecode\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mutf-8\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1481\u001b[0m \u001b[43m      \u001b[49m\u001b[43mnum_outputs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mnum_outputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1482\u001b[0m \u001b[43m      \u001b[49m\u001b[43minputs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mtensor_inputs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1483\u001b[0m \u001b[43m      \u001b[49m\u001b[43mattrs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mattrs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1484\u001b[0m \u001b[43m      \u001b[49m\u001b[43mctx\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m   1485\u001b[0m \u001b[43m  \u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   1486\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m   1487\u001b[0m   outputs \u001b[38;5;241m=\u001b[39m execute\u001b[38;5;241m.\u001b[39mexecute_with_cancellation(\n\u001b[1;32m   1488\u001b[0m       name\u001b[38;5;241m.\u001b[39mdecode(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mutf-8\u001b[39m\u001b[38;5;124m\"\u001b[39m),\n\u001b[1;32m   1489\u001b[0m       num_outputs\u001b[38;5;241m=\u001b[39mnum_outputs,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m   1493\u001b[0m       cancellation_manager\u001b[38;5;241m=\u001b[39mcancellation_context,\n\u001b[1;32m   1494\u001b[0m   )\n",
      "File \u001b[0;32m~/.local/lib/python3.10/site-packages/tensorflow/python/eager/execute.py:60\u001b[0m, in \u001b[0;36mquick_execute\u001b[0;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[1;32m     53\u001b[0m   \u001b[38;5;66;03m# Convert any objects of type core_types.Tensor to Tensor.\u001b[39;00m\n\u001b[1;32m     54\u001b[0m   inputs \u001b[38;5;241m=\u001b[39m [\n\u001b[1;32m     55\u001b[0m       tensor_conversion_registry\u001b[38;5;241m.\u001b[39mconvert(t)\n\u001b[1;32m     56\u001b[0m       \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28misinstance\u001b[39m(t, core_types\u001b[38;5;241m.\u001b[39mTensor)\n\u001b[1;32m     57\u001b[0m       \u001b[38;5;28;01melse\u001b[39;00m t\n\u001b[1;32m     58\u001b[0m       \u001b[38;5;28;01mfor\u001b[39;00m t \u001b[38;5;129;01min\u001b[39;00m inputs\n\u001b[1;32m     59\u001b[0m   ]\n\u001b[0;32m---> 60\u001b[0m   tensors \u001b[38;5;241m=\u001b[39m \u001b[43mpywrap_tfe\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mTFE_Py_Execute\u001b[49m\u001b[43m(\u001b[49m\u001b[43mctx\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_handle\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mdevice_name\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mop_name\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     61\u001b[0m \u001b[43m                                      \u001b[49m\u001b[43minputs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mattrs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mnum_outputs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     62\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m core\u001b[38;5;241m.\u001b[39m_NotOkStatusException \u001b[38;5;28;01mas\u001b[39;00m e:\n\u001b[1;32m     63\u001b[0m   \u001b[38;5;28;01mif\u001b[39;00m name \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "his = model.fit(X_train, y_train, epochs=10, batch_size=32, validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://www.kaggle.com/danbrice/keras-plot-history-full-report-and-grid-search\n",
    "def plot_history(history):\n",
    "    loss_list = [s for s in history.history.keys() if 'loss' in s and 'val' not in s]\n",
    "    val_loss_list = [s for s in history.history.keys() if 'loss' in s and 'val' in s]\n",
    "    acc_list = [s for s in history.history.keys() if 'acc' in s and 'val' not in s]\n",
    "    val_acc_list = [s for s in history.history.keys() if 'acc' in s and 'val' in s]\n",
    "    \n",
    "    if len(loss_list) == 0:\n",
    "        print('Loss is missing in history')\n",
    "        return \n",
    "    \n",
    "    ## As loss always exists\n",
    "    epochs = range(1,len(history.history[loss_list[0]]) + 1)\n",
    "    ## Loss\n",
    "    plt.figure(1)\n",
    "    for l in loss_list:\n",
    "        plt.plot(epochs, history.history[l], 'b', label='Training loss (' + str(str(format(history.history[l][-1],'.5f'))+')'))\n",
    "    for l in val_loss_list:\n",
    "        plt.plot(epochs, history.history[l], 'g', label='Validation loss (' + str(str(format(history.history[l][-1],'.5f'))+')'))\n",
    "    \n",
    "    plt.title('Loss')\n",
    "    plt.xlabel('Epochs')\n",
    "    plt.ylabel('Loss')\n",
    "    plt.legend()\n",
    "    \n",
    "    ## Accuracy\n",
    "    plt.figure(2)\n",
    "    for l in acc_list:\n",
    "        plt.plot(epochs, history.history[l], 'b', label='Training accuracy (' + str(format(history.history[l][-1],'.5f'))+')')\n",
    "    for l in val_acc_list:    \n",
    "        plt.plot(epochs, history.history[l], 'g', label='Validation accuracy (' + str(format(history.history[l][-1],'.5f'))+')')\n",
    "\n",
    "    plt.title('Accuracy')\n",
    "    plt.xlabel('Epochs')\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.legend()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(his)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.save('./models/hate-v2.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Test from scratch**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sentence = 'roses are red, sun is yellow'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess(sentence, get='stems'):\n",
