m3_assignment5.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/francescapoli98/pattern-recognition-systems/blob/main/m3_assignment5.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W_l90Obuzyy4"
      },
      "source": [
        "# **Intelligent Systems for Pattern Recognition: midterm 3**\n",
        "#### Francesca Poli [560190]\n",
        "\n",
        "\n",
        "---\n",
        "\n",
        "\n",
        "## Fake News Classification\n",
        "\n",
        "\n",
        " **Assignment 5:**\n",
        "The [dataset](https://www.kaggle.com/datasets/clmentbisaillon/fake-and-real-news-dataset) contains real and fake news, including their *title, text, subject, and date.*\n",
        "\n",
        "The objective is to **train a binary classifier to recognize fake news**. You are free to choose the model's architecture, but you should describe and justify your design choices.\n",
        "\n",
        "Notice that the fake and real news in the dataset are balanced. However, in the real world, real news are much more frequent than fake ones (hopefully).\n",
        "\n",
        "Simulate the effect of the data imbalance by undersampling/oversampling one of the classes in the training set and compute the test accuracy on a (balanced) test set.\n",
        "\n",
        "Then, try to use a mechanism to make the training robust to imbalances, such as weighting the loss for the samples depending on their class. Discuss the results of this mitigation.\n",
        "\n",
        "NOTE ON *PREPROCESSING*: you are allowed to preprocess the data however you want (e.g. using pretrained embeddings, dropping some features, just a bag-of-words), but the classification model must be trained by yourself."
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Binary Classifier"
      ],
      "metadata": {
        "id": "2whpYR7NQbgL"
      }
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dtegj_Um0NMi"
      },
      "source": [
        "## Work setup"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LGYonMno0ZeX",
        "outputId": "87a5aae3-6627-4c59-c43b-1498a3f7c602"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ttNqMWpWZjXB"
      },
      "outputs": [],
      "source": [
        "#some useful libraries\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import os\n",
        "\n",
        "#no warnings to distract us (checked before inserting the ignore filter)\n",
        "import warnings\n",
        "warnings.filterwarnings(\"ignore\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vXaboxtFZrUi"
      },
      "source": [
        "#### Loading of the dataset and work setup"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7EU1HpIzUH0i"
      },
      "outputs": [],
      "source": [
        "fake_news = pd.read_csv('/content/drive/MyDrive/ISPR/midterm 3 FAKE NEWS DET/archive/Fake.csv', header=None)\n",
        "true_news = pd.read_csv('/content/drive/MyDrive/ISPR/midterm 3 FAKE NEWS DET/archive/True.csv', header=None)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "headers = fake_news.iloc[0]\n",
        "fake_news  = pd.DataFrame(fake_news.values[1:], columns=headers)"
      ],
      "metadata": {
        "id": "UdRsNMHUwaVT"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "headers = true_news.iloc[0]\n",
        "true_news  = pd.DataFrame(true_news.values[1:], columns=headers)"
      ],
      "metadata": {
        "id": "KlL9Qd3NwcBe"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "del fake_news['date']\n",
        "del true_news['date']"
      ],
      "metadata": {
        "id": "MsQS-m4WKWk0"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "fake_news.info()\n",
        "true_news.info()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "q5AsXpz4KrEy",
        "outputId": "c2a12e67-00eb-4b9d-c32e-c390428ba0b7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 23481 entries, 0 to 23480\n",
            "Data columns (total 3 columns):\n",
            " #   Column   Non-Null Count  Dtype \n",
            "---  ------   --------------  ----- \n",
            " 0   title    23481 non-null  object\n",
            " 1   text     23481 non-null  object\n",
            " 2   subject  23481 non-null  object\n",
            "dtypes: object(3)\n",
            "memory usage: 550.5+ KB\n",
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 21417 entries, 0 to 21416\n",
            "Data columns (total 3 columns):\n",
            " #   Column   Non-Null Count  Dtype \n",
            "---  ------   --------------  ----- \n",
            " 0   title    21417 non-null  object\n",
            " 1   text     21417 non-null  object\n",
            " 2   subject  21417 non-null  object\n",
            "dtypes: object(3)\n",
            "memory usage: 502.1+ KB\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fake_news"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "oEqq_SCVKzKY",
        "outputId": "875837fd-47c4-4ad6-db8f-1a298648e4e3"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "0       Donald Trump Sends Out Embarrassing New Year’...   \n",
              "1       Drunk Bragging Trump Staffer Started Russian ...   \n",
              "2       Sheriff David Clarke Becomes An Internet Joke...   \n",
              "3       Trump Is So Obsessed He Even Has Obama’s Name...   \n",
              "4       Pope Francis Just Called Out Donald Trump Dur...   \n",
              "...                                                  ...   \n",
              "23476  McPain: John McCain Furious That Iran Treated ...   \n",
              "23477  JUSTICE? Yahoo Settles E-mail Privacy Class-ac...   \n",
              "23478  Sunnistan: US and Allied ‘Safe Zone’ Plan to T...   \n",
              "23479  How to Blow $700 Million: Al Jazeera America F...   \n",
              "23480  10 U.S. Navy Sailors Held by Iranian Military ...   \n",
              "\n",
              "0                                                   text      subject  \n",
              "0      Donald Trump just couldn t wish all Americans ...         News  \n",
              "1      House Intelligence Committee Chairman Devin Nu...         News  \n",
              "2      On Friday, it was revealed that former Milwauk...         News  \n",
              "3      On Christmas day, Donald Trump announced that ...         News  \n",
              "4      Pope Francis used his annual Christmas Day mes...         News  \n",
              "...                                                  ...          ...  \n",
              "23476  21st Century Wire says As 21WIRE reported earl...  Middle-east  \n",
              "23477  21st Century Wire says It s a familiar theme. ...  Middle-east  \n",
              "23478  Patrick Henningsen  21st Century WireRemember ...  Middle-east  \n",
              "23479  21st Century Wire says Al Jazeera America will...  Middle-east  \n",
              "23480  21st Century Wire says As 21WIRE predicted in ...  Middle-east  \n",
              "\n",
              "[23481 rows x 3 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-3556bf0a-b08c-43ab-bab1-a50571411bc2\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Donald Trump Sends Out Embarrassing New Year’...</td>\n",
              "      <td>Donald Trump just couldn t wish all Americans ...</td>\n",
              "      <td>News</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Drunk Bragging Trump Staffer Started Russian ...</td>\n",
              "      <td>House Intelligence Committee Chairman Devin Nu...</td>\n",
              "      <td>News</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Sheriff David Clarke Becomes An Internet Joke...</td>\n",
              "      <td>On Friday, it was revealed that former Milwauk...</td>\n",
              "      <td>News</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Trump Is So Obsessed He Even Has Obama’s Name...</td>\n",
              "      <td>On Christmas day, Donald Trump announced that ...</td>\n",
              "      <td>News</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Pope Francis Just Called Out Donald Trump Dur...</td>\n",
              "      <td>Pope Francis used his annual Christmas Day mes...</td>\n",
              "      <td>News</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23476</th>\n",
              "      <td>McPain: John McCain Furious That Iran Treated ...</td>\n",
              "      <td>21st Century Wire says As 21WIRE reported earl...</td>\n",
              "      <td>Middle-east</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23477</th>\n",
              "      <td>JUSTICE? Yahoo Settles E-mail Privacy Class-ac...</td>\n",
              "      <td>21st Century Wire says It s a familiar theme. ...</td>\n",
              "      <td>Middle-east</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23478</th>\n",
              "      <td>Sunnistan: US and Allied ‘Safe Zone’ Plan to T...</td>\n",
              "      <td>Patrick Henningsen  21st Century WireRemember ...</td>\n",
              "      <td>Middle-east</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23479</th>\n",
              "      <td>How to Blow $700 Million: Al Jazeera America F...</td>\n",
              "      <td>21st Century Wire says Al Jazeera America will...</td>\n",
              "      <td>Middle-east</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23480</th>\n",
              "      <td>10 U.S. Navy Sailors Held by Iranian Military ...</td>\n",
              "      <td>21st Century Wire says As 21WIRE predicted in ...</td>\n",
              "      <td>Middle-east</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>23481 rows × 3 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-3556bf0a-b08c-43ab-bab1-a50571411bc2')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-3556bf0a-b08c-43ab-bab1-a50571411bc2 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-3556bf0a-b08c-43ab-bab1-a50571411bc2');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "true_news"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "EXNh_p_dKuj0",
        "outputId": "917c9a73-eb40-4e2d-86fb-e78ca608179a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "0      As U.S. budget fight looms, Republicans flip t...   \n",
              "1      U.S. military to accept transgender recruits o...   \n",
              "2      Senior U.S. Republican senator: 'Let Mr. Muell...   \n",
              "3      FBI Russia probe helped by Australian diplomat...   \n",
              "4      Trump wants Postal Service to charge 'much mor...   \n",
              "...                                                  ...   \n",
              "21412  'Fully committed' NATO backs new U.S. approach...   \n",
              "21413  LexisNexis withdrew two products from Chinese ...   \n",
              "21414  Minsk cultural hub becomes haven from authorities   \n",
              "21415  Vatican upbeat on possibility of Pope Francis ...   \n",
              "21416  Indonesia to buy $1.14 billion worth of Russia...   \n",
              "\n",
              "0                                                   text       subject  \n",
              "0      WASHINGTON (Reuters) - The head of a conservat...  politicsNews  \n",
              "1      WASHINGTON (Reuters) - Transgender people will...  politicsNews  \n",
              "2      WASHINGTON (Reuters) - The special counsel inv...  politicsNews  \n",
              "3      WASHINGTON (Reuters) - Trump campaign adviser ...  politicsNews  \n",
              "4      SEATTLE/WASHINGTON (Reuters) - President Donal...  politicsNews  \n",
              "...                                                  ...           ...  \n",
              "21412  BRUSSELS (Reuters) - NATO allies on Tuesday we...     worldnews  \n",
              "21413  LONDON (Reuters) - LexisNexis, a provider of l...     worldnews  \n",
              "21414  MINSK (Reuters) - In the shadow of disused Sov...     worldnews  \n",
              "21415  MOSCOW (Reuters) - Vatican Secretary of State ...     worldnews  \n",
              "21416  JAKARTA (Reuters) - Indonesia will buy 11 Sukh...     worldnews  \n",
              "\n",
              "[21417 rows x 3 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-5580c9d3-3199-4d06-b82d-1bde5183e833\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>As U.S. budget fight looms, Republicans flip t...</td>\n",
              "      <td>WASHINGTON (Reuters) - The head of a conservat...</td>\n",
              "      <td>politicsNews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>U.S. military to accept transgender recruits o...</td>\n",
              "      <td>WASHINGTON (Reuters) - Transgender people will...</td>\n",
              "      <td>politicsNews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Senior U.S. Republican senator: 'Let Mr. Muell...</td>\n",
              "      <td>WASHINGTON (Reuters) - The special counsel inv...</td>\n",
              "      <td>politicsNews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>FBI Russia probe helped by Australian diplomat...</td>\n",
              "      <td>WASHINGTON (Reuters) - Trump campaign adviser ...</td>\n",
              "      <td>politicsNews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Trump wants Postal Service to charge 'much mor...</td>\n",
              "      <td>SEATTLE/WASHINGTON (Reuters) - President Donal...</td>\n",
              "      <td>politicsNews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21412</th>\n",
              "      <td>'Fully committed' NATO backs new U.S. approach...</td>\n",
              "      <td>BRUSSELS (Reuters) - NATO allies on Tuesday we...</td>\n",
              "      <td>worldnews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21413</th>\n",
              "      <td>LexisNexis withdrew two products from Chinese ...</td>\n",
              "      <td>LONDON (Reuters) - LexisNexis, a provider of l...</td>\n",
              "      <td>worldnews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21414</th>\n",
              "      <td>Minsk cultural hub becomes haven from authorities</td>\n",
              "      <td>MINSK (Reuters) - In the shadow of disused Sov...</td>\n",
              "      <td>worldnews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21415</th>\n",
              "      <td>Vatican upbeat on possibility of Pope Francis ...</td>\n",
              "      <td>MOSCOW (Reuters) - Vatican Secretary of State ...</td>\n",
              "      <td>worldnews</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21416</th>\n",
              "      <td>Indonesia to buy $1.14 billion worth of Russia...</td>\n",
              "      <td>JAKARTA (Reuters) - Indonesia will buy 11 Sukh...</td>\n",
              "      <td>worldnews</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>21417 rows × 3 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-5580c9d3-3199-4d06-b82d-1bde5183e833')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-5580c9d3-3199-4d06-b82d-1bde5183e833 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-5580c9d3-3199-4d06-b82d-1bde5183e833');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 53
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Data preprocessing"
      ],
      "metadata": {
        "id": "6o3YTvEmSeuZ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Here are randomly selected a specified fraction of the total number of rows through the *pandas* method `sample()`.\n",
        "Let’s set `frac=0.80` for the fake news in order to get a random selection of the total rows from the dataframe, according to the assessment that:\n",
        "\n",
        "*the fake and real news in the dataset are balanced. However, in the real world, real news are much more frequent than fake ones (hopefully)*"
      ],
      "metadata": {
        "id": "-GRTvBokLbK4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#true_samples = true_news.sample(frac=0.65)\n",
        "true_news['label'] = 0\n",
        "fake_news['label'] = 1\n",
        "true_news"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "KjS_4EqDLMGk",
        "outputId": "1fd6fb99-fa95-401d-be8c-dc6b3c91874d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "0      As U.S. budget fight looms, Republicans flip t...   \n",
              "1      U.S. military to accept transgender recruits o...   \n",
              "2      Senior U.S. Republican senator: 'Let Mr. Muell...   \n",
              "3      FBI Russia probe helped by Australian diplomat...   \n",
              "4      Trump wants Postal Service to charge 'much mor...   \n",
              "...                                                  ...   \n",
              "21412  'Fully committed' NATO backs new U.S. approach...   \n",
              "21413  LexisNexis withdrew two products from Chinese ...   \n",
              "21414  Minsk cultural hub becomes haven from authorities   \n",
              "21415  Vatican upbeat on possibility of Pope Francis ...   \n",
              "21416  Indonesia to buy $1.14 billion worth of Russia...   \n",
              "\n",
              "0                                                   text       subject  label  \n",
              "0      WASHINGTON (Reuters) - The head of a conservat...  politicsNews      0  \n",
              "1      WASHINGTON (Reuters) - Transgender people will...  politicsNews      0  \n",
              "2      WASHINGTON (Reuters) - The special counsel inv...  politicsNews      0  \n",
              "3      WASHINGTON (Reuters) - Trump campaign adviser ...  politicsNews      0  \n",
              "4      SEATTLE/WASHINGTON (Reuters) - President Donal...  politicsNews      0  \n",
              "...                                                  ...           ...    ...  \n",
              "21412  BRUSSELS (Reuters) - NATO allies on Tuesday we...     worldnews      0  \n",
              "21413  LONDON (Reuters) - LexisNexis, a provider of l...     worldnews      0  \n",
              "21414  MINSK (Reuters) - In the shadow of disused Sov...     worldnews      0  \n",
              "21415  MOSCOW (Reuters) - Vatican Secretary of State ...     worldnews      0  \n",
              "21416  JAKARTA (Reuters) - Indonesia will buy 11 Sukh...     worldnews      0  \n",
              "\n",
              "[21417 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-ab0ca5de-988d-4093-b942-b9cdf8ee70dc\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>As U.S. budget fight looms, Republicans flip t...</td>\n",
              "      <td>WASHINGTON (Reuters) - The head of a conservat...</td>\n",
              "      <td>politicsNews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>U.S. military to accept transgender recruits o...</td>\n",
              "      <td>WASHINGTON (Reuters) - Transgender people will...</td>\n",
              "      <td>politicsNews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Senior U.S. Republican senator: 'Let Mr. Muell...</td>\n",
              "      <td>WASHINGTON (Reuters) - The special counsel inv...</td>\n",
              "      <td>politicsNews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>FBI Russia probe helped by Australian diplomat...</td>\n",
              "      <td>WASHINGTON (Reuters) - Trump campaign adviser ...</td>\n",
              "      <td>politicsNews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Trump wants Postal Service to charge 'much mor...</td>\n",
              "      <td>SEATTLE/WASHINGTON (Reuters) - President Donal...</td>\n",
              "      <td>politicsNews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21412</th>\n",
              "      <td>'Fully committed' NATO backs new U.S. approach...</td>\n",
              "      <td>BRUSSELS (Reuters) - NATO allies on Tuesday we...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21413</th>\n",
              "      <td>LexisNexis withdrew two products from Chinese ...</td>\n",
              "      <td>LONDON (Reuters) - LexisNexis, a provider of l...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21414</th>\n",
              "      <td>Minsk cultural hub becomes haven from authorities</td>\n",
              "      <td>MINSK (Reuters) - In the shadow of disused Sov...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21415</th>\n",
              "      <td>Vatican upbeat on possibility of Pope Francis ...</td>\n",
              "      <td>MOSCOW (Reuters) - Vatican Secretary of State ...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21416</th>\n",
              "      <td>Indonesia to buy $1.14 billion worth of Russia...</td>\n",
              "      <td>JAKARTA (Reuters) - Indonesia will buy 11 Sukh...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>21417 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-ab0ca5de-988d-4093-b942-b9cdf8ee70dc')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-ab0ca5de-988d-4093-b942-b9cdf8ee70dc button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-ab0ca5de-988d-4093-b942-b9cdf8ee70dc');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 54
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fake_samples = fake_news.sample(frac=0.60)\n",
        "fake_samples"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "sX7KMcAJC2hu",
        "outputId": "9460f58f-9228-4a6e-f437-dc18d0760eab"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "15137  Colin Powell Jokes About Why He’s Still A Repu...   \n",
              "18340  DONALD TRUMP JR. Releases Emails Related To Ru...   \n",
              "13891  WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...   \n",
              "21972  Reflections on a World Gone Mad and Pushing Ba...   \n",
              "8976    Tennessee GOP Lawmaker Offers To Fund Bundy T...   \n",
              "...                                                  ...   \n",
              "16212  A MUST WATCH! JUDGE NAPOLITANO: “For the first...   \n",
              "21008  MUSLIMS CRITICIZE New State Law That Permanent...   \n",
              "1711    Trump State Campaign Chair In Hot Water For T...   \n",
              "16531  JUDGE JEANINE PIRRO: “We’re tired of being lec...   \n",
              "4225    Close Trump Ally Just ADMITTED He Communicate...   \n",
              "\n",
              "0                                                   text          subject  \\\n",
              "15137  Colin Powell believes he s fooled us all into ...         politics   \n",
              "18340  Donald Trump Jr. took to Twitter today to rele...        left-news   \n",
              "13891  It s interesting how much vitriol some women h...         politics   \n",
              "21972   Andre Vltchek Anti DiplomaticoThe following i...          US_News   \n",
              "8976   Most Republican lawmakers have remained silent...             News   \n",
              "...                                                  ...              ...   \n",
              "16212                                                     Government News   \n",
              "21008  Maybe it s time for Muslim citizens and immigr...        left-news   \n",
              "1711   Donald Trump is himself, a pervert. We all kne...             News   \n",
              "16531  Next stop after BREXIT is the US! Judge Jeanin...  Government News   \n",
              "4225   Trump s supporters, and some in the press, bel...             News   \n",
              "\n",
              "0      label  \n",
              "15137      1  \n",
              "18340      1  \n",
              "13891      1  \n",
              "21972      1  \n",
              "8976       1  \n",
              "...      ...  \n",
              "16212      1  \n",
              "21008      1  \n",
              "1711       1  \n",
              "16531      1  \n",
              "4225       1  \n",
              "\n",
              "[14089 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-21c2ddb2-f4f4-4c22-9719-4f8d2e1df584\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>15137</th>\n",
              "      <td>Colin Powell Jokes About Why He’s Still A Repu...</td>\n",
              "      <td>Colin Powell believes he s fooled us all into ...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18340</th>\n",
              "      <td>DONALD TRUMP JR. Releases Emails Related To Ru...</td>\n",
              "      <td>Donald Trump Jr. took to Twitter today to rele...</td>\n",
              "      <td>left-news</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13891</th>\n",
              "      <td>WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...</td>\n",
              "      <td>It s interesting how much vitriol some women h...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21972</th>\n",
              "      <td>Reflections on a World Gone Mad and Pushing Ba...</td>\n",
              "      <td>Andre Vltchek Anti DiplomaticoThe following i...</td>\n",
              "      <td>US_News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8976</th>\n",
              "      <td>Tennessee GOP Lawmaker Offers To Fund Bundy T...</td>\n",
              "      <td>Most Republican lawmakers have remained silent...</td>\n",
              "      <td>News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16212</th>\n",
              "      <td>A MUST WATCH! JUDGE NAPOLITANO: “For the first...</td>\n",
              "      <td></td>\n",
              "      <td>Government News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21008</th>\n",
              "      <td>MUSLIMS CRITICIZE New State Law That Permanent...</td>\n",
              "      <td>Maybe it s time for Muslim citizens and immigr...</td>\n",
              "      <td>left-news</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1711</th>\n",
              "      <td>Trump State Campaign Chair In Hot Water For T...</td>\n",
              "      <td>Donald Trump is himself, a pervert. We all kne...</td>\n",
              "      <td>News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16531</th>\n",
              "      <td>JUDGE JEANINE PIRRO: “We’re tired of being lec...</td>\n",
              "      <td>Next stop after BREXIT is the US! Judge Jeanin...</td>\n",
              "      <td>Government News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4225</th>\n",
              "      <td>Close Trump Ally Just ADMITTED He Communicate...</td>\n",
              "      <td>Trump s supporters, and some in the press, bel...</td>\n",
              "      <td>News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>14089 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-21c2ddb2-f4f4-4c22-9719-4f8d2e1df584')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-21c2ddb2-f4f4-4c22-9719-4f8d2e1df584 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-21c2ddb2-f4f4-4c22-9719-4f8d2e1df584');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 55
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "news_df = pd.concat([fake_samples, true_news], axis=0)\n",
        "news_df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "kx6Ej6g540J2",
        "outputId": "76ec303d-e7cb-44e9-86cd-1be3a48b769a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "15137  Colin Powell Jokes About Why He’s Still A Repu...   \n",
              "18340  DONALD TRUMP JR. Releases Emails Related To Ru...   \n",
              "13891  WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...   \n",
              "21972  Reflections on a World Gone Mad and Pushing Ba...   \n",
              "8976    Tennessee GOP Lawmaker Offers To Fund Bundy T...   \n",
              "...                                                  ...   \n",
              "21412  'Fully committed' NATO backs new U.S. approach...   \n",
              "21413  LexisNexis withdrew two products from Chinese ...   \n",
              "21414  Minsk cultural hub becomes haven from authorities   \n",
              "21415  Vatican upbeat on possibility of Pope Francis ...   \n",
              "21416  Indonesia to buy $1.14 billion worth of Russia...   \n",
              "\n",
              "0                                                   text    subject  label  \n",
              "15137  Colin Powell believes he s fooled us all into ...   politics      1  \n",
              "18340  Donald Trump Jr. took to Twitter today to rele...  left-news      1  \n",
              "13891  It s interesting how much vitriol some women h...   politics      1  \n",
              "21972   Andre Vltchek Anti DiplomaticoThe following i...    US_News      1  \n",
              "8976   Most Republican lawmakers have remained silent...       News      1  \n",
              "...                                                  ...        ...    ...  \n",
              "21412  BRUSSELS (Reuters) - NATO allies on Tuesday we...  worldnews      0  \n",
              "21413  LONDON (Reuters) - LexisNexis, a provider of l...  worldnews      0  \n",
              "21414  MINSK (Reuters) - In the shadow of disused Sov...  worldnews      0  \n",
              "21415  MOSCOW (Reuters) - Vatican Secretary of State ...  worldnews      0  \n",
              "21416  JAKARTA (Reuters) - Indonesia will buy 11 Sukh...  worldnews      0  \n",
              "\n",
              "[35506 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-2c6cc331-b205-435a-8b49-1d4ab0bb4bed\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>15137</th>\n",
              "      <td>Colin Powell Jokes About Why He’s Still A Repu...</td>\n",
              "      <td>Colin Powell believes he s fooled us all into ...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18340</th>\n",
              "      <td>DONALD TRUMP JR. Releases Emails Related To Ru...</td>\n",
              "      <td>Donald Trump Jr. took to Twitter today to rele...</td>\n",
              "      <td>left-news</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13891</th>\n",
              "      <td>WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...</td>\n",
              "      <td>It s interesting how much vitriol some women h...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21972</th>\n",
              "      <td>Reflections on a World Gone Mad and Pushing Ba...</td>\n",
              "      <td>Andre Vltchek Anti DiplomaticoThe following i...</td>\n",
              "      <td>US_News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8976</th>\n",
              "      <td>Tennessee GOP Lawmaker Offers To Fund Bundy T...</td>\n",
              "      <td>Most Republican lawmakers have remained silent...</td>\n",
              "      <td>News</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21412</th>\n",
              "      <td>'Fully committed' NATO backs new U.S. approach...</td>\n",
              "      <td>BRUSSELS (Reuters) - NATO allies on Tuesday we...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21413</th>\n",
              "      <td>LexisNexis withdrew two products from Chinese ...</td>\n",
              "      <td>LONDON (Reuters) - LexisNexis, a provider of l...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21414</th>\n",
              "      <td>Minsk cultural hub becomes haven from authorities</td>\n",
              "      <td>MINSK (Reuters) - In the shadow of disused Sov...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21415</th>\n",
              "      <td>Vatican upbeat on possibility of Pope Francis ...</td>\n",
              "      <td>MOSCOW (Reuters) - Vatican Secretary of State ...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21416</th>\n",
              "      <td>Indonesia to buy $1.14 billion worth of Russia...</td>\n",
              "      <td>JAKARTA (Reuters) - Indonesia will buy 11 Sukh...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>35506 rows × 4 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-2c6cc331-b205-435a-8b49-1d4ab0bb4bed')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-2c6cc331-b205-435a-8b49-1d4ab0bb4bed button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-2c6cc331-b205-435a-8b49-1d4ab0bb4bed');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Now we can create a single column containing all the useful text that we will process and feed to our binary classifier."
