error_analysis_ipynb.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "cJrrLMGtaJnE"
      },
      "outputs": [],
      "source": [
        "from sklearn.preprocessing import MultiLabelBinarizer\n",
        "import os\n",
        "from typing import List, Dict, Any, Tuple\n",
        "import pandas as pd\n",
        "import json\n",
        "from functools import reduce\n",
        "from itertools import chain\n",
        "import numpy as np"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4v3j5EyPaJnG"
      },
      "source": [
        "# LOAD"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "pZNpJtD7aJnH"
      },
      "outputs": [],
      "source": [
        "# %load ../subtask_1_2a.py\n",
        "import pdb\n",
        "import json\n",
        "import logging.handlers\n",
        "import argparse\n",
        "import os\n",
        "import numpy as np\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.preprocessing import MultiLabelBinarizer\n",
        "from networkx import DiGraph, relabel_nodes, all_pairs_shortest_path_length\n",
        "from sklearn_hierarchical_classification.constants import ROOT\n",
        "from sklearn_hierarchical_classification.metrics import h_fbeta_score, h_recall_score, h_precision_score, \\\n",
        "    fill_ancestors, multi_labeled\n",
        "import sys\n",
        "from typing import List, Dict\n",
        "sys.path.append('.')\n",
        "\n",
        "\n",
        "KEYS = ['id', 'labels']\n",
        "logger = logging.getLogger(\"subtask_1_2a_scorer\")\n",
        "ch = logging.StreamHandler(sys.stdout)\n",
        "ch.setLevel(logging.INFO)\n",
        "formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')\n",
        "ch.setFormatter(formatter)\n",
        "logger.setLevel(logging.INFO)\n",
        "# logging.basicConfig(format='%(levelname)s : %(message)s', level=logging.INFO)\n",
        "\n",
        "G = DiGraph()\n",
        "G.add_edge(ROOT, \"Logos\")\n",
        "G.add_edge(\"Logos\", \"Repetition\")\n",
        "G.add_edge(\"Logos\", \"Obfuscation, Intentional vagueness, Confusion\")\n",
        "G.add_edge(\"Logos\", \"Reasoning\")\n",
        "G.add_edge(\"Logos\", \"Justification\")\n",
        "G.add_edge('Justification', \"Slogans\")\n",
        "G.add_edge('Justification', \"Bandwagon\")\n",
        "G.add_edge('Justification', \"Appeal to authority\")\n",
        "G.add_edge('Justification', \"Flag-waving\")\n",
        "G.add_edge('Justification', \"Appeal to fear/prejudice\")\n",
        "G.add_edge('Reasoning', \"Simplification\")\n",
        "G.add_edge('Simplification', \"Causal Oversimplification\")\n",
        "G.add_edge('Simplification', \"Black-and-white Fallacy/Dictatorship\")\n",
        "G.add_edge('Simplification', \"Thought-terminating cliché\")\n",
        "G.add_edge('Reasoning', \"Distraction\")\n",
        "G.add_edge('Distraction', \"Misrepresentation of Someone's Position (Straw Man)\")\n",
        "G.add_edge('Distraction', \"Presenting Irrelevant Data (Red Herring)\")\n",
        "G.add_edge('Distraction', \"Whataboutism\")\n",
        "G.add_edge(ROOT, \"Ethos\")\n",
        "G.add_edge('Ethos', \"Appeal to authority\")\n",
        "G.add_edge('Ethos', \"Glittering generalities (Virtue)\")\n",
        "G.add_edge('Ethos', \"Bandwagon\")\n",
        "G.add_edge('Ethos', \"Ad Hominem\")\n",
        "G.add_edge('Ethos', \"Transfer\")\n",
        "G.add_edge('Ad Hominem', \"Doubt\")\n",
        "G.add_edge('Ad Hominem', \"Name calling/Labeling\")\n",
        "G.add_edge('Ad Hominem', \"Smears\")\n",
        "G.add_edge('Ad Hominem', \"Reductio ad hitlerum\")\n",
        "G.add_edge('Ad Hominem', \"Whataboutism\")\n",
        "G.add_edge(ROOT, \"Pathos\")\n",
        "G.add_edge('Pathos', \"Exaggeration/Minimisation\")\n",
        "G.add_edge('Pathos', \"Loaded Language\")\n",
        "G.add_edge('Pathos', \"Appeal to (Strong) Emotions\")\n",
        "G.add_edge('Pathos', \"Appeal to fear/prejudice\")\n",
        "G.add_edge('Pathos', \"Flag-waving\")\n",
        "G.add_edge('Pathos', \"Transfer\")\n",
        "\n",
        "\n",
        "hier = {\n",
        "    ROOT: [\"Logos\", \"Ethos\", \"Pathos\"],\n",
        "    \"Logos\": [\"Repetition\", \"Obfuscation, Intentional vagueness, Confusion\", \"Reasoning\", \"Justification\"],\n",
        "    \"Justification\": [\n",
        "        \"Slogans\",\n",
        "        \"Bandwagon\",\n",
        "        \"Appeal to authority\",\n",
        "        \"Flag-waving\",\n",
        "        \"Appeal to fear/prejudice\",\n",
        "    ],\n",
        "    \"Reasoning\": [\n",
        "        \"Simplification\",\n",
        "        \"Distraction\",\n",
        "    ],\n",
        "    \"Simplification\": [\n",
        "        \"Causal Oversimplification\",\n",
        "        \"Black-and-white Fallacy/Dictatorship\",\n",
        "        \"Thought-terminating cliché\",\n",
        "    ],\n",
        "    \"Distraction\": [\n",
        "        \"Misrepresentation of Someone's Position (Straw Man)\",\n",
        "        \"Presenting Irrelevant Data (Red Herring)\",\n",
        "        \"Whataboutism\",\n",
        "    ],\n",
        "    \"Ethos\": [\n",
        "        \"Appeal to authority\",\n",
        "        \"Glittering generalities (Virtue)\",\n",
        "        \"Bandwagon\",\n",
        "        \"Ad Hominem\",\n",
        "        \"Transfer\",\n",
        "    ],\n",
        "    \"Ad Hominem\": [\n",
        "        \"Doubt\",\n",
        "        \"Name calling/Labeling\",\n",
        "        \"Smears\",\n",
        "        \"Reductio ad hitlerum\",\n",
        "        \"Whataboutism\",\n",
        "    ],\n",
        "    \"Pathos\": [\n",
        "        \"Exaggeration/Minimisation\",\n",
        "        \"Loaded Language\",\n",
        "        \"Appeal to (Strong) Emotions\",\n",
        "        \"Appeal to fear/prejudice\",\n",
        "        \"Flag-waving\",\n",
        "        \"Transfer\",\n",
        "    ]\n",
        "}\n",
        "all_labels = []\n",
        "for k, v in hier.items():\n",
        "    all_labels.extend(v)\n",
        "    all_labels = list(set(all_labels))\n",
        "\n",
        "\n",
        "def get_dag_labels() -> List[str]:\n",
        "    return all_labels\n",
        "\n",
        "\n",
        "def get_dag() -> DiGraph:\n",
        "    return G\n",
        "\n",
        "\n",
        "def get_all_classes_from_graph(graph):\n",
        "    return [\n",
        "        node\n",
        "        for node in graph.nodes\n",
        "        if node != ROOT\n",
        "    ]\n",
        "\n",
        "\n",
        "def _h_fbeta_score(y_true, y_pred, class_hierarchy, beta=1., root=ROOT):\n",
        "    hP = _h_precision_score(y_true, y_pred, class_hierarchy, root=root)\n",
        "    hR = _h_recall_score(y_true, y_pred, class_hierarchy, root=root)\n",
        "    return (1. + beta ** 2.) * hP * hR / (beta ** 2. * hP + hR)\n",
        "\n",
        "\n",
        "def _fill_ancestors(y, graph, root, copy=True):\n",
        "    y_ = y.copy() if copy else y\n",
        "    paths = all_pairs_shortest_path_length(graph.reverse(copy=False))\n",
        "    for target, distances in paths:\n",
        "        if target == root:\n",
        "            continue\n",
        "        ix_rows = np.where(y[:, target] > 0)[0]\n",
        "        ancestors = list(filter(lambda x: x != ROOT, distances.keys()))\n",
        "        y_[tuple(np.meshgrid(ix_rows, ancestors))] = 1\n",
        "    graph.reverse(copy=False)\n",
        "    return y_\n",
        "\n",
        "\n",
        "def _h_recall_score(y_true, y_pred, class_hierarchy, root=ROOT):\n",
        "    y_true_ = _fill_ancestors(y_true, graph=class_hierarchy, root=root)\n",
        "    y_pred_ = _fill_ancestors(y_pred, graph=class_hierarchy, root=root)\n",
        "\n",
        "    ix = np.where((y_true_ != 0) & (y_pred_ != 0))\n",
        "\n",
        "    true_positives = len(ix[0])\n",
        "    all_positives = np.count_nonzero(y_true_)\n",
        "\n",
        "    return true_positives / all_positives\n",
        "\n",
        "\n",
        "def _h_precision_score(y_true, y_pred, class_hierarchy, root=ROOT):\n",
        "    y_true_ = _fill_ancestors(y_true, graph=class_hierarchy, root=root)\n",
        "    y_pred_ = _fill_ancestors(y_pred, graph=class_hierarchy, root=root)\n",
        "\n",
        "    ix = np.where((y_true_ != 0) & (y_pred_ != 0))\n",
        "\n",
        "    true_positives = len(ix[0])\n",
        "    all_results = np.count_nonzero(y_pred_)\n",
        "\n",
        "    return true_positives / all_results\n",
        "\n",
        "\n",
        "def read_classes(file_path):\n",
        "    CLASSES = []\n",
        "    with open(file_path) as f:\n",
        "        for label in f.readlines():\n",
        "            label = label.strip()\n",
        "            if label:\n",
        "                CLASSES.append(label)\n",
        "    return CLASSES\n",
        "\n",
        "\n",
        "def check_format(file_path):\n",
        "    _classes = get_all_classes_from_graph(G)\n",
        "    if not os.path.exists(file_path):\n",
        "        logging.error(\"File doesnt exists: {}\".format(file_path))\n",
        "        return False\n",
        "    submmission = ''\n",
        "    try:\n",
        "        with open(file_path, encoding='utf-8') as p:\n",
        "            submission = json.load(p)\n",
        "    except BaseException:\n",
        "        logging.error(\"File is not a valid json file: {}\".format(file_path))\n",
        "        return False\n",
        "    for i, obj in enumerate(submission):\n",
        "        for key in KEYS:\n",
        "            if key not in obj:\n",
        "                logging.error(\"Missing entry in {}:{}\".format(file_path, i))\n",
        "                return False\n",
        "    for label in list(obj['labels']):\n",
        "        if label not in _classes:\n",
        "            print(label)\n",
        "            logging.error(\"Unknown Label in {}:{}\".format(file_path, i))\n",
        "            return False\n",
        "    return True\n",
        "\n",
        "\n",
        "def _read_gold_and_pred(pred_fpath, gold_fpath):\n",
        "    \"\"\"\n",
        "    Read gold and predicted data.\n",
        "    :param pred_fpath: a json file with predictions,\n",
        "    :param gold_fpath: the original annotated gold file.\n",
        "    :return: {id:pred_labels} dict; {id:gold_labels} dict\n",
        "    \"\"\"\n",
        "\n",
        "    gold_labels = {}\n",
        "    with open(gold_fpath, encoding='utf-8') as gold_f:\n",
        "        gold = json.load(gold_f)\n",
        "        for obj in gold:\n",
        "            gold_labels[obj['id']] = obj['labels']\n",
        "\n",
        "    pred_labels = {}\n",
        "    with open(pred_fpath, encoding='utf-8') as pred_f:\n",
        "        pred = json.load(pred_f)\n",
        "        for obj in pred:\n",
        "            pred_labels[obj['id']] = obj['labels']\n",
        "\n",
        "    if set(gold_labels.keys()) != set(pred_labels.keys()):\n",
        "        logger.error(\n",
        "            'There are either missing or added examples to the prediction file. Make sure you only have the gold examples in the prediction file.')\n",
        "        raise ValueError(\n",
        "            'There are either missing or added examples to the prediction file. Make sure you only have the gold examples in the prediction file.')\n",
        "\n",
        "    return pred_labels, gold_labels\n",
        "\n",
        "\n",
        "def evaluate_h(pred_fpath, gold_fpath):\n",
        "    pred_labels, gold_labels = _read_gold_and_pred(pred_fpath, gold_fpath)\n",
        "\n",
        "    gold = []\n",
        "    pred = []\n",
        "    for id in gold_labels:\n",
        "        gold.append(gold_labels[id])\n",
        "        pred.append(pred_labels[id])\n",
        "    with multi_labeled(gold, pred, G) as (gold_, pred_, graph_):\n",
        "        return _h_precision_score(\n",
        "            gold_, pred_, graph_), _h_recall_score(\n",
        "            gold_, pred_, graph_), _h_fbeta_score(\n",
        "            gold_, pred_, graph_)\n",
        "\n",
        "\n",
        "def validate_files(pred_files, gold_files):\n",
        "    if not check_format(pred_files):\n",
        "        logger.error('Bad format for pred file {}. Cannot score.'.format(pred_file))\n",
        "        return False\n",
        "    if not check_format(gold_files):\n",
        "        logger.error('Bad format for gold file {}. Cannot score.'.format(gold_file))\n",
        "        return False\n",
        "    return True\n",
        "\n",
        "\n",
        "# if __name__ == '__main__':\n",
        "#     parser = argparse.ArgumentParser()\n",
        "#     parser.add_argument(\n",
        "#         \"--gold_file_path\",\n",
        "#         '-g',\n",
        "#         type=str,\n",
        "#         required=True,\n",
        "#         help=\"Paths to the file with gold annotations.\"\n",
        "#     )\n",
        "#     parser.add_argument(\n",
        "#         \"--pred_file_path\",\n",
        "#         '-p',\n",
        "#         type=str,\n",
        "#         required=True,\n",
        "#         help=\"Path to the file with predictions\"\n",
        "#     )\n",
        "#     parser.add_argument(\n",
        "#         \"--log_to_file\",\n",
        "#         \"-l\",\n",
        "#         action='store_true',\n",
        "#         default=False,\n",
        "#         help=\"Set flag if you want to log the execution file. The log will be appended to <pred_file>.log\"\n",
        "#     )\n",
        "#     args = parser.parse_args()\n",
        "\n",
        "#     pred_file = args.pred_file_path\n",
        "#     gold_file = args.gold_file_path\n",
        "\n",
        "#     if args.log_to_file:\n",
        "#         output_log_file = pred_file + \".log\"\n",
        "#         logger.info(\"Logging execution to file \" + output_log_file)\n",
        "#         fileLogger = logging.FileHandler(output_log_file)\n",
        "#         fileLogger.setLevel(logging.DEBUG)\n",
        "#         fileLogger.setFormatter(formatter)\n",
        "#         logger.addHandler(fileLogger)\n",
        "#         logger.setLevel(logging.DEBUG)\n",
        "\n",
        "#     if args.log_to_file:\n",
        "#         logger.info('Reading gold file')\n",
        "#     else:\n",
        "#         logger.info(\"Reading gold predictions from file {}\".format(args.gold_file_path))\n",
        "#     if args.log_to_file:\n",
        "#         logger.info('Reading predictions file')\n",
        "#     else:\n",
        "#         logger.info('Reading predictions file {}'.format(args.pred_file_path))\n",
        "\n",
        "#     if validate_files(pred_file, gold_file):\n",
        "#         logger.info('Prediction file format is correct')\n",
        "#         prec_h, rec_h, f1_h = evaluate_h(pred_file, gold_file)\n",
        "#         logger.info(\"f1_h={:.5f}\\tprec_h={:.5f}\\trec_h={:.5f}\".format(f1_h, prec_h, rec_h))\n",
        "#         if args.log_to_file:\n",
        "#             print(\"{}\\t{}\\t{}\".format(f1_h, prec_h, rec_h))\n",
        "#         else:\n",
        "#             print(\"f1_h={:.5f}\\tprec_h={:.5f}\\trec_h={:.5f}\".format(f1_h, prec_h, rec_h))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7qXHyTv-aJnJ"
      },
      "outputs": [],
      "source": [
        "# %load ./data.py\n",
        "import json\n",
        "import pandas as pd\n",
        "from typing import List\n",
        "# from src.utils.workspace import get_workdir\n",
        "\n",
        "# DATASET_DIR = f\"{get_workdir()}/dataset\"\n",
        "DATASET_DIR = \"../dataset\"\n",
        "\n",
        "\n",
        "def _load_semeval2024() -> List[pd.DataFrame]:\n",
        "    with open(f\"{DATASET_DIR}/semeval2024/subtask1/train.json\", \"r\") as f:\n",
        "        train = pd.DataFrame().from_records(json.load(f))\n",
        "    with open(f\"{DATASET_DIR}/semeval2024/subtask1/validation.json\", \"r\") as f:\n",
        "        validation = pd.DataFrame().from_records(json.load(f))\n",
        "    with open(f\"{DATASET_DIR}/semeval2024/subtask1/dev_unlabeled.json\", \"r\") as f:\n",
        "        dev_unlabeled = pd.DataFrame().from_records(json.load(f))\n",
        "    return train, validation, dev_unlabeled\n",
        "\n",
        "\n",
        "def _load_ptc2019() -> List[pd.DataFrame]:\n",
        "    train = pd.read_csv(f\"{DATASET_DIR}/ptc_adjust/ptc_preproc_train.csv\",\n",
        "                        sep=\";\").dropna(subset=[\"text\", \"label\"])[[\"text\", \"label\"]]\n",
        "    train = train.drop_duplicates(subset=[\"text\"])\n",
        "    test = pd.read_csv(f\"{DATASET_DIR}/ptc_adjust/ptc_preproc_test.csv\",\n",
        "                       sep=\";\").dropna(subset=[\"text\", \"label\"])[[\"text\", \"label\"]]\n",
        "    test = test.drop_duplicates(subset=[\"text\"])\n",
        "    dev = pd.read_csv(f\"{DATASET_DIR}/ptc_adjust/ptc_preproc_dev.csv\",\n",
        "                      sep=\";\").dropna(subset=[\"text\", \"label\"])[[\"text\", \"label\"]]\n",
        "    dev = dev.drop_duplicates(subset=[\"text\"])\n",
        "\n",
        "    def rename_label_column(df: pd.DataFrame) -> pd.DataFrame:\n",
        "        df[\"labels\"] = df[\"label\"].str.split(\",\")\n",
        "        df = df.drop(columns=[\"label\"])\n",
        "        return df\n",
        "\n",
        "    train, test, dev = map(rename_label_column, [train, test, dev])\n",
        "    return train, test, dev\n",
        "\n",
        "\n",
        "def load_dataset(dataset: str) -> List[pd.DataFrame]:\n",
        "    \"\"\"Load a given dataset, returning splits\n",
        "\n",
        "    Parameters:\n",
        "    -----------\n",
        "    dataset : str\n",
        "              name of dataset, currently only 'ptc2019' or 'semeval2024' are supported\n",
        "\n",
        "    Returns:\n",
        "    --------\n",
        "    List[pd.DataFrame]\n",
        "              train, test and dev splits. They include the columns `text` (str) and `labels` (List[str])\n",
        "    \"\"\"\n",
        "    if dataset == \"semeval2024\":\n",
        "        return _load_semeval2024()\n",
        "    elif dataset == \"ptc2019\":\n",
        "        return _load_ptc2019()\n",
        "    raise Exception(f\"{dataset} is not available\")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9syBNqjEaJnJ"
      },
      "source": [
        "# MLPClassifier - PyTorch"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4YC_EQ-5aJnJ"
      },
      "outputs": [],
      "source": [
        "GOLD_PATH = \"../dataset/semeval2024/subtask1/validation.json\"\n",
        "OUTPUT_DIR = \"../classification\"\n",
        "\n",
        "class Namespace:\n",
        "    def __init__(self, **kwargs):\n",
        "        self.__dict__.update(kwargs)\n",
        "\n",
        "args = Namespace(\n",
        "    train_features = \"../feature_extraction/1703941782_xlm-roberta-base_train_features.json\",\n",
        "    test_features = \"../feature_extraction/1703941782_xlm-roberta-base_test_features.json\",\n",
        "    dev_features = \"../feature_extraction/1703941782_xlm-roberta-base_dev_features.json\",\n",
        "    dataset = \"semeval2024\",\n",
        "    classifier = \"other\",\n",
        "    seed = 1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "9jemMhLSaJnK"
      },
      "outputs": [],
      "source": [
        "import torch\n",
        "from torch.utils.data import Dataset, DataLoader\n",