    "    sent = re.sub(r'@([a-zA-Z0-9_]+)', 'username', sentence).replace('username:', '') # replace first username\n",
    "    sent = re.sub(r'http?://\\S+', 'weblink', sent)\n",
    "    sent = re.sub(r'&amp', '&', sent)\n",
    "\n",
    "    if sent and sent[0] == '.':\n",
    "        sent = sent[1:]\n",
    "\n",
    "    remove_points(sent)\n",
    "    sent = sent   .replace('RT', '').replace('!', ' ').replace('\"', '').replace(\"\\n\", ' ')\\\n",
    "                        .replace(';', ' ').replace('-', ' ').replace(' and ', ' & ').replace('\\'', '')\\\n",
    "                        .replace('?', '.').replace(',', '').replace('~', ' ').replace('|', ' ').replace('°', ' ')\\\n",
    "                        .replace('`', ' ').replace('~', ' ').replace('*', ' ').replace('+', ' ').replace('/', ' ')\\\n",
    "                        .replace(' # ', ' ').replace('http', ' ').replace('t.co', ' ').replace('\\\\', ' ').replace('&#', ' ')\n",
    "    for _ in range(4):\n",
    "        sent = sent.replace('  ', ' ').replace('..', '.').replace(' .', '.') # remove multiple points & space\n",
    "    \n",
    "    if sent and sent[0] == ' ':\n",
    "        sent = sent[1:]\n",
    "\n",
    "    sent = sent.lower()\n",
    "\n",
    "    for old in abbr:\n",
    "        new = abbr[old]\n",
    "        sent = sent.replace( ' ' + old + ' ', ' ' + new + ' ' ) # add some space arround the world to avoid matching a part of a word\n",
    "\n",
    "    sent = reduce_repetition(sent)\n",
    "\n",
    "    if get == 'sentence':\n",
    "        return sent\n",
    "    \n",
    "    tokens = nltk.word_tokenize(sent)\n",
    "\n",
    "    for i in range(len(tokens) - 1, 0, -1):\n",
    "        if len(tokens[i].strip()) == 0 or (len(tokens[i].strip()) == 1 and tokens[i].strip() != 'a' and tokens[i].strip() != 'i' and tokens[i].strip() != '&'):\n",
    "            tokens.pop(i)\n",
    "\n",
    "    if get == 'tokens':\n",
    "        return tokens\n",
    "\n",
    "    stems = []\n",
    "\n",
    "    for tok in tokens:\n",
    "        stems.append(stemmer.stem(tok))\n",
    "\n",
    "    for i in range(len(stems) - 1, 0, -1):\n",
    "        if len(stems[i].strip()) == 0:\n",
    "            stems.pop(i)\n",
    "        elif len(stems[i].strip()) == 1:\n",
    "            if 'a' < stems[i].strip() <= 'z'and stems[i].strip() != 'i':\n",
    "                stems.pop(i)\n",
    "\n",
    "\n",
    "    for i in range(len(stems) - 1, 0, -1):\n",
    "        arr = stems[i].split('.')\n",
    "        stems[i] = arr[0]\n",
    "        for j in range(1, len(arr)):\n",
    "            stems.insert(i+j, arr[j])\n",
    "\n",
    "    for i in range(len(stems)):\n",
    "        if stems[i] in abbr.keys():\n",
    "            stems[i] = abbr[stems[i]]\n",
    "\n",
    "    return stems"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['rose', 'are', 'red', 'sun', 'is', 'yellow']\n"
     ]
    }
   ],
   "source": [
    "print(preprocess(sentence))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def encode_for_model(stems):\n",
    "    row  = []\n",
    "    code = []\n",
    "    for s in stems:\n",
    "        code.append(enc(s))\n",
    "    for c in code:\n",
    "        row.append(c)\n",
    "    elements_to_add = INPUT - len(row)\n",
    "    \n",
    "    for _ in range(elements_to_add):\n",
    "        row.append(0)\n",
    "\n",
    "    return np.array(row)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "c = encode_for_model(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def prediction(sentence:str=None, code=None):\n",
    "    if sentence is None and code is None:\n",
    "        print(\"Error, function prediction waits at least a sentence or its code\")\n",
    "    if code is None:\n",
    "        code = encode_for_model(preprocess(sentence))\n",
    "    code_reshaped = np.reshape(code, (1, INPUT))\n",
    "    predictions = model.predict(code_reshaped, verbose = 0)\n",
    "\n",
    "    return predictions[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0859596"
      ]
     },
     "execution_count": 194,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction(code=c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 100\n",
      "8 33\n",
      "8 16\n",
      "8 33\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 100\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 100\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 50\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 25\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 66\n",
      "8 33\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 16\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 66\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 100\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8 33\n",
      "8 33\n",
      "8 0\n",
      "8 0\n",
      "8 0\n",
      "8.595960587263107\n"
     ]
    }
   ],
   "source": [
    "m = 0\n",
    "for t in Tweets[:500]:\n",
    "    print(int(prediction(t.tweet)*100), int(t.hateness()*100))\n",
    "    if m < prediction(t.tweet)*100:\n",
    "        m = prediction(t.tweet)*100\n",
    "print( m )"
   ]
  }
 ],
 "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}