      ],
      "metadata": {
        "id": "PxbaOlNEsGtU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "news_df[\"whole_text\"] = news_df[\"title\"] + news_df[\"text\"]\n",
        "news_df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 519
        },
        "id": "pECQlQtF3gwh",
        "outputId": "585419fb-3604-4b26-8977-ffaf234eea7f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0                                                  title  \\\n",
              "15137  Colin Powell Jokes About Why He’s Still A Repu...   \n",
              "18340  DONALD TRUMP JR. Releases Emails Related To Ru...   \n",
              "13891  WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...   \n",
              "21972  Reflections on a World Gone Mad and Pushing Ba...   \n",
              "8976    Tennessee GOP Lawmaker Offers To Fund Bundy T...   \n",
              "...                                                  ...   \n",
              "21412  'Fully committed' NATO backs new U.S. approach...   \n",
              "21413  LexisNexis withdrew two products from Chinese ...   \n",
              "21414  Minsk cultural hub becomes haven from authorities   \n",
              "21415  Vatican upbeat on possibility of Pope Francis ...   \n",
              "21416  Indonesia to buy $1.14 billion worth of Russia...   \n",
              "\n",
              "0                                                   text    subject  label  \\\n",
              "15137  Colin Powell believes he s fooled us all into ...   politics      1   \n",
              "18340  Donald Trump Jr. took to Twitter today to rele...  left-news      1   \n",
              "13891  It s interesting how much vitriol some women h...   politics      1   \n",
              "21972   Andre Vltchek Anti DiplomaticoThe following i...    US_News      1   \n",
              "8976   Most Republican lawmakers have remained silent...       News      1   \n",
              "...                                                  ...        ...    ...   \n",
              "21412  BRUSSELS (Reuters) - NATO allies on Tuesday we...  worldnews      0   \n",
              "21413  LONDON (Reuters) - LexisNexis, a provider of l...  worldnews      0   \n",
              "21414  MINSK (Reuters) - In the shadow of disused Sov...  worldnews      0   \n",
              "21415  MOSCOW (Reuters) - Vatican Secretary of State ...  worldnews      0   \n",
              "21416  JAKARTA (Reuters) - Indonesia will buy 11 Sukh...  worldnews      0   \n",
              "\n",
              "0                                             whole_text  \n",
              "15137  Colin Powell Jokes About Why He’s Still A Repu...  \n",
              "18340  DONALD TRUMP JR. Releases Emails Related To Ru...  \n",
              "13891  WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...  \n",
              "21972  Reflections on a World Gone Mad and Pushing Ba...  \n",
              "8976    Tennessee GOP Lawmaker Offers To Fund Bundy T...  \n",
              "...                                                  ...  \n",
              "21412  'Fully committed' NATO backs new U.S. approach...  \n",
              "21413  LexisNexis withdrew two products from Chinese ...  \n",
              "21414  Minsk cultural hub becomes haven from authorit...  \n",
              "21415  Vatican upbeat on possibility of Pope Francis ...  \n",
              "21416  Indonesia to buy $1.14 billion worth of Russia...  \n",
              "\n",
              "[35506 rows x 5 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-d6b915e3-8f5e-4fe7-9190-b1bdd77b7710\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>title</th>\n",
              "      <th>text</th>\n",
              "      <th>subject</th>\n",
              "      <th>label</th>\n",
              "      <th>whole_text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>15137</th>\n",
              "      <td>Colin Powell Jokes About Why He’s Still A Repu...</td>\n",
              "      <td>Colin Powell believes he s fooled us all into ...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "      <td>Colin Powell Jokes About Why He’s Still A Repu...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18340</th>\n",
              "      <td>DONALD TRUMP JR. Releases Emails Related To Ru...</td>\n",
              "      <td>Donald Trump Jr. took to Twitter today to rele...</td>\n",
              "      <td>left-news</td>\n",
              "      <td>1</td>\n",
              "      <td>DONALD TRUMP JR. Releases Emails Related To Ru...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13891</th>\n",
              "      <td>WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...</td>\n",
              "      <td>It s interesting how much vitriol some women h...</td>\n",
              "      <td>politics</td>\n",
              "      <td>1</td>\n",
              "      <td>WATCH: FORMER TRUMP GIRLFRIEND Sets The Record...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21972</th>\n",
              "      <td>Reflections on a World Gone Mad and Pushing Ba...</td>\n",
              "      <td>Andre Vltchek Anti DiplomaticoThe following i...</td>\n",
              "      <td>US_News</td>\n",
              "      <td>1</td>\n",
              "      <td>Reflections on a World Gone Mad and Pushing Ba...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8976</th>\n",
              "      <td>Tennessee GOP Lawmaker Offers To Fund Bundy T...</td>\n",
              "      <td>Most Republican lawmakers have remained silent...</td>\n",
              "      <td>News</td>\n",
              "      <td>1</td>\n",
              "      <td>Tennessee GOP Lawmaker Offers To Fund Bundy T...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21412</th>\n",
              "      <td>'Fully committed' NATO backs new U.S. approach...</td>\n",
              "      <td>BRUSSELS (Reuters) - NATO allies on Tuesday we...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "      <td>'Fully committed' NATO backs new U.S. approach...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21413</th>\n",
              "      <td>LexisNexis withdrew two products from Chinese ...</td>\n",
              "      <td>LONDON (Reuters) - LexisNexis, a provider of l...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "      <td>LexisNexis withdrew two products from Chinese ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21414</th>\n",
              "      <td>Minsk cultural hub becomes haven from authorities</td>\n",
              "      <td>MINSK (Reuters) - In the shadow of disused Sov...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "      <td>Minsk cultural hub becomes haven from authorit...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21415</th>\n",
              "      <td>Vatican upbeat on possibility of Pope Francis ...</td>\n",
              "      <td>MOSCOW (Reuters) - Vatican Secretary of State ...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "      <td>Vatican upbeat on possibility of Pope Francis ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21416</th>\n",
              "      <td>Indonesia to buy $1.14 billion worth of Russia...</td>\n",
              "      <td>JAKARTA (Reuters) - Indonesia will buy 11 Sukh...</td>\n",
              "      <td>worldnews</td>\n",
              "      <td>0</td>\n",
              "      <td>Indonesia to buy $1.14 billion worth of Russia...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>35506 rows × 5 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d6b915e3-8f5e-4fe7-9190-b1bdd77b7710')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-d6b915e3-8f5e-4fe7-9190-b1bdd77b7710 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-d6b915e3-8f5e-4fe7-9190-b1bdd77b7710');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Let's also take a quick look at the fake and true data distribution, as to keep an eye on the statistics."
      ],
      "metadata": {
        "id": "BtHX6DY5sUZ2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "neg, pos = np.bincount(news_df['label'])\n",
        "total = neg + pos\n",
        "print('Examples:\\n    Total: {}\\n    Fake mews: {} ({:.2f}% of total)\\n'.format(total, pos, 100 * pos / total))"
      ],
      "metadata": {
        "id": "6lUjb82m73og",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ff13d3b1-27f2-4098-8304-52972cf87b77"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Examples:\n",
            "    Total: 35506\n",
            "    Fake mews: 14089 (39.68% of total)\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "sns.countplot(x=\"label\", data=news_df);\n",
        "print(\"1: Unreliable\")\n",
        "print(\"0: Reliable\")\n",
        "print(\"Distribution of labels:\")\n",
        "print(news_df.label.value_counts());"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 553
        },
        "id": "9-kqq2vEiIYB",
        "outputId": "2439b69c-544d-48f0-f86f-8c4edb71f5cb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1: Unreliable\n",
            "0: Reliable\n",
            "Distribution of labels:\n",
            "0    21417\n",
            "1    14089\n",
            "Name: label, dtype: int64\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "###Data preprocessing\n",
        "During this phase, the text at our disposal will be processed through the following methods:\n",
        "- some data cleaning in order to get rid of errors and detrimental elememts, and to lower all capital letters which are not needed for this task\n",
        "- stop words from the `nltk` library will be dropped\n",
        "- tokenization and lemmatization also through `nltk` methods"
      ],
      "metadata": {
        "id": "zfRAbgCPCzrf"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import re\n",
        "\n",
        "#nltk lib\n",
        "import nltk\n",
        "from nltk import word_tokenize\n",
        "from nltk.stem.wordnet import WordNetLemmatizer\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.corpus import wordnet\n",
        "\n",
        "nltk.download('punkt')\n",
        "nltk.download('wordnet')\n",
        "nltk.download('stopwords')\n",
        "nltk.download('omw-1.4')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OLC9UcGpA21l",
        "outputId": "aa4d4f60-f8c3-429c-c106-6385099cab03"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Package wordnet is already up-to-date!\n",
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n",
            "[nltk_data] Downloading package omw-1.4 to /root/nltk_data...\n",
            "[nltk_data]   Package omw-1.4 is already up-to-date!\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {},
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "stop_words = stopwords.words('english')\n",
        "\n",
        "def clean_text(text):\n",
        " #output = ''\n",
        " text = str(text).replace(r'http[\\w:/\\.]+','') # removing urls\n",
        " text = str(text).replace(r'[^\\.\\w\\s]','') #remove everything but characters and punctuation\n",
        " text = str(text).replace(r'\\.\\.+','.') #replace multiple periods with a single one\n",
        " text = str(text).replace(r'\\.','.') #replace periods with a single one\n",
        " text = str(text).replace(r'\\s\\s+',' ') #replace multiple white space with a single one\n",
        " text = str(text).replace('\\n', '') #removing line breaks\n",
        " text = re.sub(r'[^\\w\\s]',' ',text).lower() #lower texts\n",
        "\n",
        " return text"
      ],
      "metadata": {
        "id": "xqHsI7XP_Hp3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#text cleaning\n",
        "news_df['whole_text'] = news_df['whole_text'].apply(lambda row: clean_text(row))\n",
        "news_df['whole_text']"
      ],
      "metadata": {
        "id": "aXgVLTSRT6M_",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "eb7fc514-b07a-4035-ed80-de80ec569ccb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15137    colin powell jokes about why he s still a repu...\n",
              "18340    donald trump jr  releases emails related to ru...\n",
              "13891    watch  former trump girlfriend sets the record...\n",
              "21972    reflections on a world gone mad and pushing ba...\n",
              "8976      tennessee gop lawmaker offers to fund bundy t...\n",
              "                               ...                        \n",
              "21412     fully committed  nato backs new u s  approach...\n",
              "21413    lexisnexis withdrew two products from chinese ...\n",
              "21414    minsk cultural hub becomes haven from authorit...\n",
              "21415    vatican upbeat on possibility of pope francis ...\n",
              "21416    indonesia to buy  1 14 billion worth of russia...\n",
              "Name: whole_text, Length: 35506, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 62
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#deleting stop words\n",
        "news_df['whole_text'] = news_df['whole_text'].apply(lambda row: ' '.join([word for word in row.split() if word not in stop_words]))\n",
        "news_df['whole_text']"
      ],
      "metadata": {
        "id": "SiclMwq6fSX2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "374b6c30-234c-4190-a75c-e138f6311798"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15137    colin powell jokes still republican funny coli...\n",
              "18340    donald trump jr releases emails related russia...\n",
              "13891    watch former trump girlfriend sets record stra...\n",
              "21972    reflections world gone mad pushing back neocol...\n",
              "8976     tennessee gop lawmaker offers fund bundy terro...\n",
              "                               ...                        \n",
              "21412    fully committed nato backs new u approach afgh...\n",
              "21413    lexisnexis withdrew two products chinese marke...\n",
              "21414    minsk cultural hub becomes authoritiesminsk re...\n",
              "21415    vatican upbeat possibility pope francis visiti...\n",
              "21416    indonesia buy 1 14 billion worth russian jetsj...\n",
              "Name: whole_text, Length: 35506, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#lemmatization\n",
        "lemmatizer = WordNetLemmatizer()\n",
        "news_df['whole_text'] = news_df['whole_text'].apply(lambda row: ' '.join([lemmatizer.lemmatize(word) for word in row.split()]))\n",
        "news_df['whole_text']"
      ],
      "metadata": {
        "id": "OBcNOWEJfQGa",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "093ce01c-5455-4c33-f93e-cf567d749974"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15137    colin powell joke still republican funny colin...\n",
              "18340    donald trump jr release email related russian ...\n",
              "13891    watch former trump girlfriend set record strai...\n",
              "21972    reflection world gone mad pushing back neocolo...\n",
              "8976     tennessee gop lawmaker offer fund bundy terror...\n",
              "                               ...                        \n",
              "21412    fully committed nato back new u approach afgha...\n",
              "21413    lexisnexis withdrew two product chinese market...\n",
              "21414    minsk cultural hub becomes authoritiesminsk re...\n",
              "21415    vatican upbeat possibility pope francis visiti...\n",
              "21416    indonesia buy 1 14 billion worth russian jetsj...\n",
              "Name: whole_text, Length: 35506, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#tokenization\n",
        "news_df['whole_text'] = news_df['whole_text'].apply(lambda row: nltk.word_tokenize(row))\n",
        "news_df['whole_text']"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "x7pFLFY8_Gq-",
        "outputId": "f24ca607-d524-4e93-afe0-9fef8216400e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15137    [colin, powell, joke, still, republican, funny...\n",
              "18340    [donald, trump, jr, release, email, related, r...\n",
              "13891    [watch, former, trump, girlfriend, set, record...\n",
              "21972    [reflection, world, gone, mad, pushing, back, ...\n",
              "8976     [tennessee, gop, lawmaker, offer, fund, bundy,...\n",
              "                               ...                        \n",
              "21412    [fully, committed, nato, back, new, u, approac...\n",
              "21413    [lexisnexis, withdrew, two, product, chinese, ...\n",
              "21414    [minsk, cultural, hub, becomes, authoritiesmin...\n",
              "21415    [vatican, upbeat, possibility, pope, francis, ...\n",
              "21416    [indonesia, buy, 1, 14, billion, worth, russia...\n",
              "Name: whole_text, Length: 35506, dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "news_df.to_csv('/content/drive/MyDrive/ISPR/midterm 3 FAKE NEWS DET/news_df.csv')"
      ],
      "metadata": {
        "id": "K5CrvP6hyCvz"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Pre-trained Word embedding using Glove\n",
        "At this point it is needed to convert text into numbers to feed them to the learning machines.\n",
        "\n",
        "We will use the pre-trained word embeddings of [GloVe](https://nlp.stanford.edu/projects/glove/).\n",
        "\n",
        "We will be converting the text into numbers where each word will be represented by an array of numbers of different length depending upon the glove embedding.\n",
        "\n",
        "The [*Using pre-trained GloVe word embeddings*](https://keras.io/examples/nlp/pretrained_word_embeddings/) example from Keras documentation is taken as one reference.\n"
      ],
      "metadata": {
        "id": "3zsEGU02_XsJ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "news_df = pd.read_csv('/content/drive/MyDrive/ISPR/midterm 3 FAKE NEWS DET/news_df.csv')"
      ],
      "metadata": {
        "id": "zBj4hJTY0ekR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.utils.np_utils import to_categorical\n",
        "\n",
        "from keras.layers import TextVectorization\n",
        "from keras.layers import Dense, Input, Flatten\n",
        "from keras.layers import Conv1D, MaxPooling1D, Embedding, Dropout, Bidirectional\n",
        "from keras.models import Model\n",
        "\n",
        "from keras import regularizers, initializers, optimizers, callbacks\n",
        "from keras.layers import *\n",
        "from keras.models import Sequential\n",
        "\n",
        "from keras.utils import pad_sequences"
      ],
      "metadata": {
        "id": "JLXpKErDNqX_"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "First, we initialize the `Tokenizer` object which is imported from the Keras library as a token. Then we fit the tokenizer on the whole text where each word is assigned a unique number and every word is now represented by a number. After all that, we convert each sentence into a sequence of numbers and we make all the sentences of uniform length through padding them using the `pad_sequences` Keras method and a fixed `maxlen` value."