        "import numpy as np\n",
        "\n",
        "class TextEmbeddingDataset(Dataset):\n",
        "    def __init__(self, X, y):\n",
        "        self.embedding_vectors = X\n",
        "        self.labels = y\n",
        "\n",
        "    def __len__(self):\n",
        "        return len(self.embedding_vectors)\n",
        "\n",
        "    def __getitem__(self, idx):\n",
        "        sample = torch.from_numpy(self.embedding_vectors[idx]).float(), torch.from_numpy(self.labels[idx]).float()\n",
        "        return sample"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "iWKPlqLkaJnK",
        "outputId": "cea81a05-bd0a-428b-9d4a-d6767bc9ba75"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Loading dataset files\n",
            "Dataset Lengths 7000 500 1000\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "0                  [Black-and-white Fallacy/Dictatorship]\n",
              "1       [Loaded Language, Glittering generalities (Vir...\n",
              "2                                                      []\n",
              "3                                                      []\n",
              "4                            [Thought-terminating cliché]\n",
              "                              ...                        \n",
              "6995    [Loaded Language, Causal Oversimplification, T...\n",
              "6996                      [Loaded Language, Whataboutism]\n",
              "6997                                                   []\n",
              "6998                                                   []\n",
              "6999                                            [Slogans]\n",
              "Name: labels, Length: 7000, dtype: object"
            ]
          },
          "execution_count": 6,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "    print(\"Loading dataset files\")\n",
        "    train, dev, test = load_dataset(args.dataset)\n",
        "    print(\"Dataset Lengths\", len(train), len(dev), len(test))\n",
        "\n",
        "    train[\"labels\"]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "m32pmBX0aJnL",
        "outputId": "9ff255b4-705b-486e-e032-f85fa72ccf47"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using fixed random seed: 1\n",
            "Loading features info files\n",
            "Loading dataset files\n",
            "Dataset Lengths 7000 500 1000\n",
            "Labels: [['Obfuscation, Intentional vagueness, Confusion', 'Whataboutism', 'Presenting Irrelevant Data (Red Herring)', \"Misrepresentation of Someone's Position (Straw Man)\", 'Causal Oversimplification', 'Smears', 'Appeal to fear/prejudice', 'Black-and-white Fallacy/Dictatorship', 'Thought-terminating cliché', 'Name calling/Labeling', 'Slogans', 'Flag-waving', 'Loaded Language', 'Repetition', 'Reductio ad hitlerum', 'Exaggeration/Minimisation', 'Bandwagon', 'Appeal to authority', 'Doubt', 'Glittering generalities (Virtue)']]\n",
            "No. of labels in train+dev datasets: 20\n",
            "Loading features array files\n",
            "Features Lengths 7000 1000 500\n"
          ]
        }
      ],
      "source": [
        "from sklearn.preprocessing import MultiLabelBinarizer\n",
        "import os\n",
        "from typing import List, Dict, Any, Tuple\n",
        "import pandas as pd\n",
        "import json\n",
        "from functools import reduce\n",
        "from itertools import chain\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "import random\n",
        "import torch\n",
        "\n",
        "def seed_everything(seed):\n",
        "    random.seed(seed)\n",
        "    os.environ['PYTHONHASHSEED'] = str(seed)\n",
        "    np.random.seed(seed)\n",
        "    torch.manual_seed(seed)\n",
        "    torch.cuda.manual_seed(seed)\n",
        "    torch.backends.cudnn.deterministic = True\n",
        "\n",
        "def save_predictions(test_df: pd.DataFrame, predictions: List[List[str]], kind: str, classifier: str) -> Tuple[str, List[Dict[str, Any]]]:\n",
        "    predictions_json = []\n",
        "    for idx, row in test_df.iterrows():\n",
        "        predictions_json.append({\n",
        "            \"id\": row[\"id\"],\n",
        "            \"labels\": predictions[idx],\n",
        "        })\n",
        "\n",
        "    os.makedirs(OUTPUT_DIR, exist_ok=True)\n",
        "    # submission format has txt extension but json format\n",
        "    predictions_json_path = f\"{OUTPUT_DIR}/{kind}_predictions_{classifier}.json.txt\"\n",
        "    with open(predictions_json_path, \"w\") as f:\n",
        "        json.dump(predictions_json, f, indent=4)\n",
        "\n",
        "    return predictions_json_path, predictions_json\n",
        "\n",
        "def load_features_info(path: str) -> Dict[str, Any]:\n",
        "    with open(path, \"r\") as f:\n",
        "        return json.load(f)\n",
        "\n",
        "def load_features_array(path: str) -> np.ndarray:\n",
        "    with open(path, \"r\") as f:\n",
        "        df = pd.DataFrame().from_records(json.load(f))\n",
        "        # print(np.array(df).shape)\n",
        "        return np.array(df)\n",
        "\n",
        "def classify(args: Namespace):\n",
        "    print(f'Using fixed random seed: {args.seed}')\n",
        "\n",
        "    seed_everything(args.seed)\n",
        "\n",
        "    print(\"Loading features info files\")\n",
        "    train_ft_info, dev_ft_info, test_ft_info = map(\n",
        "        load_features_info, [\n",
        "            args.train_features, args.dev_features, args.test_features])\n",
        "\n",
        "    print(\"Loading dataset files\")\n",
        "    train, dev, test = load_dataset(args.dataset)\n",
        "    print(\"Dataset Lengths\", len(train), len(dev), len(test))\n",
        "\n",
        "    all_labels = pd.concat([train[\"labels\"], dev[\"labels\"]])\n",
        "\n",
        "    if args.classifier == \"HiMLP\":\n",
        "        labels = [get_dag_labels()]\n",
        "    else:\n",
        "        labels = [list(set(reduce(lambda x, y: x + y, all_labels.to_numpy().tolist())))]\n",
        "    print(f\"Labels: {labels}\")\n",
        "    print(f\"No. of labels in {'DAG' if args.classifier == 'HiMLP' else 'train+dev datasets'}: {len(labels[0])}\")\n",
        "\n",
        "    mlb = MultiLabelBinarizer(classes=labels[0])\n",
        "    train_labels = mlb.fit(labels).transform(train[\"labels\"].to_numpy())\n",
        "    dev_labels = mlb.transform(dev[\"labels\"].to_numpy())\n",
        "\n",
        "    print(\"Loading features array files\")\n",
        "    train_ft, test_ft, dev_ft = map(\n",
        "        load_features_array, [\n",
        "            train_ft_info[\"features\"], test_ft_info[\"features\"], dev_ft_info[\"features\"]])\n",
        "    print(\"Features Lengths\", len(train_ft), len(test_ft), len(dev_ft))\n",
        "\n",
        "    # Split the data into training and validation sets\n",
        "    X_train, X_val, y_train, y_val = train_test_split(train_ft, train_labels, test_size=0.1, random_state=42)\n",
        "\n",
        "    # Create instances of the train/test/dev datasets\n",
        "    train_dataset = TextEmbeddingDataset(X_train, y_train)\n",
        "    val_dataset = TextEmbeddingDataset(X_val, y_val)\n",
        "    dev_dataset = TextEmbeddingDataset(dev_ft, dev_labels)\n",
        "\n",
        "    return train_dataset, val_dataset, dev_dataset, mlb, labels\n",
        "\n",
        "train_dataset, val_dataset, dev_dataset, mlb, leaf_labels = classify(args)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "1pQOOte2aJnM",
        "outputId": "d15bb5a7-e1ed-4282-f385-59e32d397598"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
            "  from .autonotebook import tqdm as notebook_tqdm\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/400, Validation Loss: 0.2949240878224373\n",
            "Epoch 2/400, Validation Loss: 0.2642063945531845\n",
            "Epoch 3/400, Validation Loss: 0.25818901509046555\n",
            "Epoch 4/400, Validation Loss: 0.2555961087346077\n",
            "Epoch 5/400, Validation Loss: 0.25333284959197044\n",
            "Epoch 6/400, Validation Loss: 0.2514684125781059\n",
            "Epoch 7/400, Validation Loss: 0.2502719350159168\n",
            "Epoch 8/400, Validation Loss: 0.24861587956547737\n",
            "Epoch 9/400, Validation Loss: 0.2475341185927391\n",
            "Epoch 10/400, Validation Loss: 0.24642258882522583\n",
            "Epoch 11/400, Validation Loss: 0.24522385001182556\n",
            "Epoch 12/400, Validation Loss: 0.24416539072990417\n",
            "Epoch 13/400, Validation Loss: 0.2430606707930565\n",
            "Epoch 14/400, Validation Loss: 0.24236207827925682\n",
            "Epoch 15/400, Validation Loss: 0.24147078767418861\n",
            "Epoch 16/400, Validation Loss: 0.24077168852090836\n",
            "Epoch 17/400, Validation Loss: 0.23984159529209137\n",
            "Epoch 18/400, Validation Loss: 0.23895276337862015\n",
            "Epoch 19/400, Validation Loss: 0.23840047419071198\n",
            "Epoch 20/400, Validation Loss: 0.23788655549287796\n",
            "Epoch 21/400, Validation Loss: 0.23720724880695343\n",
            "Epoch 22/400, Validation Loss: 0.23657237365841866\n",
            "Epoch 23/400, Validation Loss: 0.235868688672781\n",
            "Epoch 24/400, Validation Loss: 0.23533840477466583\n",
            "Epoch 25/400, Validation Loss: 0.2349162958562374\n",
            "Epoch 26/400, Validation Loss: 0.23388812690973282\n",
            "Epoch 27/400, Validation Loss: 0.23372958973050117\n",
            "Epoch 28/400, Validation Loss: 0.23341776803135872\n",
            "Epoch 29/400, Validation Loss: 0.23279809951782227\n",
            "Epoch 30/400, Validation Loss: 0.2319798842072487\n",
            "Epoch 31/400, Validation Loss: 0.23198074474930763\n",
            "Epoch 32/400, Validation Loss: 0.23144417256116867\n",
            "Epoch 33/400, Validation Loss: 0.23110558092594147\n",
            "Epoch 34/400, Validation Loss: 0.23070304840803146\n",
            "Epoch 35/400, Validation Loss: 0.23040127381682396\n",
            "Epoch 36/400, Validation Loss: 0.22989178448915482\n",
            "Epoch 37/400, Validation Loss: 0.22958767786622047\n",
            "Epoch 38/400, Validation Loss: 0.22875512018799782\n",
            "Epoch 39/400, Validation Loss: 0.22876333072781563\n",
            "Epoch 40/400, Validation Loss: 0.2283441722393036\n",
            "Epoch 41/400, Validation Loss: 0.2279919795691967\n",
            "Epoch 42/400, Validation Loss: 0.2279706969857216\n",
            "Epoch 43/400, Validation Loss: 0.22759225219488144\n",
            "Epoch 44/400, Validation Loss: 0.22690889984369278\n",
            "Epoch 45/400, Validation Loss: 0.22696289792656898\n",
            "Epoch 46/400, Validation Loss: 0.2263215109705925\n",
            "Epoch 47/400, Validation Loss: 0.22584613785147667\n",
            "Epoch 48/400, Validation Loss: 0.22589947655797005\n",
            "Epoch 49/400, Validation Loss: 0.22573060914874077\n",
            "Epoch 50/400, Validation Loss: 0.2253887765109539\n",
            "Epoch 51/400, Validation Loss: 0.2251768633723259\n",
            "Epoch 52/400, Validation Loss: 0.22455991059541702\n",
            "Epoch 53/400, Validation Loss: 0.22434794157743454\n",
            "Epoch 54/400, Validation Loss: 0.22417264804244041\n",
            "Epoch 55/400, Validation Loss: 0.22371472045779228\n",
            "Epoch 56/400, Validation Loss: 0.2241356000304222\n",
            "Epoch 57/400, Validation Loss: 0.22379490360617638\n",
            "Epoch 58/400, Validation Loss: 0.22310009226202965\n",
            "Epoch 59/400, Validation Loss: 0.22292346879839897\n",
            "Epoch 60/400, Validation Loss: 0.22257490456104279\n",
            "Epoch 61/400, Validation Loss: 0.22278843820095062\n",
            "Epoch 62/400, Validation Loss: 0.22213726863265038\n",
            "Epoch 63/400, Validation Loss: 0.22193805873394012\n",
            "Epoch 64/400, Validation Loss: 0.22208714112639427\n",
            "Epoch 65/400, Validation Loss: 0.22173144668340683\n",
            "Epoch 66/400, Validation Loss: 0.22135111317038536\n",
            "Epoch 67/400, Validation Loss: 0.22117012739181519\n",
            "Epoch 68/400, Validation Loss: 0.22105197235941887\n",
            "Epoch 69/400, Validation Loss: 0.22086899355053902\n",
            "Epoch 70/400, Validation Loss: 0.22057662531733513\n",
            "Epoch 71/400, Validation Loss: 0.22054826468229294\n",
            "Epoch 72/400, Validation Loss: 0.220575712621212\n",
            "Epoch 73/400, Validation Loss: 0.21995557844638824\n",
            "Epoch 74/400, Validation Loss: 0.21966100484132767\n",
            "Epoch 75/400, Validation Loss: 0.21979115903377533\n",
            "Epoch 76/400, Validation Loss: 0.21957442164421082\n",
            "Epoch 77/400, Validation Loss: 0.21894734352827072\n",
            "Epoch 78/400, Validation Loss: 0.2192050889134407\n",
            "Epoch 79/400, Validation Loss: 0.21892667189240456\n",
            "Epoch 80/400, Validation Loss: 0.21923968568444252\n",
            "Epoch 81/400, Validation Loss: 0.21872428804636002\n",
            "Epoch 82/400, Validation Loss: 0.2188097983598709\n",
            "Epoch 83/400, Validation Loss: 0.2187357395887375\n",
            "Epoch 84/400, Validation Loss: 0.21856990084052086\n",
            "Epoch 85/400, Validation Loss: 0.21846629306674004\n",
            "Epoch 86/400, Validation Loss: 0.21799788251519203\n",
            "Epoch 87/400, Validation Loss: 0.21766235679388046\n",
            "Epoch 88/400, Validation Loss: 0.2172916866838932\n",
            "Epoch 89/400, Validation Loss: 0.21769774705171585\n",
            "Epoch 90/400, Validation Loss: 0.21708808094263077\n",
            "Epoch 91/400, Validation Loss: 0.21737513691186905\n",
            "Epoch 92/400, Validation Loss: 0.21685869991779327\n",
            "Epoch 93/400, Validation Loss: 0.21712018176913261\n",
            "Epoch 94/400, Validation Loss: 0.21671169251203537\n",
            "Epoch 95/400, Validation Loss: 0.21681611984968185\n",
            "Epoch 96/400, Validation Loss: 0.21699995920062065\n",
            "Epoch 97/400, Validation Loss: 0.21699390187859535\n",
            "Epoch 98/400, Validation Loss: 0.21638791263103485\n",
            "Epoch 99/400, Validation Loss: 0.21619608253240585\n",
            "Epoch 100/400, Validation Loss: 0.21604111045598984\n",
            "Epoch 101/400, Validation Loss: 0.2157967947423458\n",
            "Epoch 102/400, Validation Loss: 0.21581094339489937\n",
            "Epoch 103/400, Validation Loss: 0.21561575308442116\n",
            "Epoch 104/400, Validation Loss: 0.21574488654732704\n",
            "Epoch 105/400, Validation Loss: 0.215558722615242\n",
            "Epoch 106/400, Validation Loss: 0.21536079421639442\n",
            "Epoch 107/400, Validation Loss: 0.21520669013261795\n",
            "Epoch 108/400, Validation Loss: 0.21534307673573494\n",
            "Epoch 109/400, Validation Loss: 0.21515369042754173\n",
            "Epoch 110/400, Validation Loss: 0.21469514071941376\n",
            "Epoch 111/400, Validation Loss: 0.21459513530135155\n",
            "Epoch 112/400, Validation Loss: 0.2145284339785576\n",
            "Epoch 113/400, Validation Loss: 0.21409108489751816\n",
            "Epoch 114/400, Validation Loss: 0.21425936371088028\n",
            "Epoch 115/400, Validation Loss: 0.2145542986690998\n",
            "Epoch 116/400, Validation Loss: 0.21408816054463387\n",
            "Epoch 117/400, Validation Loss: 0.21410813182592392\n",
            "Epoch 118/400, Validation Loss: 0.21396716684103012\n",
            "Epoch 119/400, Validation Loss: 0.21362914517521858\n",
            "Epoch 120/400, Validation Loss: 0.2136189565062523\n",
            "Epoch 121/400, Validation Loss: 0.21349618211388588\n",
            "Epoch 122/400, Validation Loss: 0.21367372572422028\n",
            "Epoch 123/400, Validation Loss: 0.2136983536183834\n",
            "Epoch 124/400, Validation Loss: 0.21336979791522026\n",
            "Epoch 125/400, Validation Loss: 0.2135404832661152\n",
            "Epoch 126/400, Validation Loss: 0.21329744532704353\n",
            "Epoch 127/400, Validation Loss: 0.21348228305578232\n",
            "Epoch 128/400, Validation Loss: 0.2132522128522396\n",
            "Epoch 129/400, Validation Loss: 0.2132243551313877\n",
            "Epoch 130/400, Validation Loss: 0.21282191202044487\n",
            "Epoch 131/400, Validation Loss: 0.2139338254928589\n",
            "Epoch 132/400, Validation Loss: 0.21336210519075394\n",
            "Epoch 133/400, Validation Loss: 0.2129407338798046\n",
            "Epoch 134/400, Validation Loss: 0.21326104924082756\n",
            "Epoch 135/400, Validation Loss: 0.21258264034986496\n",
            "Epoch 136/400, Validation Loss: 0.2127234824001789\n",
            "Epoch 137/400, Validation Loss: 0.21214096620678902\n",
            "Epoch 138/400, Validation Loss: 0.21246947720646858\n",
            "Epoch 139/400, Validation Loss: 0.21202416345477104\n",
            "Epoch 140/400, Validation Loss: 0.21242334321141243\n",
            "Epoch 141/400, Validation Loss: 0.21186133846640587\n",
            "Epoch 142/400, Validation Loss: 0.21208657324314117\n",
            "Epoch 143/400, Validation Loss: 0.2118733674287796\n",
            "Epoch 144/400, Validation Loss: 0.21230001375079155\n",
            "Epoch 145/400, Validation Loss: 0.21182193234562874\n",
            "Epoch 146/400, Validation Loss: 0.211373008787632\n",
            "Epoch 147/400, Validation Loss: 0.21219687908887863\n",
            "Epoch 148/400, Validation Loss: 0.21185333654284477\n",
            "Epoch 149/400, Validation Loss: 0.21188048273324966\n",
            "Epoch 150/400, Validation Loss: 0.21228406578302383\n",
            "Epoch 151/400, Validation Loss: 0.2114061638712883\n",
            "Epoch 152/400, Validation Loss: 0.21161719411611557\n",
            "Epoch 153/400, Validation Loss: 0.21149102598428726\n",
            "Epoch 154/400, Validation Loss: 0.21168271079659462\n",
            "Epoch 155/400, Validation Loss: 0.21152347326278687\n",
            "Epoch 156/400, Validation Loss: 0.21120601892471313\n",
            "Epoch 157/400, Validation Loss: 0.2107676863670349\n",
            "Epoch 158/400, Validation Loss: 0.21079476922750473\n",
            "Epoch 159/400, Validation Loss: 0.21109187975525856\n",
            "Epoch 160/400, Validation Loss: 0.21048473939299583\n",
            "Epoch 161/400, Validation Loss: 0.2105426974594593\n",
            "Epoch 162/400, Validation Loss: 0.2107386514544487\n",
            "Epoch 163/400, Validation Loss: 0.2109302394092083\n",
            "Epoch 164/400, Validation Loss: 0.2104884758591652\n",
            "Epoch 165/400, Validation Loss: 0.210139911621809\n",
            "Epoch 166/400, Validation Loss: 0.21048637852072716\n",
            "Epoch 167/400, Validation Loss: 0.2107827439904213\n",
            "Epoch 168/400, Validation Loss: 0.21056430786848068\n",
            "Epoch 169/400, Validation Loss: 0.20990921929478645\n",
            "Epoch 170/400, Validation Loss: 0.21002355962991714\n",
            "Epoch 171/400, Validation Loss: 0.21029876917600632\n",
            "Epoch 172/400, Validation Loss: 0.20983168110251427\n",
            "Epoch 173/400, Validation Loss: 0.2100968286395073\n",
            "Epoch 174/400, Validation Loss: 0.21025151014328003\n",
            "Epoch 175/400, Validation Loss: 0.20998532697558403\n",
            "Epoch 176/400, Validation Loss: 0.21017349511384964\n",
            "Epoch 177/400, Validation Loss: 0.21001914888620377\n",
            "Epoch 178/400, Validation Loss: 0.20983145758509636\n",
            "Epoch 179/400, Validation Loss: 0.20964641869068146\n",
            "Epoch 180/400, Validation Loss: 0.21012292057275772\n",
            "Epoch 181/400, Validation Loss: 0.20957855135202408\n",
            "Epoch 182/400, Validation Loss: 0.20967505499720573\n",
            "Epoch 183/400, Validation Loss: 0.20976198092103004\n",
            "Epoch 184/400, Validation Loss: 0.20939218997955322\n",
            "Epoch 185/400, Validation Loss: 0.2094552405178547\n",
            "Epoch 186/400, Validation Loss: 0.20925256609916687\n",
            "Epoch 187/400, Validation Loss: 0.20986981689929962\n",
            "Epoch 188/400, Validation Loss: 0.2092910148203373\n",
            "Epoch 189/400, Validation Loss: 0.20930955931544304\n",
            "Epoch 190/400, Validation Loss: 0.20943016558885574\n",
            "Epoch 191/400, Validation Loss: 0.20918698981404305\n",
            "Epoch 192/400, Validation Loss: 0.20936359092593193\n",
            "Epoch 193/400, Validation Loss: 0.20929745957255363\n",
            "Epoch 194/400, Validation Loss: 0.20889898017048836\n",
            "Epoch 195/400, Validation Loss: 0.2087891511619091\n",
            "Epoch 196/400, Validation Loss: 0.20874838903546333\n",
            "Epoch 197/400, Validation Loss: 0.2089134193956852\n",
            "Epoch 198/400, Validation Loss: 0.20936308428645134\n",
            "Epoch 199/400, Validation Loss: 0.2092871032655239\n",
            "Epoch 200/400, Validation Loss: 0.20929492637515068\n",
            "Epoch 201/400, Validation Loss: 0.2086023911833763\n",
            "Epoch 202/400, Validation Loss: 0.2083481326699257\n",
            "Epoch 203/400, Validation Loss: 0.20850378274917603\n",
            "Epoch 204/400, Validation Loss: 0.20841556414961815\n",
            "Epoch 205/400, Validation Loss: 0.20818155258893967\n",
            "Epoch 206/400, Validation Loss: 0.2087964005768299\n",
            "Epoch 207/400, Validation Loss: 0.20892835780978203\n",
            "Epoch 208/400, Validation Loss: 0.2084278091788292\n",
            "Epoch 209/400, Validation Loss: 0.20833376049995422\n",
            "Epoch 210/400, Validation Loss: 0.20880210027098656\n",
            "Epoch 211/400, Validation Loss: 0.20881657302379608\n",
            "Epoch 212/400, Validation Loss: 0.2082795798778534\n",
            "Epoch 213/400, Validation Loss: 0.2081022672355175\n",
            "Epoch 214/400, Validation Loss: 0.20825275778770447\n",
            "Epoch 215/400, Validation Loss: 0.2087450921535492\n",
            "Epoch 216/400, Validation Loss: 0.2083471678197384\n",
            "Epoch 217/400, Validation Loss: 0.2083565779030323\n",
            "Epoch 218/400, Validation Loss: 0.20809996128082275\n",
            "Epoch 219/400, Validation Loss: 0.20798338577151299\n",
            "Epoch 220/400, Validation Loss: 0.20814169570803642\n",
            "Epoch 221/400, Validation Loss: 0.2076178453862667\n",
            "Epoch 222/400, Validation Loss: 0.20794916152954102\n",
            "Epoch 223/400, Validation Loss: 0.20831282436847687\n",
            "Epoch 224/400, Validation Loss: 0.20781762525439262\n",
            "Epoch 225/400, Validation Loss: 0.207790095359087\n",
            "Epoch 226/400, Validation Loss: 0.20812059938907623\n",
            "Epoch 227/400, Validation Loss: 0.20793916657567024\n",
            "Epoch 228/400, Validation Loss: 0.20778043940663338\n",
            "Epoch 229/400, Validation Loss: 0.20760312676429749\n",
            "Epoch 230/400, Validation Loss: 0.20756664127111435\n",
            "Epoch 231/400, Validation Loss: 0.20782344788312912\n",
            "Epoch 232/400, Validation Loss: 0.20790700241923332\n",
            "Epoch 233/400, Validation Loss: 0.2081906534731388\n",
            "Epoch 234/400, Validation Loss: 0.2076275423169136\n",
            "Epoch 235/400, Validation Loss: 0.20789146423339844\n",
            "Epoch 236/400, Validation Loss: 0.20786875113844872\n",
            "Epoch 237/400, Validation Loss: 0.20770113542675972\n",
            "Epoch 238/400, Validation Loss: 0.20802951976656914\n",
            "Epoch 239/400, Validation Loss: 0.20772451162338257\n",
            "Epoch 240/400, Validation Loss: 0.20788342878222466\n",
            "Epoch 241/400, Validation Loss: 0.2077821083366871\n",
            "Epoch 242/400, Validation Loss: 0.20709944888949394\n",
            "Epoch 243/400, Validation Loss: 0.20788740366697311\n",
            "Epoch 244/400, Validation Loss: 0.20782051980495453\n",
            "Epoch 245/400, Validation Loss: 0.207163967192173\n",
            "Epoch 246/400, Validation Loss: 0.2073189727962017\n",
            "Epoch 247/400, Validation Loss: 0.2070683315396309\n",
            "Epoch 248/400, Validation Loss: 0.20809294655919075\n",
            "Epoch 249/400, Validation Loss: 0.20732859149575233\n",
            "Epoch 250/400, Validation Loss: 0.20764805376529694\n",
            "Epoch 251/400, Validation Loss: 0.20735027268528938\n",
            "Epoch 252/400, Validation Loss: 0.2076522782444954\n",
            "Epoch 253/400, Validation Loss: 0.2073746658861637\n",
            "Epoch 254/400, Validation Loss: 0.20733052864670753\n",
            "Epoch 255/400, Validation Loss: 0.2073201946914196\n",
            "Epoch 256/400, Validation Loss: 0.2071225754916668\n",
            "Epoch 257/400, Validation Loss: 0.20752828568220139\n",
            "Epoch 258/400, Validation Loss: 0.2069995403289795\n",
            "Epoch 259/400, Validation Loss: 0.20672143623232841\n",
            "Epoch 260/400, Validation Loss: 0.20744917541742325\n",
            "Epoch 261/400, Validation Loss: 0.2066434659063816\n",
            "Epoch 262/400, Validation Loss: 0.2069198451936245\n",
            "Epoch 263/400, Validation Loss: 0.2070692479610443\n",
            "Epoch 264/400, Validation Loss: 0.2070021890103817\n",
            "Epoch 265/400, Validation Loss: 0.2074255868792534\n",
            "Epoch 266/400, Validation Loss: 0.20762329176068306\n",
            "Epoch 267/400, Validation Loss: 0.20698587968945503\n",
            "Epoch 268/400, Validation Loss: 0.2069467157125473\n",
            "Epoch 269/400, Validation Loss: 0.20674657449126244\n",
            "Epoch 270/400, Validation Loss: 0.20711011067032814\n",
            "Epoch 271/400, Validation Loss: 0.2066032662987709\n",
            "Epoch 272/400, Validation Loss: 0.20662404596805573\n",
            "Epoch 273/400, Validation Loss: 0.2067979834973812\n",
            "Epoch 274/400, Validation Loss: 0.20629627257585526\n",
            "Epoch 275/400, Validation Loss: 0.2065359577536583\n",
            "Epoch 276/400, Validation Loss: 0.2071024291217327\n",
            "Epoch 277/400, Validation Loss: 0.20635805279016495\n",
            "Epoch 278/400, Validation Loss: 0.20664197579026222\n",
            "Epoch 279/400, Validation Loss: 0.2068302519619465\n",
            "Epoch 280/400, Validation Loss: 0.20695283263921738\n",
            "Epoch 281/400, Validation Loss: 0.20621660351753235\n",
            "Epoch 282/400, Validation Loss: 0.20649771392345428\n",
            "Epoch 283/400, Validation Loss: 0.20634150132536888\n",
            "Epoch 284/400, Validation Loss: 0.20658612623810768\n",
            "Epoch 285/400, Validation Loss: 0.20664478093385696\n",
            "Epoch 286/400, Validation Loss: 0.2066587544977665\n",
            "Epoch 287/400, Validation Loss: 0.20622548088431358\n",
            "Epoch 288/400, Validation Loss: 0.20627189055085182\n",
            "Epoch 289/400, Validation Loss: 0.20640650391578674\n",
            "Epoch 290/400, Validation Loss: 0.20617952942848206\n",
            "Epoch 291/400, Validation Loss: 0.20643002539873123\n",
            "Epoch 292/400, Validation Loss: 0.20659642666578293\n",
            "Epoch 293/400, Validation Loss: 0.20658524706959724\n",
            "Epoch 294/400, Validation Loss: 0.20623945072293282\n",
            "Epoch 295/400, Validation Loss: 0.20609190687537193\n",
            "Epoch 296/400, Validation Loss: 0.20647427812218666\n",
            "Epoch 297/400, Validation Loss: 0.2058165892958641\n",
            "Epoch 298/400, Validation Loss: 0.20605583116412163\n",
            "Epoch 299/400, Validation Loss: 0.20613889023661613\n",
            "Epoch 300/400, Validation Loss: 0.20637482777237892\n",
            "Epoch 301/400, Validation Loss: 0.206113763153553\n",
            "Epoch 302/400, Validation Loss: 0.20629575476050377\n",
            "Epoch 303/400, Validation Loss: 0.20641889795660973\n",
            "Epoch 304/400, Validation Loss: 0.20626240596175194\n",
            "Epoch 305/400, Validation Loss: 0.2057209238409996\n",
            "Epoch 306/400, Validation Loss: 0.20610767975449562\n",
            "Epoch 307/400, Validation Loss: 0.2061345987021923\n",
            "Epoch 308/400, Validation Loss: 0.20630666613578796\n",
            "Epoch 309/400, Validation Loss: 0.20632139965891838\n",
            "Epoch 310/400, Validation Loss: 0.2062026634812355\n",
            "Epoch 311/400, Validation Loss: 0.20573483034968376\n",
            "Epoch 312/400, Validation Loss: 0.20605767890810966\n",
            "Epoch 313/400, Validation Loss: 0.20610005781054497\n",
            "Epoch 314/400, Validation Loss: 0.20582732930779457\n",
            "Epoch 315/400, Validation Loss: 0.2057434841990471\n",
            "Epoch 316/400, Validation Loss: 0.20598214492201805\n",
            "Epoch 317/400, Validation Loss: 0.20655455812811852\n",
            "Epoch 318/400, Validation Loss: 0.20569313690066338\n",
            "Epoch 319/400, Validation Loss: 0.2058303914964199\n",
            "Epoch 320/400, Validation Loss: 0.2056734636425972\n",
            "Epoch 321/400, Validation Loss: 0.20586049556732178\n",
            "Epoch 322/400, Validation Loss: 0.20630870759487152\n",
            "Epoch 323/400, Validation Loss: 0.20575347542762756\n",
            "Epoch 324/400, Validation Loss: 0.20578989759087563\n",
            "Epoch 325/400, Validation Loss: 0.20580938458442688\n",
            "Epoch 326/400, Validation Loss: 0.20584560930728912\n",
            "Epoch 327/400, Validation Loss: 0.2055704928934574\n",
            "Epoch 328/400, Validation Loss: 0.20625174045562744\n",
            "Epoch 329/400, Validation Loss: 0.2059127800166607\n",
            "Epoch 330/400, Validation Loss: 0.2055719532072544\n",
            "Epoch 331/400, Validation Loss: 0.2057999148964882\n",
            "Epoch 332/400, Validation Loss: 0.20580827072262764\n",
            "Epoch 333/400, Validation Loss: 0.20596666634082794\n",
            "Epoch 334/400, Validation Loss: 0.20556318387389183\n",
            "Epoch 335/400, Validation Loss: 0.20559125766158104\n",
            "Epoch 336/400, Validation Loss: 0.2052433378994465\n",
            "Epoch 337/400, Validation Loss: 0.20535001158714294\n",
            "Epoch 338/400, Validation Loss: 0.2060500532388687\n",
            "Epoch 339/400, Validation Loss: 0.20591386780142784\n",
            "Epoch 340/400, Validation Loss: 0.20532719418406487\n",
            "Epoch 341/400, Validation Loss: 0.20569069683551788\n",
            "Epoch 342/400, Validation Loss: 0.20595313608646393\n",
            "Epoch 343/400, Validation Loss: 0.20567700266838074\n",
            "Epoch 344/400, Validation Loss: 0.20548225939273834\n",
            "Epoch 345/400, Validation Loss: 0.20570845901966095\n",
            "Epoch 346/400, Validation Loss: 0.20606780797243118\n",
            "Epoch 347/400, Validation Loss: 0.20539045706391335\n",
            "Epoch 348/400, Validation Loss: 0.20518256351351738\n",
            "Epoch 349/400, Validation Loss: 0.20572953671216965\n",
            "Epoch 350/400, Validation Loss: 0.20570367202162743\n",
            "Epoch 351/400, Validation Loss: 0.2054923102259636\n",
            "Epoch 352/400, Validation Loss: 0.2049984633922577\n",
            "Epoch 353/400, Validation Loss: 0.20540577173233032\n",
            "Epoch 354/400, Validation Loss: 0.20569150522351265\n",
            "Epoch 355/400, Validation Loss: 0.20535378903150558\n",
            "Epoch 356/400, Validation Loss: 0.2054225243628025\n",
            "Epoch 357/400, Validation Loss: 0.20543503388762474\n",
            "Epoch 358/400, Validation Loss: 0.20543207973241806\n",
            "Epoch 359/400, Validation Loss: 0.20526043325662613\n",
            "Epoch 360/400, Validation Loss: 0.2057032287120819\n",
            "Epoch 361/400, Validation Loss: 0.2051374725997448\n",
            "Epoch 362/400, Validation Loss: 0.2052755281329155\n",
            "Epoch 363/400, Validation Loss: 0.20538665354251862\n",
            "Epoch 364/400, Validation Loss: 0.2053193300962448\n",
            "Epoch 365/400, Validation Loss: 0.20510851219296455\n",
            "Epoch 366/400, Validation Loss: 0.20519941300153732\n",
            "Epoch 367/400, Validation Loss: 0.2053227797150612\n",
            "Epoch 368/400, Validation Loss: 0.20553507655858994\n",
            "Epoch 369/400, Validation Loss: 0.2050330564379692\n",
            "Epoch 370/400, Validation Loss: 0.20532245934009552\n",
            "Epoch 371/400, Validation Loss: 0.20519859716296196\n",
            "Epoch 372/400, Validation Loss: 0.20483414828777313\n",
            "Epoch 373/400, Validation Loss: 0.20566948875784874\n",
            "Epoch 374/400, Validation Loss: 0.20498770475387573\n",
            "Epoch 375/400, Validation Loss: 0.20496272295713425\n",
            "Epoch 376/400, Validation Loss: 0.20518285036087036\n",
            "Epoch 377/400, Validation Loss: 0.2052507996559143\n",
            "Epoch 378/400, Validation Loss: 0.20468560978770256\n",
            "Epoch 379/400, Validation Loss: 0.20557792112231255\n",
            "Epoch 380/400, Validation Loss: 0.20500096306204796\n",
            "Epoch 381/400, Validation Loss: 0.20514264330267906\n",
            "Epoch 382/400, Validation Loss: 0.20518002286553383\n",
            "Epoch 383/400, Validation Loss: 0.20539512485265732\n",
            "Epoch 384/400, Validation Loss: 0.20521756634116173\n",
            "Epoch 385/400, Validation Loss: 0.20501062273979187\n",
            "Epoch 386/400, Validation Loss: 0.20494763180613518\n",
            "Epoch 387/400, Validation Loss: 0.20541690662503242\n",
            "Epoch 388/400, Validation Loss: 0.20518231019377708\n",
            "Epoch 389/400, Validation Loss: 0.20533397793769836\n",
            "Epoch 390/400, Validation Loss: 0.20526932179927826\n",
            "Epoch 391/400, Validation Loss: 0.20531444624066353\n",
            "Epoch 392/400, Validation Loss: 0.20505471155047417\n",
            "Epoch 393/400, Validation Loss: 0.2048446238040924\n",
            "Epoch 394/400, Validation Loss: 0.20483673736453056\n",
            "Epoch 395/400, Validation Loss: 0.2050035446882248\n",
            "Epoch 396/400, Validation Loss: 0.20511585101485252\n",
            "Epoch 397/400, Validation Loss: 0.20524784922599792\n",
            "Epoch 398/400, Validation Loss: 0.20468095317482948\n",
            "Epoch 399/400, Validation Loss: 0.20588639006018639\n",
            "Epoch 400/400, Validation Loss: 0.20484381169080734\n"
          ]
        }
      ],
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim\n",
        "from torch.utils.data import DataLoader, TensorDataset\n",
        "from sklearn.model_selection import train_test_split\n",
        "import numpy as np\n",
        "\n",
        "seed_everything(args.seed)\n",
        "batch_size = 200\n",
        "\n",
        "train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
        "val_loader = DataLoader(val_dataset, batch_size=batch_size)\n",
        "test_loader = DataLoader(dev_dataset, batch_size=batch_size)\n",
        "\n",
        "# Define the neural network model\n",
        "class MultiLabelClassifier(nn.Module):\n",
        "    def __init__(self, input_size, output_size):\n",
        "        super(MultiLabelClassifier, self).__init__()\n",
        "        self.fc = nn.Linear(input_size, output_size)\n",
        "        self.sigmoid = nn.Sigmoid()\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.fc(x)\n",
        "        x = self.sigmoid(x)\n",
        "        return x\n",
        "\n",
        "# Initialize the model, loss function, and optimizer\n",
        "input_size = 768  # Assuming 768-dimensional text embeddings\n",
        "output_size = 20  # Number of labels\n",
        "model = MultiLabelClassifier(input_size, output_size)\n",
        "criterion = nn.BCELoss()\n",
        "optimizer = optim.Adam(model.parameters(), lr=0.001)\n",
        "\n",
        "# Train the model\n",
        "num_epochs = 400\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
        "model.to(device)\n",
        "\n",
        "train_losses = []\n",
        "val_losses = []\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "    model.train()\n",
        "    for inputs, labels in train_loader:\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        optimizer.zero_grad()\n",
        "        outputs = model(inputs)\n",
        "        loss = criterion(outputs, labels)\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "\n",
        "    # Validate the model\n",
        "    model.eval()\n",
        "    with torch.no_grad():\n",
        "        val_loss = 0.0\n",
        "        for inputs, labels in val_loader:\n",
        "            inputs, labels = inputs.to(device), labels.to(device)\n",
        "            outputs = model(inputs)\n",
        "            val_loss += criterion(outputs, labels).item()\n",
        "\n",
        "        val_loss /= len(val_loader)\n",
        "        train_losses.append(loss.item())\n",
        "        val_losses.append(val_loss)\n",
        "        print(f\"Epoch {epoch+1}/{num_epochs}, Validation Loss: {val_loss}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Lsb_6oX6aJnM",
        "outputId": "52a15fe7-7a6c-4b76-8103-fe7490af7a1f"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Plot the train and validation loss values\n",
        "plt.plot(range(1, num_epochs+1), train_losses, label='Train Loss')\n",
        "plt.plot(range(1, num_epochs+1), val_losses, label='Validation Loss')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LqyBTQzQaJnM"
      },
      "source": [
        "# Error Analysis"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YeRCX72-aJnN",
        "outputId": "c40ebfbf-8f13-445a-9049-4e85dc3d9bfc"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(['Obfuscation, Intentional vagueness, Confusion',\n",
              "  'Whataboutism',\n",
              "  'Presenting Irrelevant Data (Red Herring)',\n",
              "  \"Misrepresentation of Someone's Position (Straw Man)\",\n",
              "  'Causal Oversimplification',\n",
              "  'Smears',\n",
              "  'Appeal to fear/prejudice',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Slogans',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Repetition',\n",
              "  'Reductio ad hitlerum',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Bandwagon',\n",
              "  'Appeal to authority',\n",
              "  'Doubt',\n",
              "  'Glittering generalities (Virtue)'],\n",
              " 20)"
            ]
          },
          "execution_count": 51,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "leaf_labels[0], len(leaf_labels[0])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Gw5IfqCOaJnN",
        "outputId": "03aabd65-1f9b-4126-ea0d-18f79bbba3e3"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Confusion Matrix for 'Obfuscation, Intentional vagueness, Confusion':\n",
            "[[498   0]\n",
            " [  2   0]]\n",
            "Confusion Matrix for 'Whataboutism':\n",
            "[[479   0]\n",