      ],
      "metadata": {
        "id": "gjFBDH9Du7FQ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Train test validation Split\n",
        "from sklearn.model_selection import train_test_split\n",
        "maxlen=200\n",
        "vocsize=20000\n",
        "\n",
        "x_train, x_test, y_train, y_test = train_test_split(news_df['whole_text'], news_df[\"label\"].values, test_size=0.3, random_state=42)\n",
        "\n",
        "#get validation data\n",
        "x_train, x_val, y_train, y_val = train_test_split(x_train, y_train, test_size=0.3, random_state=42)\n",
        "\n",
        "# create tokenizer object and fit it to dataset\n",
        "tokenizer = Tokenizer(num_words = vocsize,\n",
        "                      oov_token='OOV')\n",
        "\n",
        "tokenizer.fit_on_texts(x_train)\n",
        "tokenizer.fit_on_texts(x_val)\n",
        "tokenizer.fit_on_texts(x_test)\n",
        "\n",
        "# trasform text to sequence for data in train, validation and test sets\n",
        "x_train = tokenizer.texts_to_sequences(x_train)\n",
        "x_val   = tokenizer.texts_to_sequences(x_val)\n",
        "x_test  = tokenizer.texts_to_sequences(x_test)\n",
        "\n",
        "# pad sequence for data in train, validation and test sets\n",
        "x_train = pad_sequences(x_train, padding='post', maxlen=maxlen)\n",
        "x_val   = pad_sequences(x_val, padding='post', maxlen=maxlen)\n",
        "x_test  = pad_sequences(x_test, padding='post', maxlen=maxlen)"
      ],
      "metadata": {
        "id": "TqvXWqn9cbNo"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train.shape, y_train.shape, x_test.shape, y_test.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YOmtLT7ebsev",
        "outputId": "9fd595b8-e7f3-41b4-f95a-cd38f0269ceb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((17397, 200), (17397,), (10652, 200), (10652,))"
            ]
          },
          "metadata": {},
          "execution_count": 70
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# vectorizer = TextVectorization(max_tokens=5000, output_sequence_length=2000)\n",
        "# text_ds = tf.data.Dataset.from_tensor_slices(news_df['whole_text']).batch(128)\n",
        "# vectorizer.adapt(text_ds)"
      ],
      "metadata": {
        "id": "59dCQ9VQNTTv"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!wget http://nlp.stanford.edu/data/glove.6B.zip\n",
        "!unzip glove*.zip"
      ],
      "metadata": {
        "id": "VN0lqAx1LK46",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1ca68fa6-0eeb-4589-c3fe-5556df207e37"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2023-05-13 15:46:37--  http://nlp.stanford.edu/data/glove.6B.zip\n",
            "Resolving nlp.stanford.edu (nlp.stanford.edu)... 171.64.67.140\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://nlp.stanford.edu/data/glove.6B.zip [following]\n",
            "--2023-05-13 15:46:37--  https://nlp.stanford.edu/data/glove.6B.zip\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:443... connected.\n",
            "HTTP request sent, awaiting response... 301 Moved Permanently\n",
            "Location: https://downloads.cs.stanford.edu/nlp/data/glove.6B.zip [following]\n",
            "--2023-05-13 15:46:38--  https://downloads.cs.stanford.edu/nlp/data/glove.6B.zip\n",
            "Resolving downloads.cs.stanford.edu (downloads.cs.stanford.edu)... 171.64.64.22\n",
            "Connecting to downloads.cs.stanford.edu (downloads.cs.stanford.edu)|171.64.64.22|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 862182613 (822M) [application/zip]\n",
            "Saving to: ‘glove.6B.zip.1’\n",
            "\n",
            "glove.6B.zip.1      100%[===================>] 822.24M  5.00MB/s    in 2m 39s  \n",
            "\n",
            "2023-05-13 15:49:17 (5.16 MB/s) - ‘glove.6B.zip.1’ saved [862182613/862182613]\n",
            "\n",
            "Archive:  glove.6B.zip\n",
            "replace glove.6B.50d.txt? [y]es, [n]o, [A]ll, [N]one, [r]ename: "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "path_to_glove = os.path.join(\n",
        "    os.path.expanduser(\"~\"), \"/content/glove.6B.100d.txt\"\n",
        ")"
      ],
      "metadata": {
        "id": "0FMzFsNQNH1X"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Now, let's prepare a corresponding **embedding matrix** that we can use in a Keras **Embedding layer**. It's a simple NumPy matrix where entry at index `i` is the pre-trained vector for the word of index `i` in our vectorizer's vocabulary."
      ],
      "metadata": {
        "id": "0T_bvzpwtygF"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "embedding_dim = 100\n",
        "embeddings_index = {}\n",
        "\n",
        "f = open('glove.6B.100d.txt', encoding='utf-8')\n",
        "for line in f:\n",
        "    values = line.split()\n",
        "    word = values[0]\n",
        "    coefs = np.asarray(values[1:], dtype='float32')\n",
        "    embeddings_index[word] = coefs\n",
        "f.close()\n",
        "\n",
        "print('Found %s word vectors.' % len(embeddings_index))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HOTzpx9PTlLh",
        "outputId": "ac2a6eca-bcd8-4046-91ad-ac16ab4c9dd0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 400000 word vectors.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Prepare embedding matrix\n",
        "embedding_matrix = np.zeros((vocsize, embedding_dim))\n",
        "#for word, i in tokenizer.word_index.items():\n",
        "for i in range(1, vocsize):\n",
        "  word = tokenizer.index_word[i]\n",
        "  embedding_vector = embeddings_index.get(word)\n",
        "  if embedding_vector is not None:\n",
        "        # Words not found in embedding index will be all-zeros.\n",
        "        # This includes the representation for \"padding\" and \"OOV\"\n",
        "      embedding_matrix[i:] = embedding_vector\n",
        "\n",
        "embedding_matrix"
      ],
      "metadata": {
        "id": "B2QGUkJsYj25",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5ba7224f-c10e-4c78-d020-aae367c3ea6d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[ 0.        ,  0.        ,  0.        , ...,  0.        ,\n",
              "         0.        ,  0.        ],\n",
              "       [ 0.        ,  0.        ,  0.        , ...,  0.        ,\n",
              "         0.        ,  0.        ],\n",
              "       [ 0.        ,  0.        ,  0.        , ...,  0.        ,\n",
              "         0.        ,  0.        ],\n",
              "       ...,\n",
              "       [-0.65170002,  0.80484003,  0.048731  , ..., -0.78518999,\n",
              "        -0.23776001,  0.37248001],\n",
              "       [-0.65170002,  0.80484003,  0.048731  , ..., -0.78518999,\n",
              "        -0.23776001,  0.37248001],\n",
              "       [-0.65170002,  0.80484003,  0.048731  , ..., -0.78518999,\n",
              "        -0.23776001,  0.37248001]])"
            ]
          },
          "metadata": {},
          "execution_count": 75
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.initializers import Constant\n",
        "\n",
        "embedding_layer = Embedding(vocsize,\n",
        "                            embedding_dim,\n",
        "                            mask_zero = True,\n",
        "                            #weights=[embedding_matrix],\n",
        "                            embeddings_initializer = Constant(embedding_matrix),\n",
        "                            trainable = True,\n",
        "                            input_length=maxlen)"
      ],
      "metadata": {
        "id": "sRW7NGB-YrfE"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.callbacks import EarlyStopping\n",
        "earlystopping = EarlyStopping(monitor='val_loss', # monitor validation loss\n",
        "                              patience=1,\n",
        "                              restore_best_weights=True)"
      ],
      "metadata": {
        "id": "kXqjWKn3Z4BA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "### LSTM\n",
        "\n",
        "One good reason to use Long Short Term Memory model is that it is effective in memorizing important information.\n",
        "\n",
        "If we look and other non-neural network classification techniques, every single word is classified into one of the categories.\n",
        "\n",
        "This is not the same in LSTM. In LSTM we can use a multiple word string to find out the class to which it belongs. This is very helpful while working with Natural language processing. If we use appropriate layers of embedding and encoding in LSTM, the model will be able to find out the actual meaning in input string and will give the most accurate output class."
      ],
      "metadata": {
        "id": "a0uSpAHmYZZ3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#LSTM model\n",
        "def build_lstm(n_units, lr):\n",
        "\n",
        "  model_LSTM = Sequential()\n",
        "  #model_LSTM.add(Input(shape=(None,), dtype='int32'))\n",
        "  model_LSTM.add(embedding_layer)\n",
        "  model_LSTM.add(LSTM(n_units, return_sequences=True, dropout=0.8, name='lstm_layer'))\n",
        "  model_LSTM.add(Flatten())\n",
        "  model_LSTM.add(Dense(1, activation='sigmoid'))\n",
        "\n",
        "  model_LSTM.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=lr),\n",
        "                     loss='binary_crossentropy',\n",
        "                     metrics = ['accuracy'])\n",
        "\n",
        "  return model_LSTM"
      ],
      "metadata": {
        "id": "6dGKY5-0YbW4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Plot functions"
      ],
      "metadata": {
        "id": "lj7yzfKyNtrq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_loss(history):\n",
        "  %matplotlib inline\n",
        "  loss = history.history['loss']\n",
        "  val_loss = history.history['val_loss']\n",
        "  epochs = range(1, len(loss)+1)\n",
        "  plt.plot(epochs, loss, label='Training loss')\n",
        "  plt.plot(epochs, val_loss, label='Validation loss')\n",
        "  plt.title('Training and validation loss')\n",
        "  plt.xlabel('Epochs')\n",
        "  plt.ylabel('Loss')\n",
        "  plt.legend()\n",
        "  plt.show();"
      ],
      "metadata": {
        "id": "Cdx83aX_ahmb"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_acc(history, epochs):\n",
        "  accuracy = history.history['accuracy']\n",
        "  val_accuracy = history.history['val_accuracy']\n",
        "  plt.plot(epochs, accuracy, label='Training accuracy')\n",
        "  plt.plot(epochs, val_accuracy, label='Validation accuracy')\n",
        "  plt.title('Training and validation accuracy')\n",
        "  plt.ylabel('Accuracy')\n",
        "  plt.xlabel('Epochs')\n",
        "  plt.legend()\n",
        "  plt.show();"
      ],
      "metadata": {
        "id": "YuTaY2kZaorv"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Training LSTM"
      ],
      "metadata": {
        "id": "QshMn2kYNy12"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Here a very small kind of grid search is implemented, iterating through the values of `number of units` for the LSTM layer and the `learning rate` for the optimizer, Adam in our case. Moreover, class weights proportional to the balance state of the data have been added through the ScikitLearn method `compute_class_weight()`.\n",
        "\n",
        "Through this model selection phase, we're hoping to find the most suitable combination of parameters and model structure for out task."
      ],
      "metadata": {
        "id": "JPogKEU8Q0ty"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "params={\n",
        "    'n_units':[10, 30, 50],\n",
        "    'lr': [0.1, 0.01, 0.001]\n",
        "    }\n",
        "\n",
        "\n",
        "from sklearn.utils import class_weight, compute_class_weight\n",
        "class_weights = compute_class_weight(class_weight = \"balanced\",\n",
        "                                     classes = np.unique(y_train),\n",
        "                                     y = y_train)\n",
        "class_weights = dict(zip(np.unique(y_train), class_weights))\n",
        "class_weights\n",
        "\n",
        "histories = []\n",
        "min_loss = 0\n",
        "parameters = {}\n",
        "\n",
        "for i in range(len(params['n_units'])):\n",
        "  for j in range(len(params['lr'])):\n",
        "    model=build_lstm(params['n_units'][i], params['lr'][j])\n",
        "\n",
        "    history_LSTM = model.fit(x_train, y_train,\n",
        "                                epochs = 10,\n",
        "                                batch_size=16,\n",
        "                                validation_data=(x_val, y_val),\n",
        "                                shuffle=True,\n",
        "                                callbacks = [earlystopping],\n",
        "                                class_weight = class_weights)\n",
        "    # convert the history.history dict to a pandas DataFrame:\n",
        "    hist_df = pd.DataFrame(history_LSTM.history)\n",
        "    histories.append(hist_df)\n",
        "    if i==0 and j==0:\n",
        "      min_loss = hist_df['loss'].min()\n",
        "      print('First minimum loss:', min_loss)\n",
        "    else:\n",
        "      if hist_df['loss'].min() < min_loss:\n",
        "        min_loss = hist_df['loss'].min()\n",
        "        parameters[0] = i\n",
        "        parameters[1] = j\n",
        "        print('Minimum loss until now:', min_loss, 'parameters:', parameters)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yUMahHJSuFUG",
        "outputId": "2df3fcca-53c5-4291-9e53-9e576192adbd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 204s 180ms/step - loss: 5.1971 - accuracy: 0.6190 - val_loss: 3.6796 - val_accuracy: 0.7137\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 192s 177ms/step - loss: 6.5297 - accuracy: 0.5849 - val_loss: 4.8057 - val_accuracy: 0.6245\n",
            "First minimum loss: 5.197124481201172\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 184s 165ms/step - loss: 0.7651 - accuracy: 0.6470 - val_loss: 0.4503 - val_accuracy: 0.8049\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 187s 172ms/step - loss: 0.6207 - accuracy: 0.7471 - val_loss: 0.3150 - val_accuracy: 0.8710\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 189s 174ms/step - loss: 0.5852 - accuracy: 0.7741 - val_loss: 0.2869 - val_accuracy: 0.8895\n",
            "Epoch 4/10\n",
            "1088/1088 [==============================] - 189s 174ms/step - loss: 0.5541 - accuracy: 0.7952 - val_loss: 0.3055 - val_accuracy: 0.8883\n",
            "Minimum loss until now: 0.5540860295295715 parameters: {0: 0, 1: 1}\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 196s 176ms/step - loss: 0.6409 - accuracy: 0.6227 - val_loss: 0.3481 - val_accuracy: 0.8495\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 177s 163ms/step - loss: 0.4191 - accuracy: 0.8113 - val_loss: 0.2281 - val_accuracy: 0.9120\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 178s 164ms/step - loss: 0.3471 - accuracy: 0.8472 - val_loss: 0.1999 - val_accuracy: 0.9281\n",
            "Epoch 4/10\n",
            "1088/1088 [==============================] - 177s 163ms/step - loss: 0.3228 - accuracy: 0.8635 - val_loss: 0.1935 - val_accuracy: 0.9317\n",
            "Epoch 5/10\n",
            "1088/1088 [==============================] - 180s 165ms/step - loss: 0.3033 - accuracy: 0.8743 - val_loss: 0.1801 - val_accuracy: 0.9329\n",
            "Epoch 6/10\n",
            "1088/1088 [==============================] - 185s 170ms/step - loss: 0.2832 - accuracy: 0.8825 - val_loss: 0.1781 - val_accuracy: 0.9379\n",
            "Epoch 7/10\n",
            "1088/1088 [==============================] - 188s 173ms/step - loss: 0.2775 - accuracy: 0.8856 - val_loss: 0.1670 - val_accuracy: 0.9403\n",
            "Epoch 8/10\n",
            "1088/1088 [==============================] - 190s 175ms/step - loss: 0.2647 - accuracy: 0.8917 - val_loss: 0.1571 - val_accuracy: 0.9445\n",
            "Epoch 9/10\n",
            "1088/1088 [==============================] - 178s 163ms/step - loss: 0.2649 - accuracy: 0.8924 - val_loss: 0.1541 - val_accuracy: 0.9439\n",
            "Epoch 10/10\n",
            "1088/1088 [==============================] - 185s 170ms/step - loss: 0.2588 - accuracy: 0.8943 - val_loss: 0.1433 - val_accuracy: 0.9486\n",
            "Minimum loss until now: 0.25884440541267395 parameters: {0: 0, 1: 2}\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 236s 211ms/step - loss: 13.2157 - accuracy: 0.6341 - val_loss: 11.0672 - val_accuracy: 0.7495\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 223s 205ms/step - loss: 16.7660 - accuracy: 0.6569 - val_loss: 10.0051 - val_accuracy: 0.7897\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 228s 210ms/step - loss: 16.9219 - accuracy: 0.6806 - val_loss: 10.4971 - val_accuracy: 0.7955\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 217s 195ms/step - loss: 1.1773 - accuracy: 0.6983 - val_loss: 0.6562 - val_accuracy: 0.8695\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 203s 187ms/step - loss: 1.2862 - accuracy: 0.7595 - val_loss: 0.6636 - val_accuracy: 0.8856\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 221s 198ms/step - loss: 0.5393 - accuracy: 0.7166 - val_loss: 0.2214 - val_accuracy: 0.9111\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 213s 196ms/step - loss: 0.3610 - accuracy: 0.8449 - val_loss: 0.1711 - val_accuracy: 0.9356\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 205s 188ms/step - loss: 0.3193 - accuracy: 0.8672 - val_loss: 0.1594 - val_accuracy: 0.9439\n",
            "Epoch 4/10\n",
            "1088/1088 [==============================] - 215s 197ms/step - loss: 0.3086 - accuracy: 0.8735 - val_loss: 0.1541 - val_accuracy: 0.9413\n",
            "Epoch 5/10\n",
            "1088/1088 [==============================] - 215s 198ms/step - loss: 0.3006 - accuracy: 0.8763 - val_loss: 0.1564 - val_accuracy: 0.9458\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 283s 253ms/step - loss: 20.1777 - accuracy: 0.6193 - val_loss: 13.4803 - val_accuracy: 0.7892\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 273s 251ms/step - loss: 25.3138 - accuracy: 0.6593 - val_loss: 12.9845 - val_accuracy: 0.8259\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 272s 250ms/step - loss: 25.8704 - accuracy: 0.6908 - val_loss: 16.6743 - val_accuracy: 0.8116\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 244s 220ms/step - loss: 1.6262 - accuracy: 0.7086 - val_loss: 0.9056 - val_accuracy: 0.8698\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 231s 212ms/step - loss: 1.9045 - accuracy: 0.7665 - val_loss: 0.9708 - val_accuracy: 0.8896\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 231s 207ms/step - loss: 0.5242 - accuracy: 0.7348 - val_loss: 0.2084 - val_accuracy: 0.9250\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 231s 212ms/step - loss: 0.3623 - accuracy: 0.8473 - val_loss: 0.1898 - val_accuracy: 0.9346\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 240s 221ms/step - loss: 0.3311 - accuracy: 0.8669 - val_loss: 0.1791 - val_accuracy: 0.9378\n",
            "Epoch 4/10\n",
            "1088/1088 [==============================] - 231s 213ms/step - loss: 0.3224 - accuracy: 0.8730 - val_loss: 0.1716 - val_accuracy: 0.9441\n",
            "Epoch 5/10\n",
            "1088/1088 [==============================] - 229s 210ms/step - loss: 0.3274 - accuracy: 0.8745 - val_loss: 0.1696 - val_accuracy: 0.9505\n",
            "Epoch 6/10\n",
            "1088/1088 [==============================] - 226s 208ms/step - loss: 0.3101 - accuracy: 0.8804 - val_loss: 0.1719 - val_accuracy: 0.9486\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Having found the combination of parameters that leads to the lowest loss, we do a final training run to observe the results and plot loss and accuracy, and then use the chosen parameters for the evaluation and verification phase of the model's prediction capability."
      ],
      "metadata": {
        "id": "dMX0_-1xOJ6Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "i = parameters[0]\n",
        "j = parameters[1]\n",
        "model=build_lstm(params['n_units'][i], params['lr'][j])\n",
        "\n",
        "history_LSTM = model.fit(x_train, y_train,\n",
        "                            epochs = 10,\n",
        "                            batch_size=128,\n",
        "                            validation_data=(x_val, y_val),\n",
        "                            shuffle=True,\n",
        "                            callbacks = [earlystopping])\n",
        "epoch = history_LSTM.epoch\n",
        "plot_loss(history_LSTM)\n",
        "plot_acc(history_LSTM, epoch)"
      ],
      "metadata": {
        "id": "i9qYMrQJ4BbS",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "077c80a8-9b7d-4e6e-a054-2dad90e498ac"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "136/136 [==============================] - 52s 328ms/step - loss: 0.6891 - accuracy: 0.5714 - val_loss: 0.5593 - val_accuracy: 0.7136\n",
            "Epoch 2/10\n",
            "136/136 [==============================] - 41s 299ms/step - loss: 0.6104 - accuracy: 0.6668 - val_loss: 0.4132 - val_accuracy: 0.8286\n",
            "Epoch 3/10\n",
            "136/136 [==============================] - 41s 305ms/step - loss: 0.5205 - accuracy: 0.7472 - val_loss: 0.3121 - val_accuracy: 0.8746\n",
            "Epoch 4/10\n",
            "136/136 [==============================] - 40s 291ms/step - loss: 0.4436 - accuracy: 0.7997 - val_loss: 0.2521 - val_accuracy: 0.9044\n",
            "Epoch 5/10\n",
            "136/136 [==============================] - 43s 314ms/step - loss: 0.3921 - accuracy: 0.8291 - val_loss: 0.2292 - val_accuracy: 0.9110\n",
            "Epoch 6/10\n",
            "136/136 [==============================] - 40s 296ms/step - loss: 0.3552 - accuracy: 0.8480 - val_loss: 0.2215 - val_accuracy: 0.9158\n",
            "Epoch 7/10\n",
            "136/136 [==============================] - 41s 299ms/step - loss: 0.3472 - accuracy: 0.8536 - val_loss: 0.2067 - val_accuracy: 0.9216\n",
            "Epoch 8/10\n",
            "136/136 [==============================] - 41s 298ms/step - loss: 0.3302 - accuracy: 0.8614 - val_loss: 0.1988 - val_accuracy: 0.9245\n",
            "Epoch 9/10\n",
            "136/136 [==============================] - 41s 300ms/step - loss: 0.3143 - accuracy: 0.8684 - val_loss: 0.1965 - val_accuracy: 0.9281\n",
            "Epoch 10/10\n",
            "136/136 [==============================] - 42s 306ms/step - loss: 0.3177 - accuracy: 0.8667 - val_loss: 0.1918 - val_accuracy: 0.9270\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjcAAAHHCAYAAABDUnkqAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABvEklEQVR4nO3dd3hTZePG8W+StuluGaUDSsuSvWRUqAwFBQeKExRkiKjIEHnxh7wqAr6KAxQZgiCC4MIBigsEBAcgGwSEArJHy2xLW7qS/P4IRCqlFGh72vT+XFeuJicnOXcskpvnPOcck8PhcCAiIiLiJsxGBxAREREpSCo3IiIi4lZUbkRERMStqNyIiIiIW1G5EREREbeiciMiIiJuReVGRERE3IrKjYiIiLgVlRsRERFxKyo3Igbo1asX0dHRV/XakSNHYjKZCjZQMbNv3z5MJhOzZs0q0u0uX74ck8nE8uXLXcvy+7sqrMzR0dH06tWrQN8zP2bNmoXJZGLfvn1Fvm2Ra6VyI3IBk8mUr9uFX34i12rlypWMHDmSxMREo6OIuAUPowOIFCdz5szJ8Xj27NksXrz4ouW1a9e+pu1Mnz4du91+Va994YUXeO65565p+5J/1/K7yq+VK1cyatQoevXqRXBwcI7n4uLiMJv171CRK6FyI3KB7t2753j8xx9/sHjx4ouW/1taWhq+vr753o6np+dV5QPw8PDAw0P/6xaVa/ldFQSr1Wro9kVKIv1zQOQKtW3blnr16rF+/Xpat26Nr68v//3vfwH45ptvuOOOO4iIiMBqtVKtWjVefvllbDZbjvf49zyO8/M1xo4dy7Rp06hWrRpWq5VmzZqxdu3aHK/Nbc6NyWRiwIABfP3119SrVw+r1UrdunVZuHDhRfmXL19O06ZN8fb2plq1arz33nv5nsfz22+/8cADD1C5cmWsViuRkZE888wznD179qLP5+/vz+HDh+ncuTP+/v6EhIQwdOjQi/5bJCYm0qtXL4KCgggODqZnz5752j2zbt06TCYTH3744UXPLVq0CJPJxHfffQfA/v37eeqpp6hZsyY+Pj6UK1eOBx54IF/zSXKbc5PfzH/++Se9evWiatWqeHt7ExYWxqOPPsrJkydd64wcOZJnn30WgCpVqrh2fZ7Pltucmz179vDAAw9QtmxZfH19ueGGG/j+++9zrHN+/tDnn3/OK6+8QqVKlfD29qZdu3bs3r37sp/7Ut59913q1q2L1WolIiKC/v37X/TZd+3axX333UdYWBje3t5UqlSJrl27kpSU5Fpn8eLF3HjjjQQHB+Pv70/NmjVd/x+JXCv980/kKpw8eZLbbruNrl270r17d0JDQwHnJEx/f3+GDBmCv78/P//8MyNGjCA5OZk333zzsu/7ySefcObMGZ544glMJhNvvPEG9957L3v27LnsCMLvv//OvHnzeOqppwgICGDChAncd999HDhwgHLlygGwceNGOnbsSHh4OKNGjcJmszF69GhCQkLy9bm/+OIL0tLS6NevH+XKlWPNmjVMnDiRQ4cO8cUXX+RY12az0aFDB2JiYhg7dixLlixh3LhxVKtWjX79+gHgcDi4++67+f3333nyySepXbs28+fPp2fPnpfN0rRpU6pWrcrnn39+0fpz586lTJkydOjQAYC1a9eycuVKunbtSqVKldi3bx9Tpkyhbdu2/PXXX1c06nYlmRcvXsyePXvo3bs3YWFhbNu2jWnTprFt2zb++OMPTCYT9957Lzt37uTTTz/l7bffpnz58gCX/J0kJCTQsmVL0tLSGDRoEOXKlePDDz/krrvu4ssvv+See+7Jsf5rr72G2Wxm6NChJCUl8cYbb9CtWzdWr16d78983siRIxk1ahTt27enX79+xMXFMWXKFNauXcuKFSvw9PQkMzOTDh06kJGRwcCBAwkLC+Pw4cN89913JCYmEhQUxLZt27jzzjtp0KABo0ePxmq1snv3blasWHHFmURy5RCRS+rfv7/j3/+btGnTxgE4pk6detH6aWlpFy174oknHL6+vo709HTXsp49ezqioqJcj/fu3esAHOXKlXOcOnXKtfybb75xAI5vv/3Wteyll166KBPg8PLycuzevdu1bPPmzQ7AMXHiRNeyTp06OXx9fR2HDx92Ldu1a5fDw8PjovfMTW6fb8yYMQ6TyeTYv39/js8HOEaPHp1j3caNGzuaNGnievz11187AMcbb7zhWpadne1o1aqVA3DMnDkzzzzDhw93eHp65vhvlpGR4QgODnY8+uijeeZetWqVA3DMnj3btWzZsmUOwLFs2bIcn+XC39WVZM5tu59++qkDcPz666+uZW+++aYDcOzdu/ei9aOiohw9e/Z0PR48eLADcPz222+uZWfOnHFUqVLFER0d7bDZbDk+S+3atR0ZGRmudd955x0H4NiyZctF27rQzJkzc2Q6duyYw8vLy3Hrrbe6tuFwOByTJk1yAI4PPvjA4XA4HBs3bnQAji+++OKS7/322287AMfx48fzzCBytbRbSuQqWK1WevfufdFyHx8f1/0zZ85w4sQJWrVqRVpaGjt27Ljs+3bp0oUyZcq4Hrdq1Qpw7oa4nPbt21OtWjXX4wYNGhAYGOh6rc1mY8mSJXTu3JmIiAjXetWrV+e222677PtDzs+XmprKiRMnaNmyJQ6Hg40bN160/pNPPpnjcatWrXJ8lh9++AEPDw/XSA6AxWJh4MCB+crTpUsXsrKymDdvnmvZTz/9RGJiIl26dMk1d1ZWFidPnqR69eoEBwezYcOGfG3rajJfuN309HROnDjBDTfcAHDF271w+82bN+fGG290LfP39+fxxx9n3759/PXXXznW7927N15eXq7HV/Jn6kJLliwhMzOTwYMH55jg3LdvXwIDA127xYKCggDnrsG0tLRc3+v8pOlvvvmm0CdrS+mkciNyFSpWrJjjC+O8bdu2cc899xAUFERgYCAhISGuycgXzje4lMqVK+d4fL7onD59+opfe/7151977Ngxzp49S/Xq1S9aL7dluTlw4AC9evWibNmyrnk0bdq0AS7+fN7e3hftWrkwDzjnwoSHh+Pv759jvZo1a+YrT8OGDalVqxZz5851LZs7dy7ly5fn5ptvdi07e/YsI0aMIDIyEqvVSvny5QkJCSExMTFfv5cLXUnmU6dO8fTTTxMaGoqPjw8hISFUqVIFyN+fh0ttP7dtnT+Cb//+/TmWX8ufqX9vFy7+nF5eXlStWtX1fJUqVRgyZAjvv/8+5cuXp0OHDkyePDnH5+3SpQuxsbE89thjhIaG0rVrVz7//HMVHSkwmnMjchUu/Bf5eYmJibRp04bAwEBGjx5NtWrV8Pb2ZsOGDQwbNixff3FbLJZclzscjkJ9bX7YbDZuueUWTp06xbBhw6hVqxZ+fn4cPnyYXr16XfT5LpWnoHXp0oVXXnmFEydOEBAQwIIFC3jooYdyHFE2cOBAZs6cyeDBg2nRogVBQUGYTCa6du1aqF+oDz74ICtXruTZZ5+lUaNG+Pv7Y7fb6dixY5F9kRf2n4vcjBs3jl69evHNN9/w008/MWjQIMaMGcMff/xBpUqV8PHx4ddff2XZsmV8//33LFy4kLlz53LzzTfz008/FdmfHXFfKjciBWT58uWcPHmSefPm0bp1a9fyvXv3GpjqHxUqVMDb2zvXI2Xyc/TMli1b2LlzJx9++CE9evRwLV+8ePFVZ4qKimLp0qWkpKTkGAmJi4vL93t06dKFUaNG8dVXXxEaGkpycjJdu3bNsc6XX35Jz549GTdunGtZenr6VZ00L7+ZT58+zdKlSxk1ahQjRoxwLd+1a9dF73klZ5yOiorK9b/P+d2eUVFR+X6vK3H+fePi4qhatapreWZmJnv37qV9+/Y51q9fvz7169fnhRdeYOXKlcTGxjJ16lT+97//AWA2m2nXrh3t2rXjrbfe4tVXX+X5559n2bJlF72XyJXSbimRAnL+X5sX/os4MzOTd99916hIOVgsFtq3b8/XX3/NkSNHXMt3797Njz/+mK/XQ87P53A4eOedd6460+233052djZTpkxxLbPZbEycODHf71G7dm3q16/P3LlzmTt3LuHh4TnK5fns/x6pmDhx4kWHpRdk5tz+ewGMHz/+ovf08/MDyFfZuv3221mzZg2rVq1yLUtNTWXatGlER0dTp06d/H6UK9K+fXu8vLyYMGFCjs80Y8YMkpKSuOOOOwBITk4mOzs7x2vr16+P2WwmIyMDcO6u+7dGjRoBuNYRuRYauREpIC1btqRMmTL07NmTQYMGYTKZmDNnTqEO/1+pkSNH8tNPPxEbG0u/fv2w2WxMmjSJevXqsWnTpjxfW6tWLapVq8bQoUM5fPgwgYGBfPXVV1c8d+NCnTp1IjY2lueee459+/ZRp04d5s2bd8XzUbp06cKIESPw9vamT58+F53R984772TOnDkEBQVRp04dVq1axZIlS1yHyBdG5sDAQFq3bs0bb7xBVlYWFStW5Keffsp1JK9JkyYAPP/883Tt2hVPT086derkKj0Xeu655/j000+57bbbGDRoEGXLluXDDz9k7969fPXVV4V2NuOQkBCGDx/OqFGj6NixI3fddRdxcXG8++67NGvWzDW37Oeff2bAgAE88MADXHfddWRnZzNnzhwsFgv33XcfAKNHj+bXX3/ljjvuICoqimPHjvHuu+9SqVKlHBOlRa6Wyo1IASlXrhzfffcd//nPf3jhhRcoU6YM3bt3p127dq7zrRitSZMm/PjjjwwdOpQXX3yRyMhIRo8ezfbt2y97NJenpyfffvuta/6Et7c399xzDwMGDKBhw4ZXlcdsNrNgwQIGDx7MRx99hMlk4q677mLcuHE0btw43+/TpUsXXnjhBdLS0nIcJXXeO++8g8Vi4eOPPyY9PZ3Y2FiWLFlyVb+XK8n8ySefMHDgQCZPnozD4eDWW2/lxx9/zHG0GkCzZs14+eWXmTp1KgsXLsRut7N3795cy01oaCgrV65k2LBhTJw4kfT0dBo0aMC3337rGj0pLCNHjiQkJIRJkybxzDPPULZsWR5//HFeffVV13mYGjZsSIcOHfj22285fPgwvr6+NGzYkB9//NF1pNhdd93Fvn37+OCDDzhx4gTly5enTZs2jBo1ynW0lci1MDmK0z8rRcQQnTt3Ztu2bbnOBxERKWk050aklPn3pRJ27drFDz/8QNu2bY0JJCJSwDRyI1LKhIeHu653tH//fqZMmUJGRgYbN26kRo0aRscTEblmmnMjUsp07NiRTz/9lPj4eKxWKy1atODVV19VsRERt6GRGxEREXErmnMjIiIibkXlRkRERNxKqZtzY7fbOXLkCAEBAVd0ynMRERExjsPh4MyZM0RERFz2ZJWlrtwcOXKEyMhIo2OIiIjIVTh48CCVKlXKc51SV24CAgIA53+cwMBAg9OIiIhIfiQnJxMZGen6Hs9LqSs353dFBQYGqtyIiIiUMPmZUqIJxSIiIuJWVG5ERETErRSLcjN58mSio6Px9vYmJiaGNWvWXHLdtm3bYjKZLroV9tVwRUREpGQwfM7N3LlzGTJkCFOnTiUmJobx48fToUMH4uLiqFChwkXrz5s3j8zMTNfjkydP0rBhQx544IGijC0iIufYbDaysrKMjiFuwMvL67KHeeeH4ZdfiImJoVmzZkyaNAlwnocmMjKSgQMH8txzz1329ePHj2fEiBEcPXoUPz+/y66fnJxMUFAQSUlJmlAsInINHA4H8fHxJCYmGh1F3ITZbKZKlSp4eXld9NyVfH8bOnKTmZnJ+vXrGT58uGuZ2Wymffv2rFq1Kl/vMWPGDLp27ZqvYiMiIgXnfLGpUKECvr6+OjGqXJPzJ9k9evQolStXvqY/T4aWmxMnTmCz2QgNDc2xPDQ0lB07dlz29WvWrGHr1q3MmDHjkutkZGSQkZHhepycnHz1gUVEBHDuijpfbMqVK2d0HHETISEhHDlyhOzsbDw9Pa/6fYrFhOKrNWPGDOrXr0/z5s0vuc6YMWMICgpy3XR2YhGRa3d+jo2vr6/BScSdnN8dZbPZrul9DC035cuXx2KxkJCQkGN5QkICYWFheb42NTWVzz77jD59+uS53vDhw0lKSnLdDh48eM25RUTESbuipCAV1J8nQ8uNl5cXTZo0YenSpa5ldrudpUuX0qJFizxf+8UXX5CRkUH37t3zXM9qtbrORqyzEouIiLg/w3dLDRkyhOnTp/Phhx+yfft2+vXrR2pqKr179wagR48eOSYcnzdjxgw6d+6sfb0iImK46Ohoxo8fn+/1ly9fjslkKvQjzWbNmkVwcHChbqM4Mvw8N126dOH48eOMGDGC+Ph4GjVqxMKFC12TjA8cOHDRMe9xcXH8/vvv/PTTT0ZEFhGREupyuz1eeuklRo4cecXvu3bt2is6ardly5YcPXqUoKCgK96WXJ7h5QZgwIABDBgwINfnli9fftGymjVrYvDpeXK1fv8posv5Uc7fanQUERHJxdGjR133586dy4gRI4iLi3Mt8/f3d913OBzYbDY8PC7/VRkSEnJFOby8vC47t1SunuG7pdzFrzuP8/D01fScuYbkdJ2pU0SkOAoLC3PdgoKCMJlMrsc7duwgICCAH3/8kSZNmmC1Wvn999/5+++/ufvuuwkNDcXf359mzZqxZMmSHO/7791SJpOJ999/n3vuuQdfX19q1KjBggULXM//e7fU+d1HixYtonbt2vj7+9OxY8ccZSw7O5tBgwYRHBxMuXLlGDZsGD179qRz585X9N9gypQpVKtWDS8vL2rWrMmcOXNczzkcDkaOHEnlypWxWq1EREQwaNAg1/PvvvsuNWrUwNvbm9DQUO6///4r2nZRUbkpIBXL+OBv9WDr4WT6zFrL2cxrO4xNRKSkcTgcpGVmG3IryNH85557jtdee43t27fToEEDUlJSuP3221m6dCkbN26kY8eOdOrUiQMHDuT5PqNGjeLBBx/kzz//5Pbbb6dbt26cOnXqkuunpaUxduxY5syZw6+//sqBAwcYOnSo6/nXX3+djz/+mJkzZ7JixQqSk5P5+uuvr+izzZ8/n6effpr//Oc/bN26lSeeeILevXuzbNkyAL766ivefvtt3nvvPXbt2sXXX39N/fr1AVi3bh2DBg1i9OjRxMXFsXDhQlq3bn1F2y8qxWK3lDuoFuLP7D7N6TrtD9buO80TH61neo8mWD0sRkcTESkSZ7Ns1BmxyJBt/zW6A75eBfOVNnr0aG655RbX47Jly9KwYUPX45dffpn58+ezYMGCS06pAOjVqxcPPfQQAK+++ioTJkxgzZo1dOzYMdf1s7KymDp1KtWqVQOcUzZGjx7ten7ixIkMHz6ce+65B4BJkybxww8/XNFnGzt2LL169eKpp54CnAf1/PHHH4wdO5abbrqJAwcOEBYWRvv27fH09KRy5cquc8kdOHAAPz8/7rzzTgICAoiKiqJx48ZXtP2iopGbAlQ3IohZvZvh42nh153HGfzZJrJtdqNjiYjIFWjatGmOxykpKQwdOpTatWsTHByMv78/27dvv+zITYMGDVz3/fz8CAwM5NixY5dc39fX11VsAMLDw13rJyUlkZCQkOOktRaLhSZNmlzRZ9u+fTuxsbE5lsXGxrJ9+3YAHnjgAc6ePUvVqlXp27cv8+fPJzs7G4BbbrmFqKgoqlatyiOPPMLHH39MWlraFW2/qGjkpoA1iSrLtB5N6DNrHT9ujee5eVt4474GmM060ZWIuDcfTwt/je5g2LYLyr+Peho6dCiLFy9m7NixVK9eHR8fH+6//34yMzPzfJ9/Xz7AZDJht1/6H7y5rV/UB89ERkYSFxfHkiVLWLx4MU899RRvvvkmv/zyCwEBAWzYsIHly5fz008/MWLECEaOHMnatWuL3eHmGrkpBK1qhDDhocZYzCa+XH+I0d/9VSyP7hIRKUgmkwlfLw9DboV5puQVK1bQq1cv7rnnHurXr09YWBj79u0rtO3lJigoiNDQUNauXetaZrPZ2LBhwxW9T+3atVmxYkWOZStWrKBOnTquxz4+PnTq1IkJEyawfPlyVq1axZYtWwDw8PCgffv2vPHGG/z555/s27ePn3/++Ro+WeHQyE0h6VgvjDfua8B/vtjMrJX7CPT2YMitNY2OJSIiV6hGjRrMmzePTp06YTKZePHFF/McgSksAwcOZMyYMVSvXp1atWoxceJETp8+fUXF7tlnn+XBBx+kcePGtG/fnm+//ZZ58+a5jv6aNWsWNpuNmJgYfH19+eijj/Dx8SEqKorvvvuOPXv20Lp1a8qUKcMPP/yA3W6nZs3i992mclOI7mtSidTMbEZ8s40JP+8mwNuTvq2rGh1LRESuwFtvvcWjjz5Ky5YtKV++PMOGDSM5ObnIcwwbNoz4+Hh69OiBxWLh8ccfp0OHDlgs+d8l17lzZ9555x3Gjh3L008/TZUqVZg5cyZt27YFIDg4mNdee40hQ4Zgs9moX78+3377LeXKlSM4OJh58+YxcuRI0tPTqVGjBp9++il169YtpE989UyOUra/JDk5maCgIJKSkorsOlOTl+3mzUXOk0SNubc+DzWvXCTbFREpLOnp6ezdu5cqVarg7e1tdJxSyW63U7t2bR588EFefvllo+MUiLz+XF3J97dGbopA/5uqcyY9m6m//M1/52/Bz+rBXQ0jjI4lIiIlyP79+/npp59o06YNGRkZTJo0ib179/Lwww8bHa3Y0YTiIjKsY02631AZhwOGzN3EzzsSjI4kIiIliNlsZtasWTRr1ozY2Fi2bNnCkiVLqF27ttHRih2N3BQRk8nE6LvqkZKezdebjvDkRxuY1bsZLauVNzqaiIiUAJGRkRcd6SS508hNETKbTbz5QENuqRNKZradvh+uY+OB00bHEhERcSsqN0XM02Jm4kONia1ejtRMG71mrmVHfNHPuhcREXFXKjcG8Pa0MO2RpjSuHEzS2Sy6v7+GfSdSjY4lIiLiFlRuDOJn9WBWr+bUCgvgREoG3d5fzZHEs0bHEhERKfFUbgwU5OvJnD4xVCnvx+HEs3SfsZoTKRlGxxIRESnRVG4MFhJg5aPHYogI8mbP8VR6zFhD0tkso2OJiIiUWCo3xUDFYB8+eiyG8v5e/HU0mUdnrSUtM9voWCIicglt27Zl8ODBrsfR0dGMHz8+z9eYTCa+/vrra952Qb1PXkaOHEmjRo0KdRuFSeWmmKga4s+cPjEEenuwfv9pnpiznoxsm9GxRETcSqdOnejYsWOuz/3222+YTCb+/PPPK37ftWvX8vjjj19rvBwuVTCOHj3KbbfdVqDbcjcqN8VI7fBAZj3aHF8vC7/tOsGgTzeSbSv6K8+KiLirPn36sHjxYg4dOnTRczNnzqRp06Y0aNDgit83JCQEX1/fgoh4WWFhYVit1iLZVkmlclPMXF+5DNN7NMXLYmbRtgT+78s/sdtL1bVNRUQKzZ133klISAizZs3KsTwlJYUvvviCPn36cPLkSR566CEqVqyIr68v9evX59NPP83zff+9W2rXrl20bt0ab29v6tSpw+LFiy96zbBhw7juuuvw9fWlatWqvPjii2RlOedczpo1i1GjRrF582ZMJhMmk8mV+d+7pbZs2cLNN9+Mj48P5cqV4/HHHyclJcX1fK9evejcuTNjx44lPDyccuXK0b9/f9e28sNutzN69GgqVaqE1WqlUaNGLFy40PV8ZmYmAwYMIDw8HG9vb6KiohgzZgwADoeDkSNHUrlyZaxWKxEREQwaNCjf274auvxCMRRbvTyTHm5Mv483MG/jYfy9PRh1V11MJpPR0URELs3hgKw0Y7bt6Qv5+DvSw8ODHj16MGvWLJ5//nnX36tffPEFNpuNhx56iJSUFJo0acKwYcMIDAzk+++/55FHHqFatWo0b978stuw2+3ce++9hIaGsnr1apKSknLMzzkvICCAWbNmERERwZYtW+jbty8BAQH83//9H126dGHr1q0sXLiQJUuWABAUFHTRe6SmptKhQwdatGjB2rVrOXbsGI899hgDBgzIUeCWLVtGeHg4y5YtY/fu3XTp0oVGjRrRt2/fy34egHfeeYdx48bx3nvv0bhxYz744APuuusutm3bRo0aNZgwYQILFizg888/p3Llyhw8eJCDBw8C8NVXX/H222/z2WefUbduXeLj49m8eXO+tnu1VG6KqVvrhjHugYY88/kmZq/aT4C3B892qGV0LBGRS8tKg1cjjNn2f4+Al1++Vn300Ud58803+eWXX2jbti3g3CV13333ERQURFBQEEOHDnWtP3DgQBYtWsTnn3+er3KzZMkSduzYwaJFi4iIcP73ePXVVy+aJ/PCCy+47kdHRzN06FA+++wz/u///g8fHx/8/f3x8PAgLCzsktv65JNPSE9PZ/bs2fj5OT//pEmT6NSpE6+//jqhoaEAlClThkmTJmGxWKhVqxZ33HEHS5cuzXe5GTt2LMOGDaNr164AvP766yxbtozx48czefJkDhw4QI0aNbjxxhsxmUxERUW5XnvgwAHCwsJo3749np6eVK5cOV//Ha+FdksVY50bV+Tlu+sBMHnZ30z95W+DE4mIlHy1atWiZcuWfPDBBwDs3r2b3377jT59+gBgs9l4+eWXqV+/PmXLlsXf359FixZx4MCBfL3/9u3biYyMdBUbgBYtWly03ty5c4mNjSUsLAx/f39eeOGFfG/jwm01bNjQVWwAYmNjsdvtxMXFuZbVrVsXi8XiehweHs6xY8fytY3k5GSOHDlCbGxsjuWxsbFs374dcO762rRpEzVr1mTQoEH89NNPrvUeeOABzp49S9WqVenbty/z588nO7twjwjWyE0x1/2GKM6kZ/P6wh289uMO/K0edL8h6vIvFBEpap6+zhEUo7Z9Bfr06cPAgQOZPHkyM2fOpFq1arRp0waAN998k3feeYfx48dTv359/Pz8GDx4MJmZmQUWd9WqVXTr1o1Ro0bRoUMHgoKC+Oyzzxg3blyBbeNCnp6eOR6bTCbs9oI7YOX6669n7969/PjjjyxZsoQHH3yQ9u3b8+WXXxIZGUlcXBxLlixh8eLFPPXUU66Rs3/nKigauSkB+rWtRv+bqgHw4jdb+WbTYYMTiYjkwmRy7hoy4naFcxIffPBBzGYzn3zyCbNnz+bRRx91zb9ZsWIFd999N927d6dhw4ZUrVqVnTt35vu9a9euzcGDBzl69Khr2R9//JFjnZUrVxIVFcXzzz9P06ZNqVGjBvv378+xjpeXFzZb3qcEqV27Nps3byY19Z/rE65YsQKz2UzNmjXznTkvgYGBREREsGLFihzLV6xYQZ06dXKs16VLF6ZPn87cuXP56quvOHXqFAA+Pj506tSJCRMmsHz5clatWsWWLVsKJF9uNHJTQgy9tSYp6dl8uGo/Qz7fjK+XB7fUCTU6lohIieTv70+XLl0YPnw4ycnJ9OrVy/VcjRo1+PLLL1m5ciVlypThrbfeIiEhIccXeV7at2/PddddR8+ePXnzzTdJTk7m+eefz7FOjRo1OHDgAJ999hnNmjXj+++/Z/78+TnWiY6OZu/evWzatIlKlSoREBBw0SHg3bp146WXXqJnz56MHDmS48ePM3DgQB555BHXfJuC8Oyzz/LSSy9RrVo1GjVqxMyZM9m0aRMff/wxAG+99Rbh4eE0btwYs9nMF198QVhYGMHBwcyaNQubzUZMTAy+vr589NFH+Pj45JiXU9A0clNCmEwmXupUl3uvr4jN7qD/JxtYsfuE0bFEREqsPn36cPr0aTp06JBjfswLL7zA9ddfT4cOHWjbti1hYWF07tw53+9rNpuZP38+Z8+epXnz5jz22GO88sorOda56667eOaZZxgwYACNGjVi5cqVvPjiiznWue++++jYsSM33XQTISEhuR6O7uvry6JFizh16hTNmjXj/vvvp127dkyaNOnK/mNcxqBBgxgyZAj/+c9/qF+/PgsXLmTBggXUqFEDcB759cYbb9C0aVOaNWvGvn37+OGHHzCbzQQHBzN9+nRiY2Np0KABS5Ys4dtvv6VcuXIFmvFCJofDUapOopKcnExQUBBJSUkEBgYaHeeKZdvs9P9kA4u2JeDrZeGjx2K4vnIZo2OJSCmTnp7O3r17qVKlCt7e3kbHETeR15+rK/n+1shNCeNhMTPhoca0qlGetEwbvT5Yw/ajyUbHEhERKTZUbkogq4eF9x5pQpOoMiSnZ/PIjNXsOZ5y+ReKiIiUAio3JZSvlwcf9GpGnfBATqRk0v391RxOPGt0LBEREcOp3JRgQT6ezO7TnKohfhxJSqf7+6s5fibD6FgiIiKGUrkp4cr7W/moTwwVg33YeyKVR2asJikt/xdDExG5FqXsmBQpZAX150nlxg1EBPvw8WMxhARY2RF/hl6z1pCaUbinthaR0u38mWXT0gy6UKa4pfNngb7wUhFXQyfxcxPR5f2Y06c5Xd77g40HEnl8zjpm9GyGt+e1/QEREcmNxWIhODjYdX0iX19f1xl+Ra6G3W7n+PHj+Pr64uFxbfVE57lxM5sOJtJt+h+kZtq4pU4o73a7Hk+LBuhEpOA5HA7i4+NJTEw0Ooq4CbPZTJUqVfDy8rrouSv5/la5cUOr/j5Jz5lryMy2c0/jiox7oCFms/5FJSKFw2azkZWluX5y7by8vDCbc/8H+ZV8f2u3lBtqUa0cU7pdzxNz1jN/42H8rBZevruehoxFpFBYLJZrniMhUpC0v8JNtasdyltdGmEywUd/HOD1hXFGRxIRESkSKjdu7K6GEbzSuT4AU3/5m8nLdhucSEREpPCp3Li5h2Mq89/bawHw5qI4Zq/aZ2wgERGRQqZyUwo83roaA2+uDsCIb7Yxb8MhgxOJiIgUHpWbUmLILdfRq2U0AM9++ScLt8YbG0hERKSQqNyUEiaTiRF31uH+JpWw2R0M+nQjv+06bnQsERGRAqdyU4qYzSZeu7c+t9ULI9Nm5/HZ61m//5TRsURERAqUyk0p42ExM75rI1pfF8LZLBu9Zq5l25Eko2OJiIgUGJWbUsjqYeG97k1oFl2GM+nZ9Jixhr+PpxgdS0REpECo3JRSPl4WZvRqRr2KgZxMzaT7+6s5dFpX9xURkZJP5aYUC/T25MPezakW4sfRpHS6v7+aY2fSjY4lIiJyTVRuSrly/lY+fuwGKpXxYd/JNB55fw2JaZlGxxIREblqKjdCWJA3Hz8WQ4UAK3EJZ+g5cy0pGdlGxxIREbkqKjcCQFQ5Pz56LIZgX082H0zk8dnryLLZjY4lIiJyxVRuxOW60ABmP9ocPy8LK/8+ycgF24yOJCIicsVUbiSHBpWCmfBQY0wm+Hj1AeboQpsiIlLCqNzIRdrVDmVYR+eVxEd++xcrdp8wOJGIiEj+GV5uJk+eTHR0NN7e3sTExLBmzZo8109MTKR///6Eh4djtVq57rrr+OGHH4oobenxROuq3Nu4Ija7g6c+3sC+E6lGRxIREckXQ8vN3LlzGTJkCC+99BIbNmygYcOGdOjQgWPHjuW6fmZmJrfccgv79u3jyy+/JC4ujunTp1OxYsUiTu7+TCYTr95bn0aRwSSdzeKx2etITs8yOpaIiMhlmRwOh8OojcfExNCsWTMmTZoEgN1uJzIykoEDB/Lcc89dtP7UqVN588032bFjB56enle1zeTkZIKCgkhKSiIwMPCa8pcGx5LTuWvSCuKT02lbM4QZPZthMZuMjiUiIqXMlXx/GzZyk5mZyfr162nfvv0/Ycxm2rdvz6pVq3J9zYIFC2jRogX9+/cnNDSUevXq8eqrr2Kz2S65nYyMDJKTk3PcJP8qBHozvUdTvD3NLI87zusLdxgdSUREJE+GlZsTJ05gs9kIDQ3NsTw0NJT4+PhcX7Nnzx6+/PJLbDYbP/zwAy+++CLjxo3jf//73yW3M2bMGIKCgly3yMjIAv0cpUH9SkG8eX9DAKb9uocv1x8yOJGIiMilGT6h+ErY7XYqVKjAtGnTaNKkCV26dOH5559n6tSpl3zN8OHDSUpKct0OHjxYhIndR6eGEQy8uToA/523hfX7TxmcSEREJHeGlZvy5ctjsVhISEjIsTwhIYGwsLBcXxMeHs51112HxWJxLatduzbx8fFkZuZ+PSSr1UpgYGCOm1ydZ9pfR4e6oWTa7DwxZwNHEs8aHUlEROQihpUbLy8vmjRpwtKlS13L7HY7S5cupUWLFrm+JjY2lt27d2O3/3NZgJ07dxIeHo6Xl1ehZy7tzGYTbz3YiFphAZxIyaDv7HWkZeoaVCIiUrwYultqyJAhTJ8+nQ8//JDt27fTr18/UlNT6d27NwA9evRg+PDhrvX79evHqVOnePrpp9m5cyfff/89r776Kv379zfqI5Q6flYP3u/ZlHJ+Xmw7kszQLzZjtxt2wJ2IiMhFPIzceJcuXTh+/DgjRowgPj6eRo0asXDhQtck4wMHDmA2/9O/IiMjWbRoEc888wwNGjSgYsWKPP300wwbNsyoj1AqVSrjy9RHmvDw9D/4YUs8E0J3Mbj9dUbHEhERAQw+z40RdJ6bgvP52oP831d/AvBut+u5vX64wYlERMRdlYjz3EjJ92CzSB6NrQLAkM83sfVwksGJREREVG7kGv339lq0vi6E9Cw7j89ex/EzGUZHEhGRUk7lRq6Jh8XMxIcaUzXEjyNJ6TwxZx0Z2Zc+Y7SIiEhhU7mRaxbk48n7PZoS6O3BhgOJ/HfeVkrZVC4RESlGVG6kQFQN8Wdyt+uxmE18teEQ7/+21+hIIiJSSqncSIFpVSOEF+6oDcCYH7ezbMcxgxOJiEhppHIjBapXy2i6NovE7oBBn25k97EzRkcSEZFSRuVGCpTJZGL03fVoHl2WMxnZ9PlwHYlpuV/3S0REpDCo3EiB8/IwM6X79VQq48P+k2n0/2QDWTb75V8oIiJSAFRupFCU87cyvUdTfL0srNh9kpe/+8voSCIiUkqo3EihqR0eyPgujTCZYPaq/Xy8er/RkUREpBRQuZFCdWvdMIbeWhOAl77Zxqq/TxqcSERE3J3KjRS6p9pW466GEWTbHfT7eD0HTqYZHUlERNyYyo0UOpPJxBv3N6BBpSAS07J4bPZazqRnGR1LRETclMqNFAlvTwvTHmlKhQArOxNSGPzZJmx2XaJBREQKnsqNFJmwIG+m9WiKl4eZpTuO8eaiOKMjiYiIG1K5kSLVKDKYN+9vAMDUX/5m/sZDBicSERF3o3IjRe7uRhV5qm01AIZ9tYWNB04bnEhERNyJyo0YYuitNWlfO5TMbDuPz1nP0aSzRkcSERE3oXJTULLSYfV7sPEjo5OUCGazifFdG1EzNIDjZzJ4fPZ6zmbajI4lIiJuQOWmoGybBz/+HywZCZmpRqcpEfytHrzfsyllfD3ZcjiJZ7/cjMOhI6hEROTaqNwUlPoPQHAUpB6HtTOMTlNiRJb1ZUr3JniYTXz351Em/bzb6EgiIlLCqdwUFIsntBnmvL9iPGSkGBqnJLmhajlG310PgHGLd7Jwa7zBiUREpCRTuSlIDbpA2aqQdhLWTDM6TYnycExlerWMBmDI55v460iysYFERKTEUrkpSBYPaPOc8/7KCZCuL+gr8cIdtbmxennSMm30nb2OEykZRkcSEZESSOWmoNW/H8rVgLOnnUdPSb55WMxMergx0eV8OZx4ln4frScz2250LBERKWFUbgqa2QJtz43erJoIZxMNjVPSBPt68X7PZgR4e7B232le/HqrjqASEZEronJTGOreAyG1ID0J/phidJoSp3oFfyY+1BizCeauO8jMFfuMjiQiIiWIyk1huHD05o93nbuo5Iq0rVmB/95eG4D/ff8Xv+w8bnAiEREpKVRuCkvtu6FCXchIhlWTjU5TIvW5sQoPNKmE3QEDPtnA38d1eL2IiFyeyk1hMZvhpuHO+39MgbRTxuYpgUwmE/+7px5NospwJj2bvh+uIykty+hYIiJSzKncFKZad0JYA8hMcR4aLlfM6mFhavcmRAR5s+dEKgM+3UC2TUdQiYjIpancFCaTCW76r/P+6mmQonkjVyMkwMr0nk3x8bTw264TvPLDdqMjiYhIMaZyU9iu6wgRjSErFVa+Y3SaEqtuRBBvd2kIwMwV+/hszQGDE4mISHGlclPYTCZoe270Zs37cCbB2DwlWMd64TzT/joAXvxmK2v2ah6TiIhcTOWmKNS4BSo2heyzzotqylUb1K46dzQIJ8vm4MmP1nPwVJrRkUREpJhRuSkKF869WfcBJB81Nk8JZjKZGHt/Q+pVDORUaiZ9Z68jNSPb6FgiIlKMqNwUlWo3Q+QNkJ0Ov79tdJoSzcfLwrRHmlLe38qO+DMMnrsJu12XaBARESeVm6Jy4ejN+pmQdNjYPCVcRLAP03o0wctiZvFfCby1eKfRkUREpJhQuSlKVVpD1I1gy4TfxhmdpsS7vnIZxtxbH4BJy3bzzSYVRhERUbkpWibTP2ct3jAbEnU487W6r0klnmhdFYD/+/JPNh9MNDaQiIgYTuWmqEXf6BzBsWfBr2ONTuMW/q9jLW6uVYGMbDuPz1lHQnK60ZFERMRAKjdGOH/em00fw+l9hkZxBxaziXe6NqJGBX8SkjN4fPY60rNsRscSERGDqNwYIaqF8+gpezb8+qbRadxCgLcn7/dsSrCvJ5sPJTHsqz9xOHQElYhIaaRyYxTX6M2ncPJvY7O4iahyfrz78PVYzCa+2XSEKb/ov6uISGmkcmOUyGZQ41Zw2DR6U4BaVi/PyLvqAvDmojgW/6XLXYiIlDYqN0Zqe+7IqT/nwoldxmZxI4/cEEX3GyrjcMDTn21k/f7TRkcSEZEipHJjpIrXQ83bwWGHX143Oo1bealTXVrVKE9apo1eH6xhy6EkoyOJiEgRUbkxWtvnnD+3fAnHdhibxY14Wsy890gTmkeX5UxGNo98sJod8clGxxIRkSKgcmO08IZQ607AAb+8ZnQat+Lr5cGMXk1pFBlMYloW3d9fze5jKUbHEhGRQqZyUxycn3uzbT4kbDM2i5sJ8Pbkw97NqRMeyImUTLq9/wf7T6YaHUtERAqRyk1xEFYP6nR23l+u0ZuCFuTryUePxbhO8vfw9NUcTjxrdCwRESkkKjfFRdvnABNsXwBH/zQ6jdsp6+fFx31jqFLej8OJZ+k2/Q9dpkFExE2p3BQXFWpDvfuc9zV6UygqBHjz8WMxVCrjw76TaXR7fzUnUzKMjiUiIgVM5aY4aTMMTGaI+x6ObDQ6jVuKCPbh0743EBboze5jKXSfsYbEtEyjY4mISAFSuSlOQq6D+g867y8bY2wWNxZZ1pdP+sZQ3t/K9qPJ9PxgDcnpWUbHEhGRAlIsys3kyZOJjo7G29ubmJgY1qxZc8l1Z82ahclkynHz9vYuwrSFrM3/gckCuxbBoXVGp3FbVUP8+fixGMqcu9DmozPXkpaZbXQsEREpAIaXm7lz5zJkyBBeeuklNmzYQMOGDenQoQPHjh275GsCAwM5evSo67Z///4iTFzIylWDhl2d95e9amwWN1czLIA5fWII8PZg3f7TPPbhOtKzbEbHEhGRa2R4uXnrrbfo27cvvXv3pk6dOkydOhVfX18++OCDS77GZDIRFhbmuoWGhhZh4iLQ+lnn6M3fS+HAaqPTuLV6FYP48NHm+HlZWPn3SZ78aD0Z2So4IiIlmaHlJjMzk/Xr19O+fXvXMrPZTPv27Vm1atUlX5eSkkJUVBSRkZHcfffdbNvmZie+K1sFGndz3l+u0ZvCdn3lMnzQqxnenmaWxx1n0KcbybLZjY4lIiJXydByc+LECWw220UjL6GhocTHx+f6mpo1a/LBBx/wzTff8NFHH2G322nZsiWHDh3Kdf2MjAySk5Nz3EqEVkPB7Al7lsO+FUancXsxVcsxvUdTvDzMLNqWwJDPN2OzO4yOJSIiV8Hw3VJXqkWLFvTo0YNGjRrRpk0b5s2bR0hICO+9916u648ZM4agoCDXLTIysogTX6UyUXD9I877y3XkVFFoVSOEKd2ux8Ns4tvNRxj21Z/YVXBEREocQ8tN+fLlsVgsJCQk5FiekJBAWFhYvt7D09OTxo0bs3v37lyfHz58OElJSa7bwYMHrzl3kWn1H7B4wb7fYO+vRqcpFdrVDmXCQ40xm+DL9YcYsWArDocKjohISWJoufHy8qJJkyYsXbrUtcxut7N06VJatGiRr/ew2Wxs2bKF8PDwXJ+3Wq0EBgbmuJUYQZWgSS/n/WWvgr5ki8Tt9cMZ92BDTCb46I8DvPL9dhUcEZESxPDdUkOGDGH69Ol8+OGHbN++nX79+pGamkrv3r0B6NGjB8OHD3etP3r0aH766Sf27NnDhg0b6N69O/v37+exxx4z6iMUrhuHgMUKB1bBnmVGpyk17mlciTH31Afg/d/38tbinQYnEhGR/PIwOkCXLl04fvw4I0aMID4+nkaNGrFw4ULXJOMDBw5gNv/TwU6fPk3fvn2Jj4+nTJkyNGnShJUrV1KnTh2jPkLhCgyHpo/C6inO0ZuqN4HJZHSqUqFr88pkZNt5acE2Jv68G29PC/1vqm50LBERuQyTo5SNtycnJxMUFERSUlLJ2UV1JgHeaQjZZ6HbV1Cj/eVfIwXmvV/+ZsyPOwB44Y7aPNaqqsGJRERKnyv5/jZ8t5TkQ0AoNOvjvL/sFc29KWJPtKnG4PY1APjf99v56A83OiO2iIgbUrkpKWIHg6cvHNkAOxcZnabUebpdDZ5sUw2AF77eypfrcz+vkoiIGE/lpqTwD4Hmjzvva/SmyJlMJoZ1rEmvltEA/N+Xm/l28xFjQ4mISK5UbkqSloPAyx/i/4Qd3xudptQxmUy81KkODzWPxO6AwXM3sWhb7mfSFhER46jclCR+5SDmSef95WPArusfFTWTycQrnetzb+OK2OwOBnyygeVxl76CvYiIFD2Vm5KmRX+wBkLCVti+wOg0pZLZbOKN+xtwR/1wsmwOnpiznpW7TxgdS0REzlG5KWl8y8IN/Zz3l7+m0RuDeFjMjO/aiPa1K5CRbeex2etYt++U0bFERASVm5LphqfAGgTHt8Nf841OU2p5WsxMevh6WtUoT1qmjd4z1/LnoUSjY4mIlHoqNyWRTzC0HOC8v/w1sNsMjVOaeXtamPZIU2KqlOVMRjaPzFjDX0eSjY4lIlKqqdyUVDFPgncwnNgJW78yOk2p5uNlYUavZjSuHEzS2SwembGa3cfOGB1LRKTUUrkpqbwDIXaQ8/7y18CWbWyeUs7f6sGs3s2pVzGQk6mZPDx9NftOpBodS0SkVFK5KcmaPw6+5eDU37Dlc6PTlHpBPp7MeTSGmqEBHDuTQbf3V3PodJrRsURESh2Vm5LMGgCxTzvv//I62LKMzSOU8fPio8diqFrej8OJZ+n2/mrik9KNjiUiUqqo3JR0zR4DvxA4vQ82f2p0GgFCAqx83DeGyLI+7D+ZRrf3/+BESobRsURESg2Vm5LOy895UU2AX9+E7ExD44hTeJAPnzx2A+FB3vx9PJXu768mMU2/GxGRoqBy4w6aPgr+oZB4ADZ9bHQaOSeyrC+f9L2BkAArO+LP8MiMNSSna9ehiEhhU7lxB16+cOMQ5/1fx0K2doEUF1XK+/HJYzGU9fNiy+Ekes9cS2qGjmwTESlMKjfuokkvCAiH5EOwYbbRaeQCNUIDmNOnOYHeHqzff5rHPlxHepZOvCgiUlhUbtyFpze0+o/z/m/jIEtH6BQndSOCmN0nBn+rB6v2nOTxOevJyFbBEREpDCo37uT6HhBYCc4chfWzjE4j/9IoMpiZvZvh42nh153HGfDJRrJsuvCpiEhBU7lxJx5WaH1u9Ob3tyDrrLF55CLNosvyfs+meHmYWfxXAs/M3YTN7jA6loiIW1G5cTeNukNQZUhJgHUfGJ1GchFbvTzvdW+Cp8XEd38e5f++/BO7Co6ISIG5qnJz8OBBDh065Hq8Zs0aBg8ezLRp0wosmFwlDy9o86zz/u9vQ6aub1Qc3VSrAhMfaozFbOKrDYd48ZutOBwqOCIiBeGqys3DDz/MsmXLAIiPj+eWW25hzZo1PP/884wePbpAA8pVaPgQlImG1OOw9n2j08gldKwXzlsPNsRkgo9XH+Dl77ar4IiIFICrKjdbt26lefPmAHz++efUq1ePlStX8vHHHzNr1qyCzCdXw+IJbYY57/8+HjLOGBpHLu3uRhV5/d4GAHywYi9jf4ozOJGISMl3VeUmKysLq9UKwJIlS7jrrrsAqFWrFkePHi24dHL16j8IZavB2VOwRrsLi7MHm0