            " [ 21   0]]\n",
            "Confusion Matrix for 'Presenting Irrelevant Data (Red Herring)':\n",
            "[[496   0]\n",
            " [  4   0]]\n",
            "Confusion Matrix for 'Misrepresentation of Someone's Position (Straw Man)':\n",
            "[[496   0]\n",
            " [  4   0]]\n",
            "Confusion Matrix for 'Causal Oversimplification':\n",
            "[[479   0]\n",
            " [ 21   0]]\n",
            "Confusion Matrix for 'Smears':\n",
            "[[330  28]\n",
            " [104  38]]\n",
            "Confusion Matrix for 'Appeal to fear/prejudice':\n",
            "[[473   0]\n",
            " [ 27   0]]\n",
            "Confusion Matrix for 'Black-and-white Fallacy/Dictatorship':\n",
            "[[444   3]\n",
            " [ 52   1]]\n",
            "Confusion Matrix for 'Thought-terminating cliché':\n",
            "[[459   3]\n",
            " [ 37   1]]\n",
            "Confusion Matrix for 'Name calling/Labeling':\n",
            "[[376   8]\n",
            " [100  16]]\n",
            "Confusion Matrix for 'Slogans':\n",
            "[[447   3]\n",
            " [ 50   0]]\n",
            "Confusion Matrix for 'Flag-waving':\n",
            "[[455   3]\n",
            " [ 37   5]]\n",
            "Confusion Matrix for 'Loaded Language':\n",
            "[[339  26]\n",
            " [104  31]]\n",
            "Confusion Matrix for 'Repetition':\n",
            "[[477   0]\n",
            " [ 22   1]]\n",
            "Confusion Matrix for 'Reductio ad hitlerum':\n",
            "[[496   0]\n",
            " [  4   0]]\n",
            "Confusion Matrix for 'Exaggeration/Minimisation':\n",
            "[[473   0]\n",
            " [ 27   0]]\n",
            "Confusion Matrix for 'Bandwagon':\n",
            "[[493   0]\n",
            " [  7   0]]\n",
            "Confusion Matrix for 'Appeal to authority':\n",
            "[[426  11]\n",
            " [ 28  35]]\n",
            "Confusion Matrix for 'Doubt':\n",
            "[[474   2]\n",
            " [ 24   0]]\n",
            "Confusion Matrix for 'Glittering generalities (Virtue)':\n",
            "[[457   7]\n",
            " [ 34   2]]\n"
          ]
        }
      ],
      "source": [
        "import torch\n",
        "from sklearn.metrics import multilabel_confusion_matrix\n",
        "import numpy as np\n",
        "\n",
        "# Set the model to evaluation mode\n",
        "model.eval()\n",
        "\n",
        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
        "\n",
        "all_predictions = []\n",
        "all_labels = []\n",
        "\n",
        "with torch.no_grad():\n",
        "    for inputs, labels in test_loader:\n",
        "        inputs, labels = inputs.to(device), labels.to(device)\n",
        "        outputs = model(inputs)\n",
        "\n",
        "        # Convert the outputs and labels to numpy arrays\n",
        "        predictions = (outputs.cpu().numpy() > 0.5).astype(int)\n",
        "        labels_np = labels.cpu().numpy()\n",
        "\n",
        "        all_predictions.append(predictions)\n",
        "        all_labels.append(labels_np)\n",
        "\n",
        "# Concatenate predictions and labels along the samples axis\n",
        "all_predictions = np.concatenate(all_predictions, axis=0)\n",
        "all_labels = np.concatenate(all_labels, axis=0)\n",
        "\n",
        "# Compute the multilabel confusion matrix\n",
        "confusion_matrices = multilabel_confusion_matrix(all_labels, all_predictions)\n",
        "\n",
        "# Print the confusion matrices\n",
        "for i, matrix in enumerate(confusion_matrices):\n",
        "    print(f\"Confusion Matrix for '{leaf_labels[0][i]}':\\n{matrix}\")\n",
        "\n",
        "# In a multilabel confusion matrix:\n",
        "# - true negatives count is at 00,\n",
        "# - false negatives count is at 10,\n",
        "# - true positives count is at 11\n",
        "# - false positives count is at 01.\n",
        "# see: https://scikit-learn.org/stable/modules/generated/sklearn.metrics.multilabel_confusion_matrix.html"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "a7SPBIdNaJnN",
        "outputId": "88465acd-9537-4335-ff98-598b0855159d"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "((500, 20), (500, 20))"
            ]
          },
          "execution_count": 44,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "all_predictions.shape, all_labels.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "6lkBQ8_OaJnN",
        "outputId": "3aac66eb-ab02-4dd1-9aa2-60b44cef9ec2"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "[('Slogans', 'Doubt'),\n",
              " ('Smears', 'Exaggeration/Minimisation'),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " ('Thought-terminating cliché', 'Slogans', 'Appeal to authority'),\n",
              " ('Name calling/Labeling', 'Flag-waving', 'Loaded Language'),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Flag-waving',),\n",
              " ('Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Smears',),\n",
              " ('Thought-terminating cliché', 'Name calling/Labeling', 'Loaded Language'),\n",
              " ('Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Smears',),\n",
              " ('Smears', 'Name calling/Labeling', 'Glittering generalities (Virtue)'),\n",
              " ('Name calling/Labeling', 'Loaded Language', 'Doubt'),\n",
              " ('Loaded Language', 'Doubt'),\n",
              " ('Slogans',),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Repetition', 'Appeal to authority'),\n",
              " ('Doubt',),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Loaded Language',),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (\"Misrepresentation of Someone's Position (Straw Man)\",),\n",
              " ('Slogans', 'Flag-waving', 'Doubt'),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Smears', 'Exaggeration/Minimisation'),\n",
              " ('Thought-terminating cliché', 'Repetition'),\n",
              " ('Doubt',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " ('Smears', 'Loaded Language', 'Reductio ad hitlerum'),\n",
              " ('Doubt',),\n",
              " ('Smears',),\n",
              " ('Appeal to authority',),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Whataboutism', 'Loaded Language'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Smears', 'Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling', 'Slogans'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Thought-terminating cliché',\n",
              "  'Loaded Language',\n",
              "  'Repetition',\n",
              "  'Appeal to authority'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Loaded Language',),\n",
              " ('Exaggeration/Minimisation',),\n",
              " ('Name calling/Labeling', 'Slogans'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Thought-terminating cliché', 'Glittering generalities (Virtue)'),\n",
              " ('Appeal to authority',),\n",
              " ('Smears', 'Black-and-white Fallacy/Dictatorship', 'Name calling/Labeling'),\n",
              " ('Appeal to fear/prejudice',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority',\n",
              "  'Doubt'),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Slogans', 'Flag-waving', 'Loaded Language'),\n",
              " ('Whataboutism', 'Loaded Language'),\n",
              " ('Appeal to authority',),\n",
              " ('Whataboutism', 'Smears', 'Name calling/Labeling'),\n",
              " ('Appeal to fear/prejudice', 'Name calling/Labeling'),\n",
              " ('Whataboutism',),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Whataboutism', 'Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " ('Flag-waving',),\n",
              " ('Slogans',),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Smears', 'Slogans', 'Flag-waving', 'Appeal to authority'),\n",
              " ('Thought-terminating cliché',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Slogans',),\n",
              " ('Doubt',),\n",
              " ('Causal Oversimplification',\n",
              "  'Thought-terminating cliché',\n",
              "  'Appeal to authority'),\n",
              " (),\n",
              " ('Whataboutism',),\n",
              " ('Appeal to authority',),\n",
              " ('Smears',),\n",
              " ('Appeal to fear/prejudice',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Loaded Language'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Exaggeration/Minimisation'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Smears',),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Appeal to authority'),\n",
              " ('Flag-waving', 'Appeal to authority', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Name calling/Labeling', 'Slogans'),\n",
              " ('Slogans',),\n",
              " ('Slogans', 'Bandwagon'),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " ('Thought-terminating cliché', 'Loaded Language'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Slogans', 'Exaggeration/Minimisation'),\n",
              " ('Presenting Irrelevant Data (Red Herring)', 'Appeal to fear/prejudice'),\n",
              " ('Whataboutism',),\n",
              " ('Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Flag-waving',),\n",
              " ('Slogans',),\n",
              " ('Doubt',),\n",
              " ('Whataboutism', 'Smears', 'Loaded Language'),\n",
              " ('Causal Oversimplification',\n",
              "  'Appeal to fear/prejudice',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling'),\n",
              " ('Slogans',),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " ('Smears',\n",
              "  'Loaded Language',\n",
              "  'Repetition',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Doubt',),\n",
              " ('Smears', 'Repetition'),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Appeal to authority'),\n",
              " ('Whataboutism', \"Misrepresentation of Someone's Position (Straw Man)\"),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Loaded Language',),\n",
              " ('Smears', 'Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " ('Slogans',),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Doubt',),\n",
              " (),\n",
              " ('Thought-terminating cliché', 'Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving',),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears', 'Thought-terminating cliché', 'Loaded Language'),\n",
              " ('Smears', 'Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " ('Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Smears', 'Name calling/Labeling', 'Flag-waving', 'Reductio ad hitlerum'),\n",
              " (),\n",
              " ('Flag-waving',),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Slogans'),\n",
              " ('Smears',),\n",
              " ('Whataboutism',\n",
              "  'Smears',\n",
              "  'Name calling/Labeling',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Whataboutism', 'Repetition', 'Appeal to authority'),\n",
              " ('Name calling/Labeling', 'Loaded Language', 'Doubt'),\n",
              " ('Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Smears', 'Appeal to fear/prejudice', 'Name calling/Labeling'),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Loaded Language',),\n",
              " ('Smears',),\n",
              " ('Thought-terminating cliché', 'Slogans', 'Repetition'),\n",
              " ('Smears',),\n",
              " ('Causal Oversimplification', 'Repetition', 'Appeal to authority'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Name calling/Labeling'),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Name calling/Labeling'),\n",
              " ('Appeal to fear/prejudice', 'Loaded Language'),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Thought-terminating cliché', 'Doubt'),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Flag-waving'),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Appeal to fear/prejudice',),\n",
              " ('Appeal to authority',),\n",
              " ('Smears', 'Appeal to authority', 'Doubt'),\n",
              " ('Smears',),\n",
              " ('Slogans',),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language', 'Repetition'),\n",
              " (),\n",
              " (),\n",
              " ('Causal Oversimplification', 'Appeal to fear/prejudice'),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " ('Causal Oversimplification', 'Smears', 'Loaded Language'),\n",
              " (),\n",
              " ('Flag-waving', 'Loaded Language'),\n",
              " ('Name calling/Labeling', 'Doubt'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Smears', 'Repetition', 'Exaggeration/Minimisation'),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " (),\n",
              " ('Exaggeration/Minimisation', 'Glittering generalities (Virtue)'),\n",
              " ('Loaded Language', 'Doubt'),\n",
              " ('Loaded Language',),\n",
              " ('Smears',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " ('Flag-waving', 'Doubt'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Flag-waving', 'Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Appeal to authority',),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Smears',),\n",
              " ('Flag-waving', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Doubt',),\n",
              " ('Thought-terminating cliché',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Appeal to fear/prejudice',),\n",
              " ('Thought-terminating cliché', 'Name calling/Labeling', 'Slogans'),\n",
              " ('Appeal to fear/prejudice', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Causal Oversimplification', 'Loaded Language'),\n",
              " ('Smears', 'Name calling/Labeling', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " ('Smears',),\n",
              " ('Thought-terminating cliché',\n",
              "  'Flag-waving',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Slogans', 'Loaded Language'),\n",
              " ('Slogans',),\n",
              " ('Repetition', 'Exaggeration/Minimisation'),\n",
              " ('Causal Oversimplification', 'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Whataboutism', 'Name calling/Labeling'),\n",
              " ('Smears', 'Loaded Language', 'Repetition'),\n",
              " ('Name calling/Labeling', 'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Slogans',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " ('Smears', 'Black-and-white Fallacy/Dictatorship'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Smears',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Repetition'),\n",
              " ('Smears',),\n",
              " ('Smears',),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Appeal to authority'),\n",
              " ('Flag-waving',),\n",
              " ('Smears',\n",
              "  'Appeal to fear/prejudice',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Smears', 'Slogans', 'Bandwagon'),\n",
              " ('Bandwagon', 'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Smears',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Appeal to fear/prejudice', 'Black-and-white Fallacy/Dictatorship'),\n",
              " ('Flag-waving', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " ('Causal Oversimplification', 'Loaded Language'),\n",
              " (\"Misrepresentation of Someone's Position (Straw Man)\",\n",
              "  'Smears',\n",
              "  'Loaded Language'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Whataboutism', 'Smears'),\n",
              " ('Smears',),\n",
              " ('Flag-waving',),\n",
              " ('Smears',),\n",
              " ('Smears',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling', 'Exaggeration/Minimisation'),\n",
              " ('Slogans', 'Bandwagon'),\n",
              " (),\n",
              " ('Causal Oversimplification',),\n",
              " ('Repetition',),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Slogans', 'Flag-waving'),\n",
              " ('Smears',),\n",
              " ('Appeal to fear/prejudice',),\n",
              " (),\n",
              " ('Appeal to fear/prejudice', 'Loaded Language'),\n",
              " ('Name calling/Labeling', 'Flag-waving'),\n",
              " ('Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Appeal to fear/prejudice', 'Thought-terminating cliché', 'Loaded Language'),\n",
              " ('Smears', 'Slogans', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Slogans'),\n",
              " ('Smears',),\n",
              " ('Appeal to fear/prejudice',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Flag-waving', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling', 'Exaggeration/Minimisation'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Smears', 'Black-and-white Fallacy/Dictatorship'),\n",
              " ('Slogans', 'Glittering generalities (Virtue)'),\n",
              " ('Presenting Irrelevant Data (Red Herring)',\n",
              "  'Smears',\n",
              "  'Name calling/Labeling'),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " ('Smears',),\n",
              " ('Loaded Language',),\n",
              " ('Smears',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Smears', 'Reductio ad hitlerum'),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Appeal to authority'),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Loaded Language'),\n",
              " ('Slogans', 'Loaded Language'),\n",
              " (),\n",
              " ('Whataboutism', 'Name calling/Labeling', 'Loaded Language', 'Repetition'),\n",
              " ('Causal Oversimplification',\n",
              "  'Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Flag-waving',),\n",
              " ('Appeal to fear/prejudice',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Smears', 'Thought-terminating cliché', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Causal Oversimplification', 'Loaded Language', 'Repetition'),\n",
              " ('Bandwagon', 'Appeal to authority'),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Name calling/Labeling'),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " ('Loaded Language',),\n",
              " ('Loaded Language',),\n",
              " ('Causal Oversimplification',\n",
              "  'Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Repetition'),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " ('Obfuscation, Intentional vagueness, Confusion',),\n",
              " ('Presenting Irrelevant Data (Red Herring)',\n",
              "  'Smears',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Slogans', 'Glittering generalities (Virtue)'),\n",
              " ('Causal Oversimplification',),\n",
              " ('Smears',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " (),\n",
              " ('Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Slogans', 'Loaded Language'),\n",
              " ('Slogans',),\n",
              " ('Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Appeal to fear/prejudice', 'Loaded Language', 'Doubt'),\n",
              " ('Slogans',),\n",
              " ('Flag-waving',),\n",
              " ('Slogans',),\n",
              " ('Slogans',),\n",
              " ('Smears',),\n",
              " ('Appeal to fear/prejudice', 'Thought-terminating cliché', 'Loaded Language'),\n",
              " ('Causal Oversimplification', 'Smears', 'Appeal to fear/prejudice'),\n",
              " ('Thought-terminating cliché', 'Doubt'),\n",
              " ('Obfuscation, Intentional vagueness, Confusion',),\n",
              " ('Smears',),\n",
              " ('Smears',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " (),\n",