Uy+u66AExe9jcTl+4yOJGISMl2VeWmbt26TJ06ld9++43FixfTsWNHAI4cOUK5cuUKNKBcJYsHtH3OeX/FBEhPNjaP5KlHi2iev702AOMW72T6r3sMTiQiUnJdVbl5/fXXee+992jbti0PPfQQDRs2BGDBggWu3VVSDNS7D8pfB+mJsHqq0WnkMvq2rsp/brkOgFd+2M6cVfuMDSQiUkKZHFc5g9Fms5GcnEyZMmVcy/bt24evry8VKlQosIAFLTk5maCgIJKSkggMDDQ6TuHb8iV81QesQTD4T/AJNjqR5MHhcPDmojjeXf43AG/c14AHm0UanEpExHhX8v19VSM3Z8+eJSMjw1Vs9u/fz/jx44mLiyvWxaZUqnsPhNSGjCT4Y4rRaeQyTCYTz3aoyaOxVQAYNu9Pvtl02OBUIiIly1WVm7vvvpvZs50XZ0xMTCQmJoZx48bRuXNnpkzRF2ixYrb8M/fmj3ch7ZSxeeSyTCYTL95Zm4djKuNwwJDPN7Nwqybqi4jk11WVmw0bNtCqVSsAvvzyS0JDQ9m/fz+zZ89mwoQJBRpQCkDtuyC0HmQkw6rJRqeRfDCZTPzv7nrcd30lbHYH/T/ZyPglO3UtKhGRfLiqcpOWlkZAQAAAP/30E/feey9ms5kbbriB/fv3F2hAKQBmM7Qd7ry/eiqknjQ2j+SL2WzijfsbcH8TZ8EZv2QX909dxZ7jKUZHExEp1q6q3FSvXp2vv/6agwcPsmjRIm699VYAjh07Vjom6ZZEte6AsAaQmQIrNbpWUljMJsY+0JB3ujYi0NuDzQcTuX3Cb8z5Y7/OZiwicglXVW5GjBjB0KFDiY6Opnnz5rRo0QJwjuI0bty4QANKATGZ4KbnnffXTIOU48bmkStyd6OKLBzcmtjq5UjPsvPi11vpNXMtCcnpRkcTESl2rvpQ8Pj4eI4ePUrDhg0xm50dac2aNQQGBlKrVq0CDVmQSt2h4BdyOGD6zXBkA7QYAB1eMTqRXCG73cGslft4feEOMrLtBPt68uo99bm9frjR0URECtWVfH9fdbk57/zVwStVqnQtb1NkSnW5Adi1GD6+Hzx84OnNEBBqdCK5CrsSzvDM55vYeth55ul7G1dk5N11CfT2NDiZiEjhKPTz3NjtdkaPHk1QUBBRUVFERUURHBzMyy+/jN2uozmKtertoVIzyD4LK8YbnUauUo3QAOb1i2XATdUxm2DexsN0fPtXVv59wuhoIiKGu6py8/zzzzNp0iRee+01Nm7cyMaNG3n11VeZOHEiL774YkFnlIJkMsFN/3XeXzsDknX+lJLKy8PM0A41+eLJFkSV8+VIUjoPT1/N/777i/Qsm9HxREQMc1W7pSIiIpg6darrauDnffPNNzz11FMcPlx8z6ha6ndLgXPuzczb4MAqaP443P6m0YnkGqVmZPO/77fz6ZoDAFwX6s/bXRpRNyLI4GQiIgWj0HdLnTp1KtdJw7Vq1eLUKZ0Bt9i7cPRm/SxIOmRoHLl2flYPxtxbnxk9m1Le34udCSl0nryCKcv/xmbXIeMiUrpcVblp2LAhkyZNumj5pEmTaNCgwTWHkiJQpTVEtwJbJvw2zug0UkDa1Q5l0eDW3FonlCybg9cX7qDrtFUcOJlmdDQRkSJzVbulfvnlF+644w4qV67sOsfNqlWrOHjwID/88IPr0gzFkXZLXWDfCph1O5g9YeB6KBNldCIpIA6Hgy/XH2LUt3+RkpGNn5eFEZ3q8GDTSEwmk9HxRESuWKHvlmrTpg07d+7knnvuITExkcTERO699162bdvGnDlzriq0GCA6Fqq2BXsW/DbW6DRSgEwmEw80jeTHp1vRPLosqZk2hn21hcfnrOdESobR8URECtU1n+fmQps3b+b666/HZiu+R2po5OZfDqyGD24FswcMWAdlqxidSAqYze7g/d/2MPanOLJsDsr7ezHm3gbcUkfnOBKRkqPQR27EjVSOgWrtwJ4Nv2r0xh1ZzCaeaFONb/rfSK2wAE6kZNJ39jqe++pPUjKyjY4nIlLgVG7knyOnNn8KJ/82NosUmjoRgXwzIJYnWlfFZILP1h7k9nd+Y90+HeEoIu5F5UagUlOo0QEcNvjlDaPTSCGyelgYfnttPu17AxWDfThwKo0H31vFGwt3kJmts4uLiHu4ojk39957b57PJyYm8ssvv2jOTUl0ZCNMawsmMzy1GkKuMzqRFLLk9CxGLfiLrzY4z3NUJzyQ8V0bcV1ogMHJREQuVmhzboKCgvK8RUVF0aNHj2sKLwaJaAw17wCHHX553eg0UgQCvT0Z92BDpna/njK+nvx1NJk7J/7O+7/twa4T/4lICVagR0uVBBq5ycPRP+G9VoAJnloFFWobnUiKyLEz6Qz78k+WxR0HoGW1crz5QEMqBvsYnExExElHS8nVCW8AtTsBDlj2qtFppAhVCPDmg17NeOWeevh4Wlj590k6jv+V+RsPUcr+/SMibkDlRnJqOxwwwfYFsG2+0WmkCJlMJrrFRPHD061oXDmYM+nZPDN3MwM+2UhiWqbR8URE8k3lRnIKrQuthjjvL3gaTu83No8UuSrl/fjiiRb855br8DCb+H7LUW59+1d+2Xnc6GgiIvmiciMXazscKjWHjCT4qg/YsoxOJEXMw2JmYLsazHuqJdVC/Dh2JoOeH6xhxDdbOZtZfI+GFBEBlRvJjcUT7nsfvIPg0FpY9orRicQgDSoF8/2gVvRqGQ3A7FX7uWPCb2w6mGhoLhGRvKjcSO7KRMFdE533f38bdi81No8YxtvTwsi76jKnT3NCA63sOZHKfVNWMn7JTrJsOvGfiBQ/xaLcTJ48mejoaLy9vYmJiWHNmjX5et1nn32GyWSic+fOhRuwtKpzNzTt47w//0lIOWZsHjFUqxohLBrcmk4NI7DZHYxfsov7p65iz/EUo6OJiORgeLmZO3cuQ4YM4aWXXmLDhg00bNiQDh06cOxY3l+k+/btY+jQobRq1aqIkpZSHV6BCnUh9RjMfwLs+pd6aRbs68XEhxrzTtdGBHp7sPlgIrdP+I05q/bpkHERKTYMLzdvvfUWffv2pXfv3tSpU4epU6fi6+vLBx98cMnX2Gw2unXrxqhRo6hatWoRpi2FPH3g/g/Awwf+/hlWTjA6kRQDdzeqyKJnWhNbvRzpWXZe/GYbvWauJSE53ehoIiLGlpvMzEzWr19P+/btXcvMZjPt27dn1apVl3zd6NGjqVChAn369LnsNjIyMkhOTs5xkytUoRbcfu6Cmj+/DAfXGptHioXwIB/mPBrDS53qYPUw88vO43QY/yvf/3nU6GgiUsoZWm5OnDiBzWYjNDQ0x/LQ0FDi4+Nzfc3vv//OjBkzmD59er62MWbMmBzXv4qMjLzm3KVS40eg3n1gz4avHoWziUYnkmLAbDbRO7YK3w+6kXoVA0lMy6L/Jxt4Zu4mks7qFAIiYgzDd0tdiTNnzvDII48wffp0ypcvn6/XDB8+nKSkJNft4MGDhZzSTZlMcOfbEBwFiQfg26dBcyzknOoVApjXL5aBN1fHbIL5Gw9z2/hfWfn3CaOjiUgp5GHkxsuXL4/FYiEhISHH8oSEBMLCwi5a/++//2bfvn106tTJtcx+boKrh4cHcXFxVKtWLcdrrFYrVqu1ENKXQt5BcP9M+OBW+Otr2PAhNOlldCopJrw8zPzn1pq0rVmBIZ9vYv/JNB6evpo+N1bh2Q418fa0GB1RREoJQ0duvLy8aNKkCUuX/nMOFbvdztKlS2nRosVF69eqVYstW7awadMm1+2uu+7ipptuYtOmTdrlVBQqNYF2Lznv/zgMjm03No8UO02iyvDDoFY8HFMZgBm/7+WuSb+z7UiSwclEpLQwfLfUkCFDmD59Oh9++CHbt2+nX79+pKam0rt3bwB69OjB8OHDAfD29qZevXo5bsHBwQQEBFCvXj28vLyM/CilR4sBUL09ZKfDF70hM83oRFLM+Fk9ePWe+nzQqynl/a3sTEih8+QVTF62m/QsXb5BRAqX4eWmS5cujB07lhEjRtCoUSM2bdrEwoULXZOMDxw4wNGjOvqiWDGbofNU8A+F49th0XCjE0kxdXOtUBYNbkWHuqFk2Ry8uSiONm8uY9aKvSo5IlJoTI5Sduat5ORkgoKCSEpKIjAw0Og4Jdvfy2DOPYADHpgFde8xOpEUUw6Hg/kbDzN2URxHkpznwqkQYOXJNtV4OKay5uOIyGVdyfe3yo1cm6Wj4bdxYA2CJ3+FMtFGJ5JiLCPbxhfrDvHust2ukhNyruR0U8kRkTyo3ORB5aaA2bJg1h1wcDVUbAqPLnReVVwkD5nZdr5cf4jJy3ZzOPEsAOX9rTzZpirdYqLw8VLJEZGcVG7yoHJTCBIPwNQbIT0JYgfDLaOMTiQlRGa2na82OEvOodPnS44Xj7euSvcbovD1MvRsFSJSjKjc5EHlppD89Q183sN5v/s8qN7O2DxSomTZ7MzbcIhJy3Zz8JSz5JTzc5acR1qo5IiIyk2eVG4K0XdDYN0M8AuBJ1dAQOjlXyNygSybnfkbDzPp590cOOU8xUBZPy/6tqpKjxZR+FlVckRKK5WbPKjcFKKsszC9HRzbBlVvco7gmA0/24CUQFk2O19vPMykZbvZf9JZcsr4etK3dVV6tIjGXyVHpNRRucmDyk0hOx4H77WB7LPQfiTc+IzRiaQEy7bZ+WbTESb+vIt950pOsK+nayQnwFuT10VKC5WbPKjcFIENs2HBQDB7QO+FENnM6ERSwmXb7CzYfIRJP+9mz4lUwFly+sRWoVdstEqOSCmgcpMHlZsi4HDAV31g61cQXBme+A18go1OJW7AZnfw7eYjTPh5F3uOO0tOkI8nfW50lpxAlRwRt6VykweVmyKSngTvtYbT+6BOZ+cZjE0mg0OJu7DZHXz35xEmLN3F3+dKTqC3B4/eWIXesVUI8lHJEXE3Kjd5ULkpQofWwwe3gj0b7hwPTXsbnUjcjM3u4PstR5m4dBe7jqUAEODtwaOxVXj0RpUcEXeicpMHlZsitmICLH4RPLzh8eVQobbRicQN2e0Ofth6lAlLd7Ez4VzJsXrQOzaaR2+sQrCvl8EJReRaqdzkQeWmiNnt8MkDsHsJhNSGvj+Dl6/RqcRN2e0Oftwaz4Slu4hLOAOAv9WDXi2jeayVSo5ISaZykweVGwOkHIepsZCSAE16Qad3jE4kbs5ud7BoWzzvLN3Fjvh/Sk7PllE8dmNVyvip5IiUNCo3eVC5Mcie5TC7M+CA+2dCvXsNDiSlgd3u4Ke/4nln6W62H00GwM/LQo+W0fRtVZWyKjkiJYbKTR5Ubgy0dDT8Ng6sgfDkb1Am2uhEUkrY7Q4Wb0/gnSW7+OtcyfH1stCjRTR9W1WhnL/V4IQicjkqN3lQuTGQLQtm3QEHV0PFpvDoQrDoaBYpOg6HgyXbjzF+yU62HXGWHB9PCz1aRNG3dVXKq+SIFFsqN3lQuTFY4gGYeqPzPDixg+GWUUYnklLI4XCwdPsx3lm6iy2HkwBnyel+Q2Ueb12NkACVHJHiRuUmDyo3xcBfC+DzR5z3u8+D6u2MzSOllsPhYFncMcYv2cWfh5wlx9vTTPeYKB5vU5UKAd4GJxSR81Ru8qByU0x8NwTWzQC/EHhyBQSEGp1ISjGHw8HyuOOMX7qLzQcTAbB6mOkWE8WTbapSIVAlR8RoKjd5ULkpJrLOwvR2cGwbVL3JOYJjNhudSko5h8PBLzuP887SXWw8kAg4S87DMZV5sk01QlVyRAyjcpMHlZti5HgcTGsLWWnQfiTc+IzRiUQAZ8n5bdcJxi/ZyYZzJcfLw8zDzZ0lJyxIJUekqKnc5EHlppjZMBsWDASTxXn0VGRzoxOJuDgcDn7ffYJ3luxi3f7TgLPk3Nu4IrXDAwkL8iYs0JvwIG/K+VuxmHVxWJHConKTB5WbYsbhgK/6wNavIKiy8/w3PsFGpxLJweFwsPLvk4xfspO1+07nuo7FbCI0wEpokLPshAZe+NOHsEBvQoOsWD0sRZxexD2o3ORB5aYYSk+C91rD6X1QpzM8MAtM+hewFD8Oh4NVf5/kp78SiE9K52hyOglJ6Rw7k449n3+TlvXzIizQ2znqc27k58IRoLAgbwK8df4nkX9TucmDyk0xdXg9zLgV7Nlw53ho2tvoRCL5lm2zcyIlk6NJZ0lITudoUjrxyenEJ527nbufkW3P1/v5eVkuKD8+hAVZCTs3+nN+NKicnxdm7QaTUkTlJg8qN8XYigmw+EXw8Ia+yyC0jtGJRAqMw+EgMS3rn9JzrgQlXDACdDTpLMnp2fl6P0+LiQoB58pOkDfhuYwGVQjwxstDRyGKe1C5yYPKTTFmt8MnD8DuJRBSy1lwvHyNTiVSpNIys3OM+BxNSneNBp3/eSIlg/z+zV3e3+oc+Tk3AhQe5OOaDxQR7ENUWV+NAEmJoHKTB5WbYi7lOEyNhZQEaNILOr1jdCKRYifLZufYmYx/7fY6S3xyxrmf6SQkZZBpu/xusCAfT66vHEzT6LI0iSpDw0rB+Hhp0rMUPyo3eVC5KQH2LIfZnQEH3D8T6t1rcCCRksfhcHAqNTPXkZ/zPw+dTiM9K2cB8jCbqFsxiKZRZWgaVYYm0WV0GQopFlRu8qByU0IsHQ2/jQNroPPw8DLRRicScTtZNjvbjyazbt9p1u8/zbr9p0hIzrhovciyPjSNco7sNI0uw3UVArQrS4qcyk0eVG5KCFsWzLoDDq6Gik2dJ/iz6PBYkcLkcDg4dPqsq+is23eauIQzF83vCfD24PrK/4zsNIoMxtfLw5jQUmqo3ORB5aYESTwAU290ngcndjDcMsroRCKlTnJ6FhsPJLJ+3ynW7T/NpoOJpGXacqxjMZuoGxHoHNk5N8KjS1RIQVO5yYPKTQnz1wL4/BHn/e7zoHo7Y/OIlHLZNjs74s+w7lzZWb//NEeT0i9ar2KwD02jz43uRJWlZliALk8h10TlJg8qNyXQd0Ng3QzwC4EnV0BAqNGJROQChxPPsm7fKefurH2n2RGffNEZmwOsHjSqHEzTqLI0Pbcry8+qXVmSfyo3eVC5KYGyzsL0dnBsG1RtC93ng1knJhMprlIystl4wFl0Nhw4zYb9p0nNZVdW7fAAmkaV5fpzR2ZFBPsYlFhKApWbPKjclFDH42BaW8hKg3YvQashRicSkXyy2R3siE92jeys33+aw4lnL1ovIsibJtFlz+3KKkOtsAA8LPqHjDip3ORB5aYE2zAHFgwAk8V59FRkc6MTichVOpp0Nsch6NuPnsH2r31Zfl4WGlcu4zoEvXHlMvhrV1appXKTB5WbEszhgK/6wNavIKiy8/w3PsFGpxKRApCakc2mg4nOwnPgNBv3n+ZMRs7rbJlNUCss0FV2mkSVoWKwDyaTJiqXBio3eVC5KeHSk+C91nB6H9S+Cx6cDfqLTcTt2OwOdiaccR6Rde7IrEOnL96VFRboTZPoMjSODKZx5WDqRgTh7anLR7gjlZs8qNy4gcPrYcatYM+GO9+Gpo8anUhEikBCcjrr9jl3Y63ff5ptR5Iv2pXlYTZRKzyARpHBNIp0HpVVtbyfzqjsBlRu8qBy4yZWToSfXgAPb+fVw0PrGJ1IRIpYWqZzV9aGcycX3HQwkRMpmRetF+DtQaPIYBpWCnaWnsrBlPe3GpBYroXKTR5UbtyE3Q6fPAC7l0BILWfB8fI1OpWIGMjhcHA48ayz6Bxwlp0th5PIyL746uiVyvicG91x3upV1O6s4k7lJg8qN24k5ThMjYWUBLi+J9w1wehEIlLMZNnsxMWfcY3sbDqYyO5jKRetp91ZxZ/KTR5UbtzMnuUwuzPggPs/gHr3GRxIRIq75PQs/jyYxOZDiWw8cH531sVXQw/w9vhnV5Z2ZxlO5SYPKjduaOnL8NtYsAY6Dw8vE210IhEpQbQ7q2RQucmDyo0bsmXDrNvh4Gqo2AQeXQQWT6NTiUgJltvurL+Pp/Dvb8yLd2cFUbW8v3ZnFQKVmzyo3LipxAMw9UbneXBin4ZbRhudSETcTHJ6FlsOJbHpoHZnXSjLZict00ZaZjapGc6fnhYztcML9jtW5SYPKjdu7K8F8Pkjzvvdv4Lq7Y3NIyJu7cLdWZsP/rM7Kz0r991ZDSODaWzg7iy73cHZLNtFRSQ100ZahvPn2X89vnC9tEyb6znn/WzSMmxk2i7+vM2jy/L5ky0KNL/KTR5Ubtzc9/+Bte+DXwg8uQICQo1OJCKlSEHsznI4HGTa7Jy9oEz8u3CkZdpI/VfJOP8zLevS5aQweZhN+Fk98POy0KBSMFMfaVKg769ykweVGzeXdRamt4Nj26BqW+g+H8y6qrCIGCe/u7P8vCxYzCbSMm1k2wv3q9nPy4LvuSLi4+WR47Gvlwd+VudPXy8Lvl4W/KzO+35eHvhanT/9rBe81ssDL4/C/btW5SYPKjelwPE4mNYWstKg3Qho9R+jE4mIuJzfnbX5YBKbDp7Oc3eWl4fZVR58L1FA/M4VkFyfs1rw8cz52NvDUiInPKvc5EHlppTYMAcWDHDebzscWv+fRnBEpNjKstnZdyIVk8mUY9TE06K/t867ku9vjyLKJFK0GneH4ztg1SRYPgbit8A9U8EaYHQyEZGLeFrM1AjV308FRZVQ3JPJBB1egbsng8ULdnwH798CJ/82OpmIiBQylRtxb427Q68fwD8Mjm+H6TfB7qVGpxIRkUKkciPuL7IZPL4cKjVznuTv4/thxQQuOjZTRETcQrEoN5MnTyY6Ohpvb29iYmJYs2bNJdedN28eTZs2JTg4GD8/Pxo1asScOXOKMK2USIHh0Ot750iOww6LX4R5fZ2HjouIiFsxvNzMnTuXIUOG8NJLL7FhwwYaNmxIhw4dOHbsWK7rly1blueff55Vq1bx559/0rt3b3r37s2iRYuKOLmUOB5WuGsS3D4WzB6w5Qv4oAMkHjQ6mYiIFCDDDwWPiYmhWbNmTJo0CQC73U5kZCQDBw7kueeey9d7XH/99dxxxx28/PLLl11Xh4ILAPt+h897QNpJ8C0PD86G6FijU4mIyCVcyfe3oSM3mZmZrF+/nvbt/7kGkNlspn379qxateqyr3c4HCxdupS4uDhat26d6zoZGRkkJyfnuIkQfaNzHk5YA0g7AbPvgjXTNQ9HRMQNGFpuTpw4gc1mIzQ05/V/QkNDiY+Pv+TrkpKS8Pf3x8vLizvuuIOJEydyyy235LrumDFjCAoKct0iIyML9DNICRZcGR5dBPXuA3s2/DAUvh0E2RefFl1EREoOw+fcXI2AgAA2bdrE2rVreeWVVxgyZAjLly/Pdd3hw4eTlJTkuh08qPkVcgEvX7hvBrQfBZhgw2yYdSecuXS5FhGR4s3QMxSXL18ei8VCQkJCjuUJCQmEhYVd8nVms5nq1asD0KhRI7Zv386YMWNo27btRetarVasVmuB5hY3YzLBjYMhtB589SgcWuO8NlWXj6FSwV7VVkRECp+hIzdeXl40adKEpUv/Oama3W5n6dKltGjRIt/vY7fbycjQrgS5RjXaQ99lEFILzhyFmbfBpk+MTiUiIlfI8N1SQ4YMYfr06Xz44Yds376dfv36kZqaSu/evQHo0aMHw4cPd60/ZswYFi9ezJ49e9i+fTvjxo1jzpw5dO/e3aiPIO6kXDV4bAnUvANsGfB1P/hxGNiyjE4mIiL5ZPiFM7t06cLx48cZMWIE8fHxNGrUiIULF7omGR84cADzBVdzTk1N5amnnuLQoUP4+PhQq1YtPvroI7p06WLURxB3Yw2ALh/BL6/DL6/B6qmQsA0e+BD8yhmdTkRELsPw89wUNZ3nRq7I9m9h/pOQmeI8uqrrJxBW3+hUIiKlTok5z41IsVe7k3M3VZkqkHgAZtwKW+cZnUpERPKgciNyORVqw+PLoNrNkJUGX/aGJaPAbjM6mYiI5ELlRiQ/fMpAty+h5SDn49/fgk+7wtlEQ2OJiMjFVG5E8stsgVtfhnung4c37PoJ3m8Hx3canUxERC6gciNypRo86LxsQ2AlOLnbWXDiFhqdSkREzlG5EbkaEY2cF96MioWMZOcuql/e1IU3RUSKAZUbkavlHwI9voFmjwEOWPY/+LwHZKQYnUxEpFRTuRG5FhZPuGMcdJoAZk/YvsB5uPipvUYnExEptVRuRApCk57Q63vwD4Vj22D6TfD3MqNTiYiUSio3IgWlcoxzHk7FJnD2NHx0L6yarHk4IiJFTOVGpCAFRkCvH6BRN3DYYdF/nZdvyDprdDIRkVJD5UakoHl6w92ToePrYLLAn5/BzNsg6ZDRyURESgWVG5HCYDLBDU/CI/PBpywc2QjT2sL+VUYnExFxeyo3IoWpahvndalC60HqcfiwE6z7wOhUIiJuTeVGpLCViYY+P0Hde8CeBd89A98OhuxMo5OJiLgllRuRouDlB/fPhHYvASZYPxNm3wUpx4xOJiLidlRuRIqKyQSthsDDn4M1CA6scs7DObzB6GQiIm5F5UakqF13K/RdCuWvg+TDziOpNs81OpWIiNtQuRExQvka8NgSuO42yE6H+Y/DoufBlm10MhGREk/lRsQo3kHQ9RNo/azz8apJ8PF9kHbK2FwiIiWcyo2IkcxmuPkFeHA2ePrBnuXOeTgJ24xOJiJSYqnciBQHde6GxxZDcBQk7of3b4G/vjE6lYhIiaRyI1JchNZ1XnizShvISoXPe8DP/wO73ehkIiIlisqNSHHiWxa6z4MWA5yPf30TPnsY0pONzSUiUoKo3IgUNxYP6PAK3PMeWKyw80eYEgtLRsGhdRrJERG5DJPD4XAYHaIoJScnExQURFJSEoGBgUbHEcnb4Q0wt7vzfDjn+YdBzY5Q606o0ho8rMblExEpIlfy/a1yI1LcZZyBnYtgx/ewazFknvnnOS9/qN4Oat7hPDmgTxnjcoqIFCKVmzyo3EiJlp0B+36DHT9A3A9w5ug/z5ksEB3rLDq1bofgysblFBEpYCo3eVC5Ebdht8PRjc6is+N7OL495/Nh9f8pOmENnNe2EhEpoVRu8qByI27r1J5/RnQOrALHBROPgyKh5u3OohMVCxZP43KKiFwFlZs8qNxIqZB6EnYudBad3Ush++w/z3kHQY0OzqJTvT1YA4zLKSKSTyo3eVC5kVInM815WYe47yFuIaSd+Oc5i5fziKtadzhHdgLCDIspIpIXlZs8qNxIqWa3wcE1zqKz4wc49XfO5ys2Obf76k4Iqal5OiJSbKjc5EHlRuQchwNO7IQd3zmLzuF1OZ8vW/Vc0bkDImPAbDEmp4gIKjd5UrkRuYQz8RD3o/PIq72/gC3zn+d8y8N1HZ3zdKreBF6+xuUUkVJJ5SYPKjci+ZBxxjkRecf3sGsRpCf985yHD1S72Vl0rusIfuWNyykipYbKTR5UbkSukC0L9q90Hnm143tIOvjPcyazc5fV+QnJ5aoZl1NE3