              " ('Causal Oversimplification', 'Appeal to fear/prejudice', 'Loaded Language'),\n",
              " ('Slogans', 'Loaded Language'),\n",
              " ('Whataboutism',\n",
              "  'Thought-terminating cliché',\n",
              "  'Loaded Language',\n",
              "  'Repetition'),\n",
              " ('Causal Oversimplification',),\n",
              " (),\n",
              " ('Whataboutism',\n",
              "  'Appeal to fear/prejudice',\n",
              "  'Name calling/Labeling',\n",
              "  'Repetition'),\n",
              " ('Exaggeration/Minimisation', 'Bandwagon'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Flag-waving', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Whataboutism', 'Smears', 'Name calling/Labeling', 'Repetition'),\n",
              " ('Causal Oversimplification',\n",
              "  'Smears',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Slogans', 'Loaded Language'),\n",
              " ('Slogans', 'Loaded Language'),\n",
              " ('Thought-terminating cliché', 'Flag-waving'),\n",
              " ('Slogans', 'Flag-waving'),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Smears',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Name calling/Labeling', 'Slogans'),\n",
              " ('Smears', 'Appeal to fear/prejudice'),\n",
              " ('Causal Oversimplification', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Appeal to fear/prejudice', 'Loaded Language'),\n",
              " ('Presenting Irrelevant Data (Red Herring)',\n",
              "  'Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Thought-terminating cliché', 'Appeal to authority'),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " (\"Misrepresentation of Someone's Position (Straw Man)\",\n",
              "  'Smears',\n",
              "  'Name calling/Labeling'),\n",
              " ('Slogans', 'Bandwagon'),\n",
              " ('Whataboutism',),\n",
              " ('Whataboutism', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling', 'Glittering generalities (Virtue)'),\n",
              " ('Flag-waving',),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Repetition'),\n",
              " ('Smears',),\n",
              " ('Smears', 'Thought-terminating cliché'),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Appeal to authority'),\n",
              " ('Causal Oversimplification',\n",
              "  'Appeal to fear/prejudice',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Smears',\n",
              "  'Name calling/Labeling',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Loaded Language', 'Glittering generalities (Virtue)'),\n",
              " ('Whataboutism', 'Loaded Language', 'Doubt'),\n",
              " ('Doubt',),\n",
              " ('Appeal to fear/prejudice',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language'),\n",
              " ('Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " ('Loaded Language',),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " ('Smears', 'Exaggeration/Minimisation'),\n",
              " ('Whataboutism', 'Loaded Language'),\n",
              " ('Smears',),\n",
              " ('Doubt',),\n",
              " (),\n",
              " ('Loaded Language',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Appeal to authority'),\n",
              " ('Loaded Language', 'Reductio ad hitlerum'),\n",
              " ('Thought-terminating cliché', 'Name calling/Labeling', 'Loaded Language'),\n",
              " ('Smears',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation'),\n",
              " ('Smears',\n",
              "  'Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Flag-waving',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Smears',),\n",
              " ('Flag-waving',),\n",
              " ('Smears',),\n",
              " ('Causal Oversimplification',\n",
              "  'Loaded Language',\n",
              "  'Exaggeration/Minimisation',\n",
              "  'Appeal to authority'),\n",
              " ('Slogans', 'Flag-waving'),\n",
              " ('Name calling/Labeling',),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " (),\n",
              " ('Smears', 'Repetition'),\n",
              " ('Name calling/Labeling', 'Slogans', 'Repetition'),\n",
              " (),\n",
              " (),\n",
              " ('Causal Oversimplification', 'Loaded Language'),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Name calling/Labeling',\n",
              "  'Appeal to authority')]"
            ]
          },
          "execution_count": 46,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "mlb.inverse_transform(all_labels)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "zBe_GYODaJnN",
        "outputId": "6ab1478a-a03d-4d15-f056-947d729a9bca"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "[('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Slogans',),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " ('Appeal to authority',),\n",
              " ('Smears',),\n",
              " ('Smears',),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language', 'Repetition', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Slogans',),\n",
              " ('Loaded Language',),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " ('Smears',),\n",
              " ('Smears',),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Name calling/Labeling',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Slogans', 'Flag-waving'),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Doubt',),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving', 'Appeal to authority', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving',),\n",
              " (),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Appeal to authority'),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " ('Doubt',),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving', 'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " ('Thought-terminating cliché',),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling'),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Flag-waving', 'Appeal to authority'),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " ('Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Thought-terminating cliché'),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',\n",
              "  'Thought-terminating cliché',\n",
              "  'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Appeal to authority'),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',\n",
              "  'Appeal to authority',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " ('Loaded Language',),\n",
              " ('Smears', 'Appeal to authority'),\n",
              " ('Smears',),\n",
              " ('Smears',\n",
              "  'Thought-terminating cliché',\n",
              "  'Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Appeal to authority'),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Name calling/Labeling',\n",
              "  'Loaded Language',\n",
              "  'Glittering generalities (Virtue)'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship', 'Appeal to authority'),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " ('Flag-waving',),\n",
              " ('Smears', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " (),\n",
              " ('Smears',),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " ('Smears',),\n",
              " ('Smears',),\n",
              " (),\n",
              " ('Glittering generalities (Virtue)',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',),\n",
              " ('Loaded Language',),\n",
              " ('Flag-waving',),\n",
              " (),\n",
              " (),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " ('Smears', 'Name calling/Labeling', 'Loaded Language'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Loaded Language', 'Appeal to authority'),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Black-and-white Fallacy/Dictatorship',),\n",
              " (),\n",
              " (),\n",
              " (),\n",
              " ('Appeal to authority',)]"
            ]
          },
          "execution_count": 45,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "mlb.inverse_transform(all_predictions)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "6o0XDMspaJnO",
        "outputId": "986fe0f1-c5d6-46d7-e655-dc873f42e0e8"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "<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>Text</th>\n",
              "      <th>Actual</th>\n",
              "      <th>Predicted</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Critical Thinking Essentials\\n\\Are my biases a...</td>\n",
              "      <td>(Slogans, Doubt)</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Trying to think of a single accomplishment...</td>\n",
              "      <td>(Smears, Exaggeration/Minimisation)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Corporate needs you to find the difference bet...</td>\n",
              "      <td>(Thought-terminating cliché,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>KYLE RITTENHOUSE\\nALL CHARGES\\nNOT GUILTY</td>\n",
              "      <td>(Glittering generalities (Virtue),)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Al Franken explains why America should tax the...</td>\n",
              "      <td>(Thought-terminating cliché, Slogans, Appeal t...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>\\ I left my parents. I left my brothers and si...</td>\n",
              "      <td>(Name calling/Labeling, Flag-waving, Loaded La...</td>\n",
              "      <td>(Loaded Language, Appeal to authority)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>Grandson: Grandpa, whats that thing on the wal...</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>President Obama,\\nWe Built It-You Broke It - W...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Slogans,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>A SINGLE-PHOTO DEPICTION\\n\\nOF HOW MUCH PRESID...</td>\n",
              "      <td>(Flag-waving,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>ORWELL CALLED THEM \\THOUGHT POLICE\\\\n\\n\\nBIG T...</td>\n",
              "      <td>(Name calling/Labeling, Appeal to authority)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>\\I AM TIRED OF WATCHING POLITICIANS SCREW UP T...</td>\n",
              "      <td>(Loaded Language, Appeal to authority)</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>WE WILL LEAVE THE COUNTRY IF TRUMP IS ELECTED\\...</td>\n",
              "      <td>(Black-and-white Fallacy/Dictatorship,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>Everyone is all excited that after 4 years the...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>IT'S A SAD DAY IN AMERICA\\n\\nWHEN A BIGOTED, R...</td>\n",
              "      <td>(Thought-terminating cliché, Name calling/Labe...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Language)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>\\This is the most important election in the hi...</td>\n",
              "      <td>(Name calling/Labeling, Flag-waving, Exaggerat...</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>My Dad got to watch Armstrong walk\\non the moo...</td>\n",
              "      <td>(Name calling/Labeling,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>ALEXANDRA OCASIO-CORTEZ\\n\\nWAS FIRED FROM \\HOT...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>The composure, poise and grace that Ketanji Br...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Glittering gen...</td>\n",
              "      <td>(Name calling/Labeling,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>BILL CLINTON SPEAKS AT DNC WEEKS AFTER COURT\\n...</td>\n",
              "      <td>(Name calling/Labeling, Loaded Language, Doubt)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>BREAKING:\\nOSHA JUST SUSPENDED THE ENFORCEMENT...</td>\n",
              "      <td>(Loaded Language, Doubt)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>I STAND WITH MIKE LINDELL\\n\\n1. VISIT MYPILLOW...</td>\n",
              "      <td>(Slogans,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>That face you make when...\\n\\nTrump opens his ...</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>\\[C]lass struggle is not merely an economic st...</td>\n",
              "      <td>(Name calling/Labeling, Repetition, Appeal to ...</td>\n",
              "      <td>(Loaded Language, Repetition, Appeal to author...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>OBAMA LOVES AMERICA LIKE OJ LOVED NICOLE</td>\n",
              "      <td>(Doubt,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>24</th>\n",
              "      <td>YOU CANNOT WIN ON WEDNESDAY\\n\\nBECAUSE WE HAVE...</td>\n",
              "      <td>(Thought-terminating cliché,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>Thank God California banned plastic straws\\n</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>This is where the My Pillow guy could actually...</td>\n",
              "      <td>(Black-and-white Fallacy/Dictatorship,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>Who sank the Titanic?\\n\\nA pro-Russian submari...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Language)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>28</th>\n",
              "      <td>Every Day I thank God For These Two Men\\n\\nWin...</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>29</th>\n",
              "      <td>Stay with me.\\n\\nTogether we can weather any s...</td>\n",
              "      <td>(Black-and-white Fallacy/Dictatorship, Loaded ...</td>\n",
              "      <td>(Slogans,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>30</th>\n",
              "      <td>when the one with the rifle get killed, the on...</td>\n",
              "      <td>()</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>Live: Russia says first phase of invasion almo...</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>SO IS THE DEFENDAT A PEDOPHILE?\\n\\nI DON'T KNO...</td>\n",
              "      <td>(Misrepresentation of Someone's Position (Stra...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>33</th>\n",
              "      <td>$10 MILLION A DAY TO ISRAEL?\\n\\nOur money is n...</td>\n",
              "      <td>(Slogans, Flag-waving, Doubt)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>34</th>\n",
              "      <td>THAT WAS THE ONLY HACKING USAW DURING THE ELEC...</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>35</th>\n",
              "      <td>STILL LOOKING FOR ANSWERS\\n\\nEXTORTION 17\\nDEM...</td>\n",
              "      <td>(Black-and-white Fallacy/Dictatorship,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>BIDEN\\nGARBAGE\\nREMOVAL</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>37</th>\n",
              "      <td>The England Football team visited an orphanage...</td>\n",
              "      <td>(Smears, Exaggeration/Minimisation)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>38</th>\n",
              "      <td>He doesn't read.\\nHe doesn't think.\\nHe doesn'...</td>\n",
              "      <td>(Thought-terminating cliché, Repetition)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>39</th>\n",
              "      <td>ONLY TWO HOUSE DEMOCRATS SHOW UP FOR\\nIMPEACH ...</td>\n",
              "      <td>(Doubt,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>40</th>\n",
              "      <td>COPE CAGE\\n\\nCOPE PYRAMIDS</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>41</th>\n",
              "      <td>I Stand With My President!\\n\\n100%</td>\n",
              "      <td>(Slogans,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>42</th>\n",
              "      <td>OBAMA AND PUTIN ARE BOTH WORKING TOWARD THE SA...</td>\n",
              "      <td>(Smears, Loaded Language, Reductio ad hitlerum)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>43</th>\n",
              "      <td>To manufacture each EV auto battery, you must ...</td>\n",
              "      <td>(Doubt,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>44</th>\n",
              "      <td>DONALD TRUMP\\nIS A\\nRA IST\\nC P\\n</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>45</th>\n",
              "      <td>Michelle Obama says there's 'zero chance' she'...</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>46</th>\n",
              "      <td>DON'T WAIT.\\nVACCINATE.\\nWE WERE ALL STUPID YO...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>47</th>\n",
              "      <td>LET'S BE HONEST\\nTHE MEDIA WOULD HAVE TOLD THE...</td>\n",
              "      <td>(Whataboutism, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>Russia missile strikes that led to 7 of 11 pen...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49</th>\n",
              "      <td>HILL RY \\nS CKS\\n</td>\n",
              "      <td>()</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                 Text  \\\n",
              "0   Critical Thinking Essentials\\n\\Are my biases a...   \n",
              "1       Trying to think of a single accomplishment...   \n",
              "2   Corporate needs you to find the difference bet...   \n",
              "3           KYLE RITTENHOUSE\\nALL CHARGES\\nNOT GUILTY   \n",
              "4   Al Franken explains why America should tax the...   \n",
              "5   \\ I left my parents. I left my brothers and si...   \n",
              "6   Grandson: Grandpa, whats that thing on the wal...   \n",
              "7   President Obama,\\nWe Built It-You Broke It - W...   \n",
              "8   A SINGLE-PHOTO DEPICTION\\n\\nOF HOW MUCH PRESID...   \n",
              "9   ORWELL CALLED THEM \\THOUGHT POLICE\\\\n\\n\\nBIG T...   \n",
              "10  \\I AM TIRED OF WATCHING POLITICIANS SCREW UP T...   \n",
              "11  WE WILL LEAVE THE COUNTRY IF TRUMP IS ELECTED\\...   \n",
              "12  Everyone is all excited that after 4 years the...   \n",
              "13  IT'S A SAD DAY IN AMERICA\\n\\nWHEN A BIGOTED, R...   \n",
              "14  \\This is the most important election in the hi...   \n",
              "15  My Dad got to watch Armstrong walk\\non the moo...   \n",
              "16  ALEXANDRA OCASIO-CORTEZ\\n\\nWAS FIRED FROM \\HOT...   \n",
              "17  The composure, poise and grace that Ketanji Br...   \n",
              "18  BILL CLINTON SPEAKS AT DNC WEEKS AFTER COURT\\n...   \n",
              "19  BREAKING:\\nOSHA JUST SUSPENDED THE ENFORCEMENT...   \n",
              "20  I STAND WITH MIKE LINDELL\\n\\n1. VISIT MYPILLOW...   \n",
              "21  That face you make when...\\n\\nTrump opens his ...   \n",
              "22  \\[C]lass struggle is not merely an economic st...   \n",
              "23           OBAMA LOVES AMERICA LIKE OJ LOVED NICOLE   \n",
              "24  YOU CANNOT WIN ON WEDNESDAY\\n\\nBECAUSE WE HAVE...   \n",
              "25       Thank God California banned plastic straws\\n   \n",
              "26  This is where the My Pillow guy could actually...   \n",
              "27  Who sank the Titanic?\\n\\nA pro-Russian submari...   \n",
              "28  Every Day I thank God For These Two Men\\n\\nWin...   \n",
              "29  Stay with me.\\n\\nTogether we can weather any s...   \n",
              "30  when the one with the rifle get killed, the on...   \n",
              "31  Live: Russia says first phase of invasion almo...   \n",
              "32  SO IS THE DEFENDAT A PEDOPHILE?\\n\\nI DON'T KNO...   \n",
              "33  $10 MILLION A DAY TO ISRAEL?\\n\\nOur money is n...   \n",
              "34  THAT WAS THE ONLY HACKING USAW DURING THE ELEC...   \n",
              "35  STILL LOOKING FOR ANSWERS\\n\\nEXTORTION 17\\nDEM...   \n",
              "36                            BIDEN\\nGARBAGE\\nREMOVAL   \n",
              "37  The England Football team visited an orphanage...   \n",
              "38  He doesn't read.\\nHe doesn't think.\\nHe doesn'...   \n",
              "39  ONLY TWO HOUSE DEMOCRATS SHOW UP FOR\\nIMPEACH ...   \n",
              "40                         COPE CAGE\\n\\nCOPE PYRAMIDS   \n",
              "41                 I Stand With My President!\\n\\n100%   \n",
              "42  OBAMA AND PUTIN ARE BOTH WORKING TOWARD THE SA...   \n",
              "43  To manufacture each EV auto battery, you must ...   \n",
              "44                  DONALD TRUMP\\nIS A\\nRA IST\\nC P\\n   \n",
              "45  Michelle Obama says there's 'zero chance' she'...   \n",
              "46  DON'T WAIT.\\nVACCINATE.\\nWE WERE ALL STUPID YO...   \n",
              "47  LET'S BE HONEST\\nTHE MEDIA WOULD HAVE TOLD THE...   \n",
              "48  Russia missile strikes that led to 7 of 11 pen...   \n",
              "49                                  HILL RY \\nS CKS\\n   \n",
              "\n",
              "                                               Actual  \\\n",
              "0                                    (Slogans, Doubt)   \n",
              "1                 (Smears, Exaggeration/Minimisation)   \n",
              "2                       (Thought-terminating cliché,)   \n",
              "3                 (Glittering generalities (Virtue),)   \n",
              "4   (Thought-terminating cliché, Slogans, Appeal t...   \n",
              "5   (Name calling/Labeling, Flag-waving, Loaded La...   \n",
              "6                                                  ()   \n",
              "7                                           (Smears,)   \n",
              "8                                      (Flag-waving,)   \n",
              "9        (Name calling/Labeling, Appeal to authority)   \n",
              "10             (Loaded Language, Appeal to authority)   \n",
              "11            (Black-and-white Fallacy/Dictatorship,)   \n",
              "12                                          (Smears,)   \n",
              "13  (Thought-terminating cliché, Name calling/Labe...   \n",
              "14  (Name calling/Labeling, Flag-waving, Exaggerat...   \n",
              "15                           (Name calling/Labeling,)   \n",
              "16                                          (Smears,)   \n",
              "17  (Smears, Name calling/Labeling, Glittering gen...   \n",
              "18    (Name calling/Labeling, Loaded Language, Doubt)   \n",
              "19                           (Loaded Language, Doubt)   \n",
              "20                                         (Slogans,)   \n",
              "21                                                 ()   \n",
              "22  (Name calling/Labeling, Repetition, Appeal to ...   \n",
              "23                                           (Doubt,)   \n",
              "24                      (Thought-terminating cliché,)   \n",
              "25                                 (Loaded Language,)   \n",
              "26            (Black-and-white Fallacy/Dictatorship,)   \n",
              "27   (Smears, Name calling/Labeling, Loaded Language)   \n",
              "28                                                 ()   \n",
              "29  (Black-and-white Fallacy/Dictatorship, Loaded ...   \n",
              "30                                                 ()   \n",
              "31                                 (Loaded Language,)   \n",
              "32  (Misrepresentation of Someone's Position (Stra...   \n",
              "33                      (Slogans, Flag-waving, Doubt)   \n",
              "34                                                 ()   \n",
              "35            (Black-and-white Fallacy/Dictatorship,)   \n",
              "36                                                 ()   \n",
              "37                (Smears, Exaggeration/Minimisation)   \n",
              "38           (Thought-terminating cliché, Repetition)   \n",
              "39                                           (Doubt,)   \n",
              "40                                                 ()   \n",
              "41                                         (Slogans,)   \n",
              "42    (Smears, Loaded Language, Reductio ad hitlerum)   \n",
              "43                                           (Doubt,)   \n",
              "44                                          (Smears,)   \n",
              "45                             (Appeal to authority,)   \n",
              "46                    (Smears, Name calling/Labeling)   \n",
              "47                    (Whataboutism, Loaded Language)   \n",
              "48                    (Smears, Name calling/Labeling)   \n",
              "49                                                 ()   \n",
              "\n",
              "                                            Predicted  \n",
              "0                              (Appeal to authority,)  \n",
              "1                                                  ()  \n",
              "2                                                  ()  \n",
              "3                                                  ()  \n",
              "4                                                  ()  \n",
              "5              (Loaded Language, Appeal to authority)  \n",
              "6                                                  ()  \n",
              "7                                          (Slogans,)  \n",
              "8                                                  ()  \n",
              "9                                                  ()  \n",
              "10                             (Appeal to authority,)  \n",
              "11                                                 ()  \n",
              "12                                          (Smears,)  \n",
              "13   (Smears, Name calling/Labeling, Loaded Language)  \n",
              "14                             (Appeal to authority,)  \n",
              "15                                          (Smears,)  \n",
              "16                                          (Smears,)  \n",
              "17                           (Name calling/Labeling,)  \n",
              "18                                                 ()  \n",
              "19                                                 ()  \n",
              "20                                                 ()  \n",
              "21                                                 ()  \n",
              "22  (Loaded Language, Repetition, Appeal to author...  \n",
              "23                                                 ()  \n",
              "24                                                 ()  \n",
              "25                                                 ()  \n",
              "26                                                 ()  \n",
              "27                                          (Smears,)  \n",
              "28                                                 ()  \n",
              "29                                         (Slogans,)  \n",
              "30                                 (Loaded Language,)  \n",
              "31                                 (Loaded Language,)  \n",
              "32                                                 ()  \n",
              "33                                                 ()  \n",
              "34                                                 ()  \n",
              "35                                                 ()  \n",
              "36                                                 ()  \n",
              "37                                                 ()  \n",
              "38                                                 ()  \n",
              "39                                                 ()  \n",
              "40                                                 ()  \n",
              "41                                                 ()  \n",
              "42                                                 ()  \n",
              "43                                                 ()  \n",
              "44                                                 ()  \n",
              "45                                                 ()  \n",
              "46                                                 ()  \n",
              "47                                                 ()  \n",
              "48                          (Smears, Loaded Language)  \n",
              "49                                                 ()  "
            ]
          },
          "execution_count": 89,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "_, dev, _ = load_dataset(args.dataset)\n",
        "\n",
        "# Creating a pandas DataFrame\n",
        "df_error_analysis = pd.DataFrame({'Text': dev.text, 'Actual': mlb.inverse_transform(all_labels), 'Predicted': mlb.inverse_transform(all_predictions)})\n",
        "\n",
        "# Displaying the DataFrame\n",
        "df_error_analysis.head(50)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dm0VCAgiaJnO",
        "outputId": "a0d046f4-c941-4127-d15d-8619477f0826"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(74, 3)\n"
          ]
        }
      ],
      "source": [
        "lbl_of_interest = 'Appeal to authority'\n",
        "# lbl_of_interest = 'Whataboutism'\n",
        "# lbl_of_interest = 'Obfuscation, Intentional vagueness, Confusion'\n",
        "\n",
        "# Selecting rows containing the label of interest in any column\n",
        "result_df = df_error_analysis[df_error_analysis.apply(lambda row: row.astype(str).str.contains(lbl_of_interest).any(), axis=1)]\n",
        "print(result_df.shape)\n",
        "result_df.head(50)\n",
        "result_df.to_csv('Appeal_to_authority.csv')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "oKTv9DU_aJnO",
        "outputId": "a981b31f-b8cc-4ae5-b847-c76fb512f851"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(142, 3)\n"
          ]
        },
        {
          "data": {
            "text/html": [
              "<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>Text</th>\n",
              "      <th>Actual</th>\n",
              "      <th>Predicted</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Trying to think of a single accomplishment...</td>\n",
              "      <td>(Smears, Exaggeration/Minimisation)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>President Obama,\\nWe Built It-You Broke It - W...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Slogans,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>Everyone is all excited that after 4 years the...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>ALEXANDRA OCASIO-CORTEZ\\n\\nWAS FIRED FROM \\HOT...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>The composure, poise and grace that Ketanji Br...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Glittering gen...</td>\n",
              "      <td>(Name calling/Labeling,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>Who sank the Titanic?\\n\\nA pro-Russian submari...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Language)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>37</th>\n",
              "      <td>The England Football team visited an orphanage...</td>\n",
              "      <td>(Smears, Exaggeration/Minimisation)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>42</th>\n",
              "      <td>OBAMA AND PUTIN ARE BOTH WORKING TOWARD THE SA...</td>\n",
              "      <td>(Smears, Loaded Language, Reductio ad hitlerum)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>44</th>\n",
              "      <td>DONALD TRUMP\\nIS A\\nRA IST\\nC P\\n</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>46</th>\n",
              "      <td>DON'T WAIT.\\nVACCINATE.\\nWE WERE ALL STUPID YO...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>Russia missile strikes that led to 7 of 11 pen...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>51</th>\n",
              "      <td>By now, many have seen the photo on the right....</td>\n",
              "      <td>(Smears, Black-and-white Fallacy/Dictatorship,...</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>52</th>\n",
              "      <td>Never forget this mountain was sacred to the l...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53</th>\n",
              "      <td>SAUDI ARABIA\\n\\nAS SOON AS THE G20 SPOTLIGHT O...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>58</th>\n",
              "      <td>Jacob Chansley, AKA Jake Angeli, Arizona man m...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>65</th>\n",
              "      <td>THIS IS DONALD.\\n\\nHe has,\\nThe appeal of an i...</td>\n",
              "      <td>(Smears, Black-and-white Fallacy/Dictatorship,...</td>\n",
              "      <td>(Name calling/Labeling,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>72</th>\n",
              "      <td>THE MODERN SOCIALIST\\n\\nTAKES AN UBER TO THE S...</td>\n",
              "      <td>(Whataboutism, Smears, Name calling/Labeling)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75</th>\n",
              "      <td>DONALD TRUMP: ILHAN OMAR 'EXTREMELY UNPATRIOTI...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Languag...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>76</th>\n",
              "      <td>REAL ESTATE BUBBLE BURSTING, BANKING CRISIS\\n\\...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>78</th>\n",
              "      <td>Russia demands from UN, Ukraine to return arms...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>84</th>\n",
              "      <td>Obama could learn from this speech.\\n\\nIn Russ...</td>\n",
              "      <td>(Smears, Slogans, Flag-waving, Appeal to autho...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>96</th>\n",
              "      <td>There is Hope Young Scientists are searching f...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Doubt,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>99</th>\n",
              "      <td>MISS UNIVERSE MISTAKE \\n\\nMEDIA FREAK OUT\\n\\nP...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Language)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>103</th>\n",
              "      <td>HEY DEMOCRATS: IF YOU RE-ELECT TRUMP.\\n\\nTHAT'...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>119</th>\n",
              "      <td>WELCOME TO THE 2041 ANNUAL\\n\\nMARICOPA COUNTY ...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Slogans, Exagg...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>126</th>\n",
              "      <td>DEMOCRAT CONVENTION:\\nMICHELLE O DECRIES KIDS ...</td>\n",
              "      <td>(Whataboutism, Smears, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>135</th>\n",
              "      <td>TALKING THE TALK\\n\\n5 children with 3 women\\nC...</td>\n",
              "      <td>(Smears, Loaded Language, Repetition, Glitteri...</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>136</th>\n",
              "      <td>\\ILLEGALLY SHOOT THE SON OF A B-\\\\n\\nHillary C...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Languag...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>137</th>\n",
              "      <td>Who else wants to impeach these three justices...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>138</th>\n",
              "      <td>Tomi Lahren, Ann Coulter, and Kellyanne Conway...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>142</th>\n",
              "      <td>Here's a great example of why I'm proud to be ...</td>\n",
              "      <td>(Smears, Repetition)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>147</th>\n",
              "      <td>The only upside to Trump passing his cognitive...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>149</th>\n",
              "      <td>\\REPUBLICANS:\\nTHE NO. 1 REASON THE REST OF TH...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Appeal to auth...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>152</th>\n",
              "      <td>\\Not one penny more to countries that are burn...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Appeal to auth...</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>154</th>\n",
              "      <td>IF PEOPLE ARE RISKING THEIR LIVES TO ESCAPE SO...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>161</th>\n",
              "      <td>WHEN YOU'RE SO WOKE.....\\n\\nTHAT A BLACKFEET C...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>169</th>\n",
              "      <td>The fact \\ISIS\\ has a twitter account should t...</td>\n",
              "      <td>(Smears, Thought-terminating cliché, Loaded La...</td>\n",
              "      <td>(Smears, Name calling/Labeling)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>170</th>\n",
              "      <td>\\I believe [Trump] knows he's an illegitimate ...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Appeal to auth...</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>174</th>\n",
              "      <td>Help Buy LeBron a Helicopter</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>177</th>\n",
              "      <td>WE ARE GOING TO PUT AMERICA FIRST!!\\n\\nDID YOU...</td>\n",
              "      <td>(Smears, Name calling/Labeling, Flag-waving, R...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>181</th>\n",
              "      <td>Congratulations to Nancy Pelosi!\\n\\nFor being ...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Glittering generalities (Virtue),)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>182</th>\n",
              "      <td>Trump donates ALL his salary.\\n\\nMelania has a...</td>\n",
              "      <td>(Whataboutism, Smears, Name calling/Labeling, ...</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>186</th>\n",
              "      <td>The International #Cat Federation has banned #...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>187</th>\n",
              "      <td>DEMOCRAT...\\nONCE A POLITICAL PARTY\\n\\nNOW AN ...</td>\n",
              "      <td>(Smears, Appeal to fear/prejudice, Name callin...</td>\n",
              "      <td>(Loaded Language,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>190</th>\n",
              "      <td>Someone should tell the people in this food li...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>193</th>\n",
              "      <td>\\IF BIDEN WAS TRYING TO DESTROY THE COUNTRY,\\n...</td>\n",
              "      <td>(Smears, Loaded Language, Appeal to authority)</td>\n",
              "      <td>(Appeal to authority,)</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>195</th>\n",
              "      <td>Will Smith hit me too</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>197</th>\n",
              "      <td>Trump has repeatedly said he'd like to see me ...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>199</th>\n",
              "      <td>SALON OWNER NANCY PELOSI ATTACKED RAISES OVER ...</td>\n",
              "      <td>(Smears, Loaded Language)</td>\n",
              "      <td>()</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>209</th>\n",
              "      <td>Hang this in the Louvre\\n\\nTory MP Nadine Dorr...</td>\n",
              "      <td>(Smears,)</td>\n",
              "      <td>(Smears,)</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                  Text  \\\n",
              "1        Trying to think of a single accomplishment...   \n",