JrKTR5UbkSugcMB8VvOFZ3vnPcvFFLrnwnJEY3BrGvzikjBULnJg8qNSAFKPPDPPJ19v4PD9s9z/mFQ87ZzF/hspQt8isg1UbnJg8qNSCE5e9p5Yc8d38PuJZCZ8s9zXgHOC3xWb+c8W3JghPOcOtZAHW4uIvmicpMHlRuRIpCdAXt/dRaduB8hJT739Tz9nCXnfNkJCD93+9cyjfqIlHoqN3lQuREpYnY7HNngLDpHNjgPOU8+ChlJl3/teT5lLyg7YRAQ8a8CFOE8akvn4hFxW1fy/e1RRJlEpLQym6FSU+ftQpmpzqJz5ui5wnPk3OMjOR/bMuDsKectYeult2OyXFB+chsBOvfTO0i7wkTcnMqNiBjDy8956Hheh487HM65PGeOOm/JR3MWoPPLUo85JzMnH3be8uLpe/kCFBAOnt4F+3lFpMio3IhI8WUygW9Z5y207qXXs2U7C46rAB29YFTogmXpiZCVBqf2OG958SlzQQEKh8DwfwpQUEXnYe+aCyRSLKnciEjJZ/Fwjr4ERkDFPNbLTHNObs6xG+xfBejMUchOd44YnT0Nx/7K/b3MnlChtvN8PhGNILyRs4Cp8IgYThOKRUQu5HA4R3hynQd0rvyc3ussPv9m9oTQOs6iE9HIWXwq1FHhESkAOloqDyo3InLNHA7nCQyPboIjG+HIJuf9vApPRON/Sk+FuuDhVaSRRUo6lZs8qNyISKFwOCBx/z9F53zpSU+8eF2Ll3NE5/zuLNcIjwqPyKWo3ORB5UZEisyFhefIxnOlZ9NlCs8Fc3hUeERcVG7yoHIjIoZyOOD0vn+KzvnSk57LSQ0tXs5JyudHdyIaQUhtFR4plVRu8qByIyLFzvnCc+HozuUKT445PHXA4lmUiUWKnMpNHlRuRKREcDicR2XlmMOzOffLVlis5wpPo39KT4XaKjziVlRu8qByIyIllqvwXHCE1mULz4VzeFR4pORSucmDyo2IuBW73Vl4cszh+fPShSes3j+7s8IbQlCk82zMut6WFHMqN3lQuRERt3e+8OSYw7MZMpJzX9/D+9wlJiIuuNRERM6f/mGayCyGKnHlZvLkybz55pvEx8fTsGFDJk6cSPPmzXNdd/r06cyePZutW51XB27SpAmvvvrqJdf/N5UbESmVLiw8RzY6y07CNufV1vPFBH4huRefC4uRrrouhaRElZu5c+fSo0cPpk6dSkxMDOPHj+eLL74gLi6OChUqXLR+t27diI2NpWXLlnh7e/P6668zf/58tm3bRsWKeV1UxknlRkTkAllnc15bK/nIv36eW27Pyt/7efr+U3ZcI0H/+ukf6rwemMgVKFHlJiYmhmbNmjFp0iQA7HY7kZGRDBw4kOeee+6yr7fZbJQpU4ZJkybRo0ePy66vciMicoXsdkg76bzGVvJRSD58QfE58s/P3A5dz43JDH4VLj36c74YWQMK93NJiXIl39+GVufMzEzWr1/P8OHDXcvMZjPt27dn1apV+XqPtLQ0srKyKFu2bK7PZ2RkkJGR4XqcnHyJfc4iIpI7sxn8Q5y38IaXXi8z7RKjPxeMAqXEgz3b+TMlHth46ffzCsi9+Fw4L8i/ApgtBf6RpWQztNycOHECm81GaGhojuWhoaHs2LEjX+8xbNgwIiIiaN++fa7PjxkzhlGjRl1zVhERuQwvXyhXzXm7FLsdUo/nHPG5cJfY+SKUkQyZZ+DEGTix89LvZ7I4C46Xn3NitIc1j58+l3neO+d9T+/c17F4aV5RMVeid3q+9tprfPbZZyxfvhxvb+9c1xk+fDhDhgxxPU5OTiYyMrKoIoqIyIXMZggIdd4iGl96vYyUfIwCJYDD5nxc1K6oOOWy7FLFyRronJR9/mYNUJG6CoaWm/Lly2OxWEhISMixPCEhgbCwsDxfO3bsWF577TWWLFlCgwYNLrme1WrFarUWSF4RESkiVn+w1oDyNS69jt0GKcecu7ey0iE7HbIz/vUzr+UXLMs6m8s6F/48m3Pb59+DfM4zulomc86y4x188WOfXJadv+/pUyrLkaHlxsvLiyZNmrB06VI6d+4MOCcUL126lAEDBlzydW+88QavvPIKixYtomnTpkWUVkREihWzxTn3JjC88LflcIAty1lyLlmCLrh/ybKVR/nKOuvcHZeeBGcTnUeoOexw9rTzdjXMns6Sk1cBOn/zCf7X8uASe24jw3dLDRkyhJ49e9K0aVOaN2/O+PHjSU1NpXfv3gD06NGDihUrMmbMGABef/11RowYwSeffEJ0dDTx8fEA+Pv74+/vb9jnEBERN2YyOb/oi+rL3uFwlp7zRSc96YJb4rnbBcsuWifJucvOngVpJ5y3q+Hh86/yE5S/kuRTxnkziOHlpkuXLhw/fpwRI0YQHx9Po0aNWLhwoWuS8YEDBzCbza71p0yZQmZmJvfff3+O93nppZcYOXJkUUYXEREpHCaTc5eSpw8E5D1NI1cOB2SmXFx4ci1K/7p/Numfy3dkn4WUs+eObLsC4Q3hiV+vPHcBMfw8N0VN57kRERG5DLsNMs7kUoDyGCm6cL2K10PPbws0Uok5z42IiIgUQ2bLud1LwUDUlb/ebi/gQFfGfPlVRERERK6A2dh6oXIjIiIibkXlRkRERNyKyo2IiIi4FZUbERERcSsqNyIiIuJWVG5ERETErajciIiIiFtRuRERERG3onIjIiIibkXlRkRERNyKyo2IiIi4FZUbERERcSsqNyIiIuJWPIwOUNQcDgcAycnJBicRERGR/Dr/vX3+ezwvpa7cnDlzBoDIyEiDk4iIiMiVOnPmDEFBQXmuY3LkpwK5EbvdzpEjRwgICMBkMhkdp1hKTk4mMjKSgwcPEhgYaHScUk+/j+JFv4/iR7+T4qWwfh8Oh4MzZ84QERGB2Zz3rJpSN3JjNpupVKmS0TFKhMDAQP1FUYzo91G86PdR/Oh3UrwUxu/jciM252lCsYiIiLgVlRsRERFxKyo3chGr1cpLL72E1Wo1Ooqg30dxo99H8aPfSfFSHH4fpW5CsYiIiLg3jdyIiIiIW1G5EREREbeiciMiIiJuReVGRERE3IrKjbiMGTOGZs2aERAQQIUKFejcuTNxcXFGxxLgtddew2QyMXjwYKOjlGqHDx+me/fulCtXDh8fH+rXr8+6deuMjlUq2Ww2XnzxRapUqYKPjw/VqlXj5Zdfztd1h+Ta/frrr3Tq1ImIiAhMJhNff/11jucdDgcjRowgPDwcHx8f2rdvz65du4osn8qNuPzyyy/079+fP/74g8WLF5OVlcWtt95Kamqq0dFKtbVr1/Lee+/RoEEDo6OUaqdPnyY2NhZPT09+/PFH/vrrL8aNG0eZMmWMjlYqvf7660yZMoVJkyaxfft2Xn/9dd544w0mTpxodLRSITU1lYYNGzJ58uRcn3/jjTeYMGECU6dOZfXq1fj5+dGhQwfS09OLJJ8OBZdLOn78OBUqVOCXX36hdevWRscplVJSUrj++ut59913+d///kejRo0YP3680bFKpeeee44VK1bw22+/GR1FgDvvvJPQ0FBmzJjhWnbffffh4+PDRx99ZGCy0sdkMjF//nw6d+4MOEdtIiIi+M9//sPQoUMBSEpKIjQ0lFmzZtG1a9dCz6SRG7mkpKQkAMqWLWtwktKrf//+3HHHHbRv397oKKXeggULaNq0KQ888AAVKlSgcePGTJ8+3ehYpVbLli1ZunQpO3fuBGDz5s38/vvv3HbbbQYnk7179xIfH5/j762goCBiYmJYtWpVkWQodRfOlPyx2+0MHjyY2NhY6tWrZ3ScUumzzz5jw4YNrF271ugoAuzZs4cpU6YwZMgQ/vvf/7J27VoGDRqEl5cXPXv2NDpeqfPcc8+RnJxMrVq1sFgs2Gw2XnnlFbp162Z0tFIvPj4egNDQ0BzLQ0NDXc8VNpUbyVX//v3ZunUrv//+u9FRSqWDBw/y9NNPs3jxYry9vY2OIzgLf9OmTXn11VcBaNy4MVu3bmXq1KkqNwb4/PPP+fjjj/nkk0+oW7cumzZtYvDgwUREROj3IdotJRcbMGAA3333HcuWLaNSpUpGxymV1q9fz7Fjx7j++uvx8PDAw8ODX375hQkTJuDh4YHNZjM6YqkTHh5OnTp1ciyrXbs2Bw4cMChR6fbss8/y3HPP0bVrV+rXr88jjzzCM888w5gxY4yOVuqFhYUBkJCQkGN5QkKC67nCpnIjLg6HgwEDBjB//nx+/vlnqlSpYnSkUqtdu3Zs2bKFTZs2uW5NmzalW7dubNq0CYvFYnTEUic2NvaiUyPs3LmTqKgogxKVbmlpaZjNOb/CLBYLdrvdoERyXpUqVQgLC2Pp0qWuZcnJyaxevZoWLVoUSQbtlhKX/v3788knn/DNN98QEBDg2jcaFBSEj4+PwelKl4CAgIvmOvn5+VGuXDnNgTLIM888Q8uWLXn11Vd58MEHWbNmDdOmTWPatGlGRyuVOnXqxCuvvELlypWpW7cuGzdu5K233uLRRx81OlqpkJKSwu7du12P9+7dy6ZNmyhbtiyVK1dm8ODB/O9//6NGjRpUqVKFF198kYiICNcRVYXOIXIOkOtt5syZRkcTh8PRpk0bx9NPP210jFLt22+/ddSrV89htVodtWrVckybNs3oSKVWcnKy4+mnn3ZUrlzZ4e3t7ahatarj+eefd2RkZBgdrVRYtmxZrt8XPXv2dDgcDofdbne8+OKLjtDQUIfVanW0a9fOERcXV2T5dJ4bERERcSuacyMiIiJuReVGRERE3IrKjYiIiLgVlRsRERFxKyo3IiIi4lZUbkRERMStqNyIiIiIW1G5EZFSyWQy8fXXXxsdQ0QKgcqNiBS5Xr16YTKZLrp17NjR6Ggi4gZ0bSkRMUTHjh2ZOXNmjmVWq9WgNCLiTjRyIyKGsFqthIWF5biVKVMGcO4ymjJlCrfddhs+Pj5UrVqVL7/8Msfrt2zZws0334yPjw/lypXj8ccfJyUlJcc6H3zwAXXr1sVqtRIeHs6AAQNyPH/ixAnuuecefH19qVGjBgsWLHA9d/r0abp160ZISAg+Pj7UqFHjojImIsWTyo2IFEsvvvgi9913H5s3b6Zbt2507dqV7du3A5CamkqHDh0oU6YMa9eu5YsvvmDJkiU5ysuUKVPo378/jz/+OFu2bGHBggVUr149xzZGjRrFgw8+yJ9//sntt99Ot27dOHXqlGv7f/31Fz/++CPbt29nypQplC9fvuj+A4jI1SuyS3SKiJzTs2dPh8Vicfj5+eW4vfLKKw6Hw3mF+ieffDLHa2JiYhz9+vVzOBwOx7Rp0xxlypRxpKSkuJ7//vvvHWaz2REfH+9wOByOiIgIx/PPP3/JDIDjhRdecD1OSUlxAI4ff/zR4XA4HJ06dXL07t27YD6wiBQpzbkREUPcdNNNTJkyJceysmXLuu63aNEix3MtWrRg06ZNAGzfvp2GDRvi5+fnej42Nha73U5cXBwmk4kjR47Qrl27PDM0aNDAdd/Pz4/AwECOHTsGQL9+/bjvvvvYsGEDt956K507d6Zly5ZX9VlFpGip3IiIIfz8/C7aTVRQfHx88rWep6dnjscmkwm73Q7Abbfdxv79+/nhhx9YvHgx7dq1o3///owdO7bA84pIwdKcGxEplv7444+LHteuXRuA2rVrs3nzZlJTU13Pr1ixArPZTM2aNQkICCA6OpqlS5deU4aQkBB69uzJRx99xPjx45k2bdo1vZ+IFA2N3IiIITIyMoiPj8+xzMPDwzVp94svvqBp06bceOONfPzxx6xZs4YZM2YA0K1bN1566SV69uzJyJEjOX78OAMHDuSRRx4hNDQUgJEjR/Lkk09SoUIFbrvtNs6cOcOKFSsYOHBgvvKNGDGCJk2aULduXTIyMvjuu+9c5UpEijeVGxExxMKFCwkPD8+xrGbNmuzYsQNwHsn02Wef8dRTTxEeHs6nn35KnTp1APD19WXRokU8/fTTNGvWDF9fX+677z7eeust13v17NmT9PR03n77bYYOHUr58uW5//77853Py8uL4cOHs2/fPnx8fGjVqhWfffZZAXxyESlsJofD4TA6hIjIhUwmE/Pnz6dz585GRxGREkhzbkRERMStqNyIiIiIW9GcGxEpdrS3XESuhUZuRERExK2o3IiIiIhbUbkRERERt6JyIyIiIm5F5UZERETcisqNiIiIuBWVGxEREXErKjciIiLiVlRuRERExK38P42fEvemPZXlAAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Testing: model evaluation and prediction skills"
      ],
      "metadata": {
        "id": "C8qtxrOEN6B9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"Evaluate model on test data\")\n",
        "model=build_lstm(params['n_units'][i], params['lr'][j])\n",
        "results = model.evaluate(x_test, y_test)\n",
        "print(\"test loss, test acc:\", results)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SVsbtkJceA3L",
        "outputId": "28d34d5e-bb60-412b-ee23-cfe7015223d9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Evaluate model on test data\n",
            "333/333 [==============================] - 22s 56ms/step - loss: 0.7049 - accuracy: 0.5356\n",
            "test loss, test acc: [0.7048978209495544, 0.5355801582336426]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Generate a prediction using model.predict()\n",
        "# and calculate it's shape:\n",
        "print(\"Generate a prediction\")\n",
        "prediction = model.predict(x_test)\n",
        "print(\"prediction shape:\", prediction.shape)"
      ],
      "metadata": {
        "id": "ib8jMbEo1Nu9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fdee2887-e381-456f-a9aa-6ded04288d9d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Generate a prediction\n",
            "333/333 [==============================] - 10s 31ms/step\n",
            "prediction shape: (10652, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = []\n",
        "\n",
        "for i in prediction:\n",
        "  if i >= 0.5:\n",
        "    y_pred.append(1)\n",
        "  else:\n",
        "    y_pred.append(0)"
      ],
      "metadata": {
        "id": "g0Gy-WTFeG2i"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import accuracy_score, precision_score, recall_score, confusion_matrix\n",
        "\n",
        "print('Accuracy on test set:', accuracy_score(y_pred, y_test))\n",
        "print('Precision on test set:', precision_score(y_pred, y_test))\n",
        "print('Recall on test set:', recall_score(y_pred, y_test))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aikwHY7KeL5k",
        "outputId": "23248643-83f9-448d-89c9-ae164e691d6b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy on test set: 0.9228313931656027\n",
            "Precision on test set: 0.8541519642437073\n",
            "Recall on test set: 0.9472997651969737\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "confusion_matrix(y_test, y_pred)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yG3c-JCAnZyi",
        "outputId": "b0dbaf23-24a0-42c3-df46-f21fd9048861"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[6199,  202],\n",
              "       [ 620, 3631]])"
            ]
          },
          "metadata": {},
          "execution_count": 88
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Bidirectional LSTM\n",
        "Bidirectional LSTM (BiLSTM) is a recurrent neural network used primarily on natural language processing. Unlike standard LSTM, the input flows in both directions, and it’s capable of utilizing information from both sides.\n",
        "\n",
        "In summary, BiLSTM adds one more LSTM layer, which reverses the direction of information flow, meaning that the input sequence flows backward in the additional LSTM layer."
      ],
      "metadata": {
        "id": "sndHLTbbpFwk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#BidirectionalLSTM model\n",
        "def build_bdlstm(n_units, lr):\n",
        "\n",
        "  model = Sequential()\n",
        "  #model_LSTM.add(Input(shape=(None,), dtype='int32'))\n",
        "  model.add(embedding_layer)\n",
        "  model.add(Bidirectional(LSTM(n_units, return_sequences=True, dropout=0.8, name='lstm_layer')))\n",
        "  model.add(Flatten())\n",
        "  model.add(Dense(1, activation='sigmoid'))\n",
        "\n",
        "  model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=lr),\n",
        "                     loss='binary_crossentropy',\n",
        "                     metrics = ['accuracy'])\n",
        "\n",
        "  return model"
      ],
      "metadata": {
        "id": "kZKKhIs-pE-Q"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "histories = []\n",
        "min_loss = 0\n",
        "parameters = {}\n",
        "\n",
        "for i in range(len(params['n_units'])):\n",
        "  for j in range(len(params['lr'])):\n",
        "    model=build_bdlstm(params['n_units'][i], params['lr'][j])\n",
        "\n",
        "    history = model.fit(x_train, y_train,\n",
        "                                epochs = 10,\n",
        "                                batch_size=16,\n",
        "                                validation_data=(x_val, y_val),\n",
        "                                shuffle=True,\n",
        "                                callbacks = [earlystopping],\n",
        "                                class_weight = class_weights)\n",
        "    # convert the history.history dict to a pandas DataFrame:\n",
        "    hist_df = pd.DataFrame(history.history)\n",
        "    histories.append(hist_df)\n",
        "    if i==0 and j==0:\n",
        "      min_loss = hist_df['loss'].min()\n",
        "      print('First minimum loss:', min_loss)\n",
        "    else:\n",
        "      if hist_df['loss'].min() < min_loss:\n",
        "        min_loss = hist_df['loss'].min()\n",
        "        parameters[0] = i\n",
        "        parameters[1] = j\n",
        "        print('Minimum loss until now:', min_loss, 'parameters:', parameters)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Jd0pMyAkNQCt",
        "outputId": "05523c0f-5e20-4f13-f377-e01729255468"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 326s 289ms/step - loss: 7.5741 - accuracy: 0.6654 - val_loss: 5.3221 - val_accuracy: 0.8224\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 307s 282ms/step - loss: 9.2211 - accuracy: 0.7160 - val_loss: 5.0000 - val_accuracy: 0.8557\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 303s 278ms/step - loss: 9.1312 - accuracy: 0.7398 - val_loss: 5.0000 - val_accuracy: 0.8773\n",
            "First minimum loss: 7.574073314666748\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 317s 283ms/step - loss: 0.6703 - accuracy: 0.7764 - val_loss: 0.4066 - val_accuracy: 0.9189\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 300s 275ms/step - loss: 0.6914 - accuracy: 0.8407 - val_loss: 0.4250 - val_accuracy: 0.9292\n",
            "Minimum loss until now: 0.6702916026115417 parameters: {0: 0, 1: 1}\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 334s 297ms/step - loss: 0.4901 - accuracy: 0.7505 - val_loss: 0.2040 - val_accuracy: 0.9281\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 309s 284ms/step - loss: 0.2678 - accuracy: 0.8953 - val_loss: 0.1925 - val_accuracy: 0.9399\n",
            "Epoch 3/10\n",
            "1088/1088 [==============================] - 309s 284ms/step - loss: 0.2231 - accuracy: 0.9140 - val_loss: 0.1725 - val_accuracy: 0.9504\n",
            "Epoch 4/10\n",
            "1088/1088 [==============================] - 308s 283ms/step - loss: 0.2124 - accuracy: 0.9191 - val_loss: 0.1618 - val_accuracy: 0.9551\n",
            "Epoch 5/10\n",
            "1088/1088 [==============================] - 307s 283ms/step - loss: 0.1943 - accuracy: 0.9279 - val_loss: 0.1749 - val_accuracy: 0.9527\n",
            "Minimum loss until now: 0.1943148672580719 parameters: {0: 0, 1: 2}\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 363s 325ms/step - loss: 14.4311 - accuracy: 0.6950 - val_loss: 9.3981 - val_accuracy: 0.8805\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 370s 339ms/step - loss: 18.9824 - accuracy: 0.7627 - val_loss: 11.6198 - val_accuracy: 0.8895\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 401s 357ms/step - loss: 0.9737 - accuracy: 0.8131 - val_loss: 0.6495 - val_accuracy: 0.9370\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 394s 362ms/step - loss: 1.2307 - accuracy: 0.8633 - val_loss: 0.8434 - val_accuracy: 0.9425\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 384s 343ms/step - loss: 0.3870 - accuracy: 0.8209 - val_loss: 0.1862 - val_accuracy: 0.9447\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 347s 319ms/step - loss: 0.2197 - accuracy: 0.9175 - val_loss: 0.1887 - val_accuracy: 0.9515\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 459s 410ms/step - loss: 18.1401 - accuracy: 0.7267 - val_loss: 10.9890 - val_accuracy: 0.9075\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 420s 386ms/step - loss: 21.5086 - accuracy: 0.8012 - val_loss: 14.4663 - val_accuracy: 0.9115\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 407s 365ms/step - loss: 1.3308 - accuracy: 0.8074 - val_loss: 0.8687 - val_accuracy: 0.9355\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 402s 370ms/step - loss: 1.6732 - accuracy: 0.8630 - val_loss: 1.1902 - val_accuracy: 0.9418\n",
            "Epoch 1/10\n",
            "1088/1088 [==============================] - 451s 403ms/step - loss: 0.3165 - accuracy: 0.8643 - val_loss: 0.2245 - val_accuracy: 0.9429\n",
            "Epoch 2/10\n",
            "1088/1088 [==============================] - 424s 389ms/step - loss: 0.2068 - accuracy: 0.9245 - val_loss: 0.2467 - val_accuracy: 0.9496\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "i = parameters[0]\n",
        "j = parameters[1]\n",
        "\n",
        "model=build_bdlstm(params['n_units'][i], params['lr'][j])\n",
        "\n",
        "history = model.fit(x_train, y_train,\n",
        "                            epochs = 10,\n",
        "                            batch_size=128,\n",
        "                            validation_data=(x_val, y_val),\n",
        "                            shuffle=True,\n",
        "                            callbacks = [earlystopping])\n",
        "epoch = history.epoch\n",
        "plot_loss(history)\n",
        "plot_acc(history, epoch)"
      ],
      "metadata": {
        "id": "ePFi3YWqTXt6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "f639a5c9-efc0-43ad-c493-754ebf09ad07"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "136/136 [==============================] - 79s 514ms/step - loss: 0.6389 - accuracy: 0.6338 - val_loss: 0.4192 - val_accuracy: 0.8454\n",
            "Epoch 2/10\n",
            "136/136 [==============================] - 65s 478ms/step - loss: 0.4403 - accuracy: 0.8108 - val_loss: 0.2556 - val_accuracy: 0.9067\n",
            "Epoch 3/10\n",
            "136/136 [==============================] - 69s 512ms/step - loss: 0.3035 - accuracy: 0.8818 - val_loss: 0.2088 - val_accuracy: 0.9276\n",
            "Epoch 4/10\n",
            "136/136 [==============================] - 66s 486ms/step - loss: 0.2488 - accuracy: 0.9035 - val_loss: 0.1888 - val_accuracy: 0.9391\n",
            "Epoch 5/10\n",
            "136/136 [==============================] - 67s 492ms/step - loss: 0.2242 - accuracy: 0.9133 - val_loss: 0.1793 - val_accuracy: 0.9452\n",
            "Epoch 6/10\n",
            "136/136 [==============================] - 72s 528ms/step - loss: 0.2052 - accuracy: 0.9225 - val_loss: 0.1796 - val_accuracy: 0.9466\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"Evaluate model on test data\")\n",
        "model=build_bdlstm(params['n_units'][i], params['lr'][j])\n",
        "results = model.evaluate(x_test, y_test)\n",
        "print(\"test loss, test acc:\", results)"
      ],
      "metadata": {
        "id": "JlcYfNGfTXt6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9a0a0398-5868-4d41-85bd-62e9faf85c6a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Evaluate model on test data\n",
            "333/333 [==============================] - 26s 54ms/step - loss: 0.7233 - accuracy: 0.4959\n",
            "test loss, test acc: [0.7233254313468933, 0.49586930871009827]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Generate a prediction using model.predict()\n",
        "# and calculate its shape:\n",
        "print(\"Generate a prediction\")\n",
        "prediction = model.predict(x_test)\n",
        "print(\"prediction shape:\", prediction.shape)"
      ],
      "metadata": {
        "id": "qz8nCvwyTXt6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "34d30c5d-8c64-4143-90e0-737f03e144d2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Generate a prediction\n",
            "333/333 [==============================] - 20s 51ms/step\n",
            "prediction shape: (10652, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = []\n",
        "\n",
        "for i in prediction:\n",
        "  if i >= 0.5:\n",
        "    y_pred.append(1)\n",
        "  else:\n",
        "    y_pred.append(0)"
      ],
      "metadata": {
        "id": "wnuaydq9Y-g5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#from sklearn.metrics import accuracy_score, precision_score, recall_score, confusion_matrix\n",
        "\n",
        "print('Accuracy on test set:', accuracy_score(y_pred, y_test))\n",
        "print('Precision on test set:', precision_score(y_pred, y_test))\n",
        "print('Recall on test set:', recall_score(y_pred, y_test))"
      ],
      "metadata": {
        "id": "vIqRzLkfpmtY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6457a7df-3920-4e81-d5e5-c2b030298428"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy on test set: 0.49586932031543374\n",
            "Precision on test set: 0.44389555398729713\n",
            "Recall on test set: 0.38565297363580625\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "confusion_matrix(y_pred, y_test)"
      ],
      "metadata": {
        "id": "hlc4rE2Spo8d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "74a54ad8-27ec-4990-8029-16fb353f1a3d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[3395, 2364],\n",
              "       [3006, 1887]])"
            ]
          },
          "metadata": {},
          "execution_count": 96
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Final considerations"
      ],
      "metadata": {
        "id": "tHjAEACtYShf"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "The decision to test LSTM models stems from their peculiar adaptability, due to their ability to store what is considered to be important information. In this project, a Unidirectional LSTM and a Bidirectional LSTM were compared in order to see which one actually was the best fit for a binary text classification task.\n",
        "\n",
        "The results tell us that this model has a tendency to overfit, as a broader and heavier grid search for hyperparameters is surely needed. Another useful observation could be directred to the data preprocessing, which might also be expanded even though not absolutely necessary: a text vectorization might be a good solution, often useful before applying the word embeddings to the dataset, in order to help the network lighten as much as possible.\n",
        "\n",
        "Surely, a downside of this otherwise very efficient model (and all its variants) is the time and computational complexity that brings with it and that slow maybe too much all the work.\n",
        "\n",
        "In the development of this project, the intention was to look at networks that had already been tested and approved for Natural Language processing as a study experiment, but a very interesting future approach could be a shift to extremely elastic models such as Bayesian Networks, which would have on its side its probabilistic nature, excellent for emulating semantic analysis and the relationship between words for classification purposes.\n",
        "\n",
        "\n",
        "Or again, as suggested in the paper [*Detecting Fake News With Machine Learning*](https://iopscience.iop.org/article/10.1088/1742-6596/1693/1/012158/pdf) (Jeffrey Huang 2020 J. Phys.: Conf. Ser. 1693 012158), a shift to complex networks such as Convolutional Neural Networks, which might be fit for the task considering that, in poor words, a CNN basically joins sets of information. Sliding the filter (or kernel) over the word embeddings to find convolutions might be a practical and good solution, and the fact that these are further dimensionally reduced through the Max Pooling layer reduces complexity and computation and improves drastically the heaviness of LSTM computations, BiLSTM especially.\n"
      ],
      "metadata": {
        "id": "wx8tP2EVzr_b"
      }
    }
  ],
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "mount_file_id": "1xUKZwCwfoEjgR2b_nKV4rp5eVSAB6emG",
      "authorship_tag": "ABX9TyPpjh05Gw8cq+Uq7RcWIeNP",
      "include_colab_link": true
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}