              "7    President Obama,\\nWe Built It-You Broke It - W...   \n",
              "12   Everyone is all excited that after 4 years the...   \n",
              "16   ALEXANDRA OCASIO-CORTEZ\\n\\nWAS FIRED FROM \\HOT...   \n",
              "17   The composure, poise and grace that Ketanji Br...   \n",
              "27   Who sank the Titanic?\\n\\nA pro-Russian submari...   \n",
              "37   The England Football team visited an orphanage...   \n",
              "42   OBAMA AND PUTIN ARE BOTH WORKING TOWARD THE SA...   \n",
              "44                   DONALD TRUMP\\nIS A\\nRA IST\\nC P\\n   \n",
              "46   DON'T WAIT.\\nVACCINATE.\\nWE WERE ALL STUPID YO...   \n",
              "48   Russia missile strikes that led to 7 of 11 pen...   \n",
              "51   By now, many have seen the photo on the right....   \n",
              "52   Never forget this mountain was sacred to the l...   \n",
              "53   SAUDI ARABIA\\n\\nAS SOON AS THE G20 SPOTLIGHT O...   \n",
              "58   Jacob Chansley, AKA Jake Angeli, Arizona man m...   \n",
              "65   THIS IS DONALD.\\n\\nHe has,\\nThe appeal of an i...   \n",
              "72   THE MODERN SOCIALIST\\n\\nTAKES AN UBER TO THE S...   \n",
              "75   DONALD TRUMP: ILHAN OMAR 'EXTREMELY UNPATRIOTI...   \n",
              "76   REAL ESTATE BUBBLE BURSTING, BANKING CRISIS\\n\\...   \n",
              "78   Russia demands from UN, Ukraine to return arms...   \n",
              "84   Obama could learn from this speech.\\n\\nIn Russ...   \n",
              "96   There is Hope Young Scientists are searching f...   \n",
              "99   MISS UNIVERSE MISTAKE \\n\\nMEDIA FREAK OUT\\n\\nP...   \n",
              "103  HEY DEMOCRATS: IF YOU RE-ELECT TRUMP.\\n\\nTHAT'...   \n",
              "119  WELCOME TO THE 2041 ANNUAL\\n\\nMARICOPA COUNTY ...   \n",
              "126  DEMOCRAT CONVENTION:\\nMICHELLE O DECRIES KIDS ...   \n",
              "135  TALKING THE TALK\\n\\n5 children with 3 women\\nC...   \n",
              "136  \\ILLEGALLY SHOOT THE SON OF A B-\\\\n\\nHillary C...   \n",
              "137  Who else wants to impeach these three justices...   \n",
              "138  Tomi Lahren, Ann Coulter, and Kellyanne Conway...   \n",
              "142  Here's a great example of why I'm proud to be ...   \n",
              "147  The only upside to Trump passing his cognitive...   \n",
              "149  \\REPUBLICANS:\\nTHE NO. 1 REASON THE REST OF TH...   \n",
              "152  \\Not one penny more to countries that are burn...   \n",
              "154  IF PEOPLE ARE RISKING THEIR LIVES TO ESCAPE SO...   \n",
              "161  WHEN YOU'RE SO WOKE.....\\n\\nTHAT A BLACKFEET C...   \n",
              "169  The fact \\ISIS\\ has a twitter account should t...   \n",
              "170  \\I believe [Trump] knows he's an illegitimate ...   \n",
              "174                       Help Buy LeBron a Helicopter   \n",
              "177  WE ARE GOING TO PUT AMERICA FIRST!!\\n\\nDID YOU...   \n",
              "181  Congratulations to Nancy Pelosi!\\n\\nFor being ...   \n",
              "182  Trump donates ALL his salary.\\n\\nMelania has a...   \n",
              "186  The International #Cat Federation has banned #...   \n",
              "187  DEMOCRAT...\\nONCE A POLITICAL PARTY\\n\\nNOW AN ...   \n",
              "190  Someone should tell the people in this food li...   \n",
              "193  \\IF BIDEN WAS TRYING TO DESTROY THE COUNTRY,\\n...   \n",
              "195                              Will Smith hit me too   \n",
              "197  Trump has repeatedly said he'd like to see me ...   \n",
              "199  SALON OWNER NANCY PELOSI ATTACKED RAISES OVER ...   \n",
              "209  Hang this in the Louvre\\n\\nTory MP Nadine Dorr...   \n",
              "\n",
              "                                                Actual  \\\n",
              "1                  (Smears, Exaggeration/Minimisation)   \n",
              "7                                            (Smears,)   \n",
              "12                                           (Smears,)   \n",
              "16                                           (Smears,)   \n",
              "17   (Smears, Name calling/Labeling, Glittering gen...   \n",
              "27    (Smears, Name calling/Labeling, Loaded Language)   \n",
              "37                 (Smears, Exaggeration/Minimisation)   \n",
              "42     (Smears, Loaded Language, Reductio ad hitlerum)   \n",
              "44                                           (Smears,)   \n",
              "46                     (Smears, Name calling/Labeling)   \n",
              "48                     (Smears, Name calling/Labeling)   \n",
              "51   (Smears, Black-and-white Fallacy/Dictatorship,...   \n",
              "52                           (Smears, Loaded Language)   \n",
              "53                                           (Smears,)   \n",
              "58                           (Smears, Loaded Language)   \n",
              "65   (Smears, Black-and-white Fallacy/Dictatorship,...   \n",
              "72       (Whataboutism, Smears, Name calling/Labeling)   \n",
              "75   (Smears, Name calling/Labeling, Loaded Languag...   \n",
              "76                           (Smears, Loaded Language)   \n",
              "78                           (Smears, Loaded Language)   \n",
              "84   (Smears, Slogans, Flag-waving, Appeal to autho...   \n",
              "96                                           (Smears,)   \n",
              "99                           (Smears, Loaded Language)   \n",
              "103                                          (Smears,)   \n",
              "119  (Smears, Name calling/Labeling, Slogans, Exagg...   \n",
              "126            (Whataboutism, Smears, Loaded Language)   \n",
              "135  (Smears, Loaded Language, Repetition, Glitteri...   \n",
              "136                          (Smears, Loaded Language)   \n",
              "137                                          (Smears,)   \n",
              "138                    (Smears, Name calling/Labeling)   \n",
              "142                               (Smears, Repetition)   \n",
              "147                                          (Smears,)   \n",
              "149  (Smears, Name calling/Labeling, Appeal to auth...   \n",
              "152  (Smears, Name calling/Labeling, Appeal to auth...   \n",
              "154   (Smears, Name calling/Labeling, Loaded Language)   \n",
              "161                    (Smears, Name calling/Labeling)   \n",
              "169  (Smears, Thought-terminating cliché, Loaded La...   \n",
              "170  (Smears, Name calling/Labeling, Appeal to auth...   \n",
              "174                                          (Smears,)   \n",
              "177  (Smears, Name calling/Labeling, Flag-waving, R...   \n",
              "181                                          (Smears,)   \n",
              "182  (Whataboutism, Smears, Name calling/Labeling, ...   \n",
              "186                          (Smears, Loaded Language)   \n",
              "187  (Smears, Appeal to fear/prejudice, Name callin...   \n",
              "190                                          (Smears,)   \n",
              "193     (Smears, Loaded Language, Appeal to authority)   \n",
              "195                                          (Smears,)   \n",
              "197                                          (Smears,)   \n",
              "199                          (Smears, Loaded Language)   \n",
              "209                                          (Smears,)   \n",
              "\n",
              "                                             Predicted  \n",
              "1                                                   ()  \n",
              "7                                           (Slogans,)  \n",
              "12                                           (Smears,)  \n",
              "16                                           (Smears,)  \n",
              "17                            (Name calling/Labeling,)  \n",
              "27                                           (Smears,)  \n",
              "37                                                  ()  \n",
              "42                                                  ()  \n",
              "44                                                  ()  \n",
              "46                                                  ()  \n",
              "48                           (Smears, Loaded Language)  \n",
              "51                                  (Loaded Language,)  \n",
              "52                                                  ()  \n",
              "53                                                  ()  \n",
              "58                                           (Smears,)  \n",
              "65                            (Name calling/Labeling,)  \n",
              "72                                                  ()  \n",
              "75                                                  ()  \n",
              "76                                                  ()  \n",
              "78                                  (Loaded Language,)  \n",
              "84                                                  ()  \n",
              "96                                            (Doubt,)  \n",
              "99    (Smears, Name calling/Labeling, Loaded Language)  \n",
              "103                                                 ()  \n",
              "119                                                 ()  \n",
              "126                                                 ()  \n",
              "135                                          (Smears,)  \n",
              "136  (Smears, Name calling/Labeling, Loaded Languag...  \n",
              "137                                                 ()  \n",
              "138                                          (Smears,)  \n",
              "142                                                 ()  \n",
              "147                                                 ()  \n",
              "149                                                 ()  \n",
              "152                             (Appeal to authority,)  \n",
              "154                                                 ()  \n",
              "161                                                 ()  \n",
              "169                    (Smears, Name calling/Labeling)  \n",
              "170                             (Appeal to authority,)  \n",
              "174                                                 ()  \n",
              "177                                                 ()  \n",
              "181                (Glittering generalities (Virtue),)  \n",
              "182                                                 ()  \n",
              "186                                                 ()  \n",
              "187                                 (Loaded Language,)  \n",
              "190                                                 ()  \n",
              "193                             (Appeal to authority,)  \n",
              "195                                                 ()  \n",
              "197                                                 ()  \n",
              "199                                                 ()  \n",
              "209                                          (Smears,)  "
            ]
          },
          "execution_count": 92,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Selecting rows containing the given label in the selected column\n",
        "\n",
        "# lbl_of_interest = 'Whataboutism' # idx = 1\n",
        "lbl_of_interest = 'Smears' # idx = 5\n",
        "\n",
        "col_of_interest = 'Actual'\n",
        "result_df = df_error_analysis[df_error_analysis[col_of_interest].apply(lambda tup: lbl_of_interest in (','.join([str(x) for x in tup])))]\n",
        "print(result_df.shape)\n",
        "result_df.head(50)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Vq3YsmeSaJnO",
        "outputId": "b5fbcd52-36b4-4083-e457-60d4dcecf2c7"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: mlcm in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (0.0.1)\n",
            "Requirement already satisfied: numpy in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from mlcm) (1.26.2)\n"
          ]
        }
      ],
      "source": [
        "# M. Heydarian, T. Doyle, and R. Samavi, MLCM: Multi-Label Confusion Matrix, IEEE Access, Feb. 2022, DOI: 10.1109/ACCESS.2022.3151048\n",
        "\n",
        "# !pip install mlcm"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dLVdJGDUaJnO",
        "outputId": "c2cb6e9c-2bca-474d-f3e1-8ceb8492dcd5"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<module 'mlcm.mlcm' from '/home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages/mlcm/mlcm.py'>"
            ]
          },
          "execution_count": 29,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# import mlcm\n",
        "import importlib\n",
        "\n",
        "# make changes to example.py file\n",
        "\n",
        "importlib.reload(mlcm)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "jOVRHgQlaJnP",
        "outputId": "a7c98886-bfed-4ae4-d0a2-feb0baf283e8"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "MLCM has one extra row (NTL) and one extra column (NPL).        \n",
            "Please read the following paper for more information:\n",
            "        Heydarian et al., MLCM: Multi-Label Confusion Matrix, IEEE Access,2022        \n",
            "To skip this message, please add parameter \"print_note=False\"\n",
            "        e.g., conf_mat,normal_conf_mat = mlcm.cm(label_true,label_pred,False)\n",
            "\n",
            "Raw confusion Matrix:\n",
            "[[ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  2]\n",
            " [ 0  0  0  0  0  6  0  0  1  0  0  0  3  0  0  0  0  1  0  1 13]\n",
            " [ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  1  0  0  3]\n",
            " [ 0  0  0  0  0  1  0  0  0  0  0  0  0  0  0  0  0  0  0  0  3]\n",
            " [ 0  0  0  0  0  4  0  1  1  0  0  0  1  0  0  0  0  1  0  0 16]\n",
            " [ 0  0  0  0  0 38  0  0  1  2  1  0  9  0  0  0  0  3  2  3 93]\n",
            " [ 0  0  0  0  0  3  0  0  0  0  0  0  2  0  0  0  0  2  0  0 21]\n",
            " [ 0  0  0  0  0  1  0  1  0  2  1  1  2  0  0  0  0  2  0  1 43]\n",
            " [ 0  0  0  0  0  4  0  0  1  3  0  1  2  0  0  0  0  1  0  0 26]\n",
            " [ 0  0  0  0  0  6  0  1  0 16  0  0 11  0  0  0  0  4  0  0 81]\n",
            " [ 0  0  0  0  0  1  0  2  0  0  0  1  2  0  0  0  0  1  0  0 44]\n",
            " [ 0  0  0  0  0  3  0  0  0  2  0  5  1  0  0  0  0  2  0  1 30]\n",
            " [ 0  0  0  0  0  6  0  1  1  4  1  0 31  0  0  0  0  4  0  1 93]\n",
            " [ 0  0  0  0  0  2  0  2  2  0  0  1  4  1  0  0  0  2  1  0 13]\n",
            " [ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  4]\n",
            " [ 0  0  0  0  0  1  0  0  0  1  0  0  0  0  0  0  0  2  1  2 21]\n",
            " [ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  7]\n",
            " [ 0  0  0  0  0  2  0  0  0  1  0  1  2  0  0  0  0 35  0  1 22]\n",
            " [ 0  0  0  0  0  2  0  0  0  0  0  0  0  0  0  0  0  1  0  0 21]\n",
            " [ 0  0  0  0  0  9  0  1  0  2  1  1  0  0  0  0  0  1  0  2 22]\n",
            " [ 0  0  0  0  0  2  0  0  0  0  0  0  3  0  0  0  0  0  0  0 84]]\n",
            "\n",
            "Normalized confusion Matrix (%):\n",
            "[[  0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.   0.   0.   0. 100.]\n",
            " [  0.   0.   0.   0.   0.  24.   0.   0.   4.   0.   0.   0.  12.   0.\n",
            "    0.   0.   0.   4.   0.   4.  52.]\n",
            " [  0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.  25.   0.   0.  75.]\n",
            " [  0.   0.   0.   0.   0.  25.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.   0.   0.   0.  75.]\n",
            " [  0.   0.   0.   0.   0.  17.   0.   4.   4.   0.   0.   0.   4.   0.\n",
            "    0.   0.   0.   4.   0.   0.  67.]\n",
            " [  0.   0.   0.   0.   0.  25.   0.   0.   1.   1.   1.   0.   6.   0.\n",
            "    0.   0.   0.   2.   1.   2.  61.]\n",
            " [  0.   0.   0.   0.   0.  11.   0.   0.   0.   0.   0.   0.   7.   0.\n",
            "    0.   0.   0.   7.   0.   0.  75.]\n",
            " [  0.   0.   0.   0.   0.   2.   0.   2.   0.   4.   2.   2.   4.   0.\n",
            "    0.   0.   0.   4.   0.   2.  80.]\n",
            " [  0.   0.   0.   0.   0.  11.   0.   0.   3.   8.   0.   3.   5.   0.\n",
            "    0.   0.   0.   3.   0.   0.  68.]\n",
            " [  0.   0.   0.   0.   0.   5.   0.   1.   0.  13.   0.   0.   9.   0.\n",
            "    0.   0.   0.   3.   0.   0.  68.]\n",
            " [  0.   0.   0.   0.   0.   2.   0.   4.   0.   0.   0.   2.   4.   0.\n",
            "    0.   0.   0.   2.   0.   0.  86.]\n",
            " [  0.   0.   0.   0.   0.   7.   0.   0.   0.   5.   0.  11.   2.   0.\n",
            "    0.   0.   0.   5.   0.   2.  68.]\n",
            " [  0.   0.   0.   0.   0.   4.   0.   1.   1.   3.   1.   0.  22.   0.\n",
            "    0.   0.   0.   3.   0.   1.  65.]\n",
            " [  0.   0.   0.   0.   0.   7.   0.   7.   7.   0.   0.   4.  14.   4.\n",
            "    0.   0.   0.   7.   4.   0.  46.]\n",
            " [  0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.   0.   0.   0. 100.]\n",
            " [  0.   0.   0.   0.   0.   4.   0.   0.   0.   4.   0.   0.   0.   0.\n",
            "    0.   0.   0.   7.   4.   7.  75.]\n",
            " [  0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.   0.   0.   0. 100.]\n",
            " [  0.   0.   0.   0.   0.   3.   0.   0.   0.   2.   0.   2.   3.   0.\n",
            "    0.   0.   0.  55.   0.   2.  34.]\n",
            " [  0.   0.   0.   0.   0.   8.   0.   0.   0.   0.   0.   0.   0.   0.\n",
            "    0.   0.   0.   4.   0.   0.  88.]\n",
            " [  0.   0.   0.   0.   0.  23.   0.   3.   0.   5.   3.   3.   0.   0.\n",
            "    0.   0.   0.   3.   0.   5.  56.]\n",
            " [  0.   0.   0.   0.   0.   2.   0.   0.   0.   0.   0.   0.   3.   0.\n",
            "    0.   0.   0.   0.   0.   0.  94.]]\n"
          ]
        }
      ],
      "source": [
        "from mlcm import mlcm\n",
        "\n",
        "conf_mat,normal_conf_mat = mlcm.cm(y_true,y_pred)\n",
        "print('\\nRaw confusion Matrix:')\n",
        "print(conf_mat)\n",
        "print('\\nNormalized confusion Matrix (%):')\n",
        "print(normal_conf_mat)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Tv69-9qfaJnP",
        "outputId": "8976f86c-094c-4dd4-acef-38479ef01c28"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting seaborn\n",
            "  Downloading seaborn-0.13.1-py3-none-any.whl.metadata (5.4 kB)\n",
            "Requirement already satisfied: numpy!=1.24.0,>=1.20 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from seaborn) (1.26.2)\n",
            "Requirement already satisfied: pandas>=1.2 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from seaborn) (2.1.4)\n",
            "Requirement already satisfied: matplotlib!=3.6.1,>=3.4 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from seaborn) (3.8.2)\n",
            "Requirement already satisfied: contourpy>=1.0.1 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (1.2.0)\n",
            "Requirement already satisfied: cycler>=0.10 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (0.12.1)\n",
            "Requirement already satisfied: fonttools>=4.22.0 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (4.47.0)\n",
            "Requirement already satisfied: kiwisolver>=1.3.1 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (1.4.5)\n",
            "Requirement already satisfied: packaging>=20.0 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (23.2)\n",
            "Requirement already satisfied: pillow>=8 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (10.2.0)\n",
            "Requirement already satisfied: pyparsing>=2.3.1 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (3.1.1)\n",
            "Requirement already satisfied: python-dateutil>=2.7 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2020.1 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from pandas>=1.2->seaborn) (2023.3.post1)\n",
            "Requirement already satisfied: tzdata>=2022.1 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from pandas>=1.2->seaborn) (2023.3)\n",
            "Requirement already satisfied: six>=1.5 in /home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages (from python-dateutil>=2.7->matplotlib!=3.6.1,>=3.4->seaborn) (1.16.0)\n",
            "Downloading seaborn-0.13.1-py3-none-any.whl (294 kB)\n",
            "\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m294.8/294.8 kB\u001b[0m \u001b[31m13.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hInstalling collected packages: seaborn\n",
            "Successfully installed seaborn-0.13.1\n",
            "Note: you may need to restart the kernel to use updated packages.\n"
          ]
        }
      ],
      "source": [
        "# pip install seaborn"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "XDB791azaJnP",
        "outputId": "58a22e35-e5c3-4571-efc6-23f05f793ef3"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 800x600 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import seaborn as sns\n",
        "\n",
        "# Change figure size and increase dpi for better resolution\n",
        "plt.figure(figsize=(8,6), dpi=100)\n",
        "# Scale up the size of all text\n",
        "sns.set(font_scale = 1.1)\n",
        "\n",
        "# Plot Confusion Matrix using Seaborn heatmap()\n",
        "# Parameters:\n",
        "# first param - confusion matrix in array format\n",
        "# annot = True: show the numbers in each heatmap cell\n",
        "# fmt = 'd': show numbers as integers.\n",
        "ax = sns.heatmap(conf_mat, annot=True, fmt='d', )\n",
        "\n",
        "# set x-axis label and ticks.\n",
        "ax.set_xlabel(\"Predicted\", fontsize=14, labelpad=20)\n",
        "# lbls = leaf_labels[0] + [\"None\"]\n",
        "# ax.xaxis.set_ticklabels(lbls)\n",
        "\n",
        "# set y-axis label and ticks\n",
        "ax.set_ylabel(\"Actual\", fontsize=14, labelpad=20)\n",
        "\n",
        "# set plot title\n",
        "ax.set_title(\"Confusion Matrix\", fontsize=14, pad=20)\n",
        "\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "p0lw4C_eaJnP",
        "outputId": "4b0a6c14-40ae-4c16-af36-c4b0cf6aa630"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "                                                     precision    recall  f1-score   support\n",
            "\n",
            "      Obfuscation, Intentional vagueness, Confusion       0.00      0.00      0.00         2\n",
            "                                       Whataboutism       0.00      0.00      0.00        21\n",
            "           Presenting Irrelevant Data (Red Herring)       0.00      0.00      0.00         4\n",
            "Misrepresentation of Someone's Position (Straw Man)       0.00      0.00      0.00         4\n",
            "                          Causal Oversimplification       0.00      0.00      0.00        21\n",
            "                                             Smears       0.58      0.27      0.37       142\n",
            "                           Appeal to fear/prejudice       0.00      0.00      0.00        27\n",
            "               Black-and-white Fallacy/Dictatorship       0.25      0.02      0.04        53\n",
            "                         Thought-terminating cliché       0.25      0.03      0.05        38\n",
            "                              Name calling/Labeling       0.67      0.14      0.23       116\n",
            "                                            Slogans       0.00      0.00      0.00        50\n",
            "                                        Flag-waving       0.62      0.12      0.20        42\n",
            "                                    Loaded Language       0.54      0.23      0.32       135\n",
            "                                         Repetition       1.00      0.04      0.08        23\n",
            "                               Reductio ad hitlerum       0.00      0.00      0.00         4\n",
            "                          Exaggeration/Minimisation       0.00      0.00      0.00        27\n",
            "                                          Bandwagon       0.00      0.00      0.00         7\n",
            "                                Appeal to authority       0.76      0.56      0.64        63\n",
            "                                              Doubt       0.00      0.00      0.00        24\n",
            "                   Glittering generalities (Virtue)       0.22      0.06      0.09        36\n",
            "\n",
            "                                          micro avg       0.58      0.15      0.24       839\n",
            "                                          macro avg       0.24      0.07      0.10       839\n",
            "                                       weighted avg       0.43      0.15      0.21       839\n",
            "                                        samples avg       0.16      0.10      0.12       839\n",
            "\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages/sklearn/metrics/_classification.py:1471: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
            "  _warn_prf(average, modifier, msg_start, len(result))\n",
            "/home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages/sklearn/metrics/_classification.py:1471: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in samples with no predicted labels. Use `zero_division` parameter to control this behavior.\n",
            "  _warn_prf(average, modifier, msg_start, len(result))\n",
            "/home/ebezerra/anaconda3/envs/bambas/lib/python3.12/site-packages/sklearn/metrics/_classification.py:1471: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in samples with no true labels. Use `zero_division` parameter to control this behavior.\n",
            "  _warn_prf(average, modifier, msg_start, len(result))\n"
          ]
        }
      ],
      "source": [
        "# see https://stackoverflow.com/questions/68374999/is-this-the-correct-use-of-sklearn-classification-report-for-multi-label-classif\n",
        "\n",
        "from sklearn.metrics import classification_report\n",
        "import numpy as np\n",
        "\n",
        "y_true, y_pred = all_labels, all_predictions\n",
        "\n",
        "print(classification_report(y_true, y_pred, target_names=leaf_labels[0]))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RyBVMtGOaJnP"
      },
      "outputs": [],
      "source": [
        "_, dev, _ = load_dataset(args.dataset)\n",
        "dev_predicted_labels = all_predictions\n",
        "print(f\"dev_predicted_labels.shape = {dev_predicted_labels.shape}\")\n",
        "print(f\"dev.shape = {dev.shape}\")\n",
        "\n",
        "if args.classifier != \"HiMLP\":\n",
        "    dev_predicted_labels = mlb.inverse_transform(dev_predicted_labels)\n",
        "\n",
        "pred_path, _ = save_predictions(dev, dev_predicted_labels, \"dev\", args.classifier)\n",
        "print(f\"Finished successfully. val predictions saved at {pred_path}\")\n",
        "\n",
        "print()\n",
        "prec, rec, f1 = evaluate_h(pred_path, GOLD_PATH)\n",
        "print(f\"\\nValidation set:\\n\\tPrecision: {prec}\\n\\tRecall: {rec}\\n\\tF1: {f1}\\n\")"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import re\n",
        "\n",
        "def extract_author(quote):\n",
        "    # Define a regular expression pattern to match the author's name in quotes\n",
        "    author_pattern = r'— (.+)$|(?<=\\\\\")[^\\\\\"]+(?=\\\\\")'\n",
        "\n",
        "    # Search for the pattern in the given quote\n",
        "    match = re.search(author_pattern, quote)\n",
        "\n",
        "    # If a match is found, return the author's name\n",
        "    if match:\n",
        "        return match.group(0).strip()\n",
        "\n",
        "    # If no match is found, return None\n",
        "    return None\n",
        "\n",
        "# Examples\n",
        "quote1 = \"I'm selfish, impatient and a little insecure. I make mistakes, I am out of control and at times hard to handle. But if you can't handle me at my worst, then you sure as hell don't deserve me at my best. — Marilyn Monroe\"\n",
        "\n",
        "quote2 = 'Kurt Cobain quote from 1993\\\\n\\\\n\\\\In the end I believe my generation will surprise everyone. We already know that both political parties are playing both sides from the middle and we\\'ll elect a true outsider when we fully mature. I wouldn\\'t be surprised if it\\'s not a business tycoon who can\\'t be bought and who does what\\'s right for the people. Someone like Donald Trump as crazy as that sounds.\\\\\"'\n",
        "\n",
        "# Extract authors\n",
        "author1 = extract_author(quote1)\n",
        "author2 = extract_author(quote2)\n",
        "\n",
        "# Display results\n",
        "print(\"Author 1:\", author1)\n",
        "print(\"Author 2:\", author2)\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bxCn-cQAhDXm",
        "outputId": "6e3ff002-1cbd-4cfb-ad9c-f8b83e0ee051"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Author 1: — Marilyn Monroe\n",
            "Author 2: None\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import nltk\n",
        "nltk.download('punkt')\n",
        "nltk.download('averaged_perceptron_tagger')\n",
        "nltk.download('maxent_ne_chunker')\n",
        "nltk.download('words')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jwrl1pTth5tM",
        "outputId": "1e9689ed-d8d2-4ab4-e644-ebb6d56a4f93"
      },
      "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 averaged_perceptron_tagger to\n",
            "[nltk_data]     /root/nltk_data...\n",
            "[nltk_data]   Package averaged_perceptron_tagger is already up-to-\n",
            "[nltk_data]       date!\n",
            "[nltk_data] Downloading package maxent_ne_chunker to\n",
            "[nltk_data]     /root/nltk_data...\n",
            "[nltk_data]   Package maxent_ne_chunker is already up-to-date!\n",
            "[nltk_data] Downloading package words to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/words.zip.\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import re\n",
        "from nltk import sent_tokenize, word_tokenize, pos_tag, ne_chunk\n",
        "\n",
        "def extract_proper_nouns(text):\n",
        "    # Using NLTK for named entity recognition to identify potential author names\n",
        "    sentences = sent_tokenize(text)\n",
        "    authors = []\n",
        "    for sentence in sentences:\n",
        "        words = word_tokenize(sentence)\n",
        "        pos_tags = pos_tag(words)\n",
        "        named_entities = ne_chunk(pos_tags)\n",
        "        for chunk in named_entities:\n",
        "            if hasattr(chunk, 'label') and chunk.label() == 'PERSON':\n",
        "                authors.append(' '.join(c[0] for c in chunk))\n",
        "\n",
        "    return authors\n",
        "\n",
        "text_examples = ['John and Mary went to the market. Steve joined them later.',\n",
        "                 \"I'm selfish, impatient and a little insecure. I make mistakes, I am out of control and at times hard to handle. But if you can't handle me at my worst, then you sure as hell don't deserve me at my best. — Marilyn Monroe\",\n",
        "                 'Kurt Cobain quote from 1993\\\\n\\\\n\\\\In the end I believe my generation will surprise everyone. We already know that both political parties are playing both sides from the middle and we\\'ll elect a true outsider when we fully mature. I wouldn\\'t be surprised if it\\'s not a business tycoon who can\\'t be bought and who does what\\'s right for the people. Someone like Donald Trump as crazy as that sounds.\\\\\"',\n",
        "                 \"When you look at serial number of your AK-47 and its the same one you had in Afganistan\",\n",
        "                 \"BREAKING: Andrew Cuomo Will Be Arrested Next Week\\\\nFor Sex Crimes; Here's What We Know\",\n",
        "                 \"Statistically, you're more likely to die from living under\\\\nsocialism than by an \\\\Assault Rifle.\\\\\\\\n- Reality\"]\n",
        "\n",
        "for text in text_examples:\n",
        "  authors = extract_proper_nouns(text)\n",
        "  print(\"Authors:\", authors)\n",
        "  print()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rbssNWD9hyun",
        "outputId": "99da3036-1cdf-4396-c7f4-c39102fa8aa0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Authors: ['John', 'Mary', 'Steve']\n",
            "\n",
            "Authors: ['Marilyn Monroe']\n",
            "\n",
            "Authors: ['Kurt', 'Donald Trump']\n",
            "\n",
            "Authors: []\n",
            "\n",
            "Authors: ['Andrew Cuomo', 'Next']\n",
            "\n",
            "Authors: []\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "_, dev, _ = load_dataset(args.dataset)\n",
        "dev.head()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 193
        },
        "id": "CPVUgH-r1ChH",
        "outputId": "335c2963-03ef-4769-d85e-217ccd8b2b20"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'load_dataset' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-24-7d124652b0db>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdev\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mload_dataset\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mdev\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhead\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'load_dataset' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "pip install flair"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1fsVGkDzCXYF",
        "outputId": "6d1367fb-388c-4ffd-b505-c17163e9c7c2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting flair\n",
            "  Downloading flair-0.13.1-py3-none-any.whl (388 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m388.3/388.3 kB\u001b[0m \u001b[31m4.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting boto3>=1.20.27 (from flair)\n",
            "  Downloading boto3-1.34.17-py3-none-any.whl (139 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m139.3/139.3 kB\u001b[0m \u001b[31m5.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting bpemb>=0.3.2 (from flair)\n",
            "  Downloading bpemb-0.3.4-py3-none-any.whl (19 kB)\n",
            "Collecting conllu>=4.0 (from flair)\n",
            "  Downloading conllu-4.5.3-py2.py3-none-any.whl (16 kB)\n",
            "Collecting deprecated>=1.2.13 (from flair)\n",
            "  Downloading Deprecated-1.2.14-py2.py3-none-any.whl (9.6 kB)\n",
            "Collecting ftfy>=6.1.0 (from flair)\n",
            "  Downloading ftfy-6.1.3-py3-none-any.whl (53 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m53.4/53.4 kB\u001b[0m \u001b[31m6.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: gdown>=4.4.0 in /usr/local/lib/python3.10/dist-packages (from flair) (4.6.6)\n",
            "Requirement already satisfied: gensim>=4.2.0 in /usr/local/lib/python3.10/dist-packages (from flair) (4.3.2)\n",
            "Requirement already satisfied: huggingface-hub>=0.10.0 in /usr/local/lib/python3.10/dist-packages (from flair) (0.20.2)\n",
            "Collecting janome>=0.4.2 (from flair)\n",
            "  Downloading Janome-0.5.0-py2.py3-none-any.whl (19.7 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m19.7/19.7 MB\u001b[0m \u001b[31m16.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting langdetect>=1.0.9 (from flair)\n",
            "  Downloading langdetect-1.0.9.tar.gz (981 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m981.5/981.5 kB\u001b[0m \u001b[31m22.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: lxml>=4.8.0 in /usr/local/lib/python3.10/dist-packages (from flair) (4.9.4)\n",
            "Requirement already satisfied: matplotlib>=2.2.3 in /usr/local/lib/python3.10/dist-packages (from flair) (3.7.1)\n",
            "Requirement already satisfied: more-itertools>=8.13.0 in /usr/local/lib/python3.10/dist-packages (from flair) (10.1.0)\n",
            "Collecting mpld3>=0.3 (from flair)\n",
            "  Downloading mpld3-0.5.10-py3-none-any.whl (202 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m202.6/202.6 kB\u001b[0m \u001b[31m16.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting pptree>=3.1 (from flair)\n",
            "  Downloading pptree-3.1.tar.gz (3.0 kB)\n",
            "  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: python-dateutil>=2.8.2 in /usr/local/lib/python3.10/dist-packages (from flair) (2.8.2)\n",
            "Collecting pytorch-revgrad>=0.2.0 (from flair)\n",
            "  Downloading pytorch_revgrad-0.2.0-py3-none-any.whl (4.6 kB)\n",
            "Requirement already satisfied: regex>=2022.1.18 in /usr/local/lib/python3.10/dist-packages (from flair) (2023.6.3)\n",
            "Requirement already satisfied: scikit-learn>=1.0.2 in /usr/local/lib/python3.10/dist-packages (from flair) (1.2.2)\n",
            "Collecting segtok>=1.5.11 (from flair)\n",
            "  Downloading segtok-1.5.11-py3-none-any.whl (24 kB)\n",
            "Collecting sqlitedict>=2.0.0 (from flair)\n",
            "  Downloading sqlitedict-2.1.0.tar.gz (21 kB)\n",
            "  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: tabulate>=0.8.10 in /usr/local/lib/python3.10/dist-packages (from flair) (0.9.0)\n",
            "Requirement already satisfied: torch!=1.8,>=1.5.0 in /usr/local/lib/python3.10/dist-packages (from flair) (2.1.0+cu121)\n",
            "Requirement already satisfied: tqdm>=4.63.0 in /usr/local/lib/python3.10/dist-packages (from flair) (4.66.1)\n",
            "Collecting transformer-smaller-training-vocab>=0.2.3 (from flair)\n",
            "  Downloading transformer_smaller_training_vocab-0.3.3-py3-none-any.whl (14 kB)\n",
            "Requirement already satisfied: transformers[sentencepiece]<5.0.0,>=4.18.0 in /usr/local/lib/python3.10/dist-packages (from flair) (4.35.2)\n",
            "Collecting urllib3<2.0.0,>=1.0.0 (from flair)\n",
            "  Downloading urllib3-1.26.18-py2.py3-none-any.whl (143 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m143.8/143.8 kB\u001b[0m \u001b[31m11.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting wikipedia-api>=0.5.7 (from flair)\n",
            "  Downloading Wikipedia_API-0.6.0-py3-none-any.whl (14 kB)\n",
            "Collecting semver<4.0.0,>=3.0.0 (from flair)\n",
            "  Downloading semver-3.0.2-py3-none-any.whl (17 kB)\n",
            "Collecting botocore<1.35.0,>=1.34.17 (from boto3>=1.20.27->flair)\n",
            "  Downloading botocore-1.34.17-py3-none-any.whl (11.9 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m11.9/11.9 MB\u001b[0m \u001b[31m36.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting jmespath<2.0.0,>=0.7.1 (from boto3>=1.20.27->flair)\n",
            "  Downloading jmespath-1.0.1-py3-none-any.whl (20 kB)\n",
            "Collecting s3transfer<0.11.0,>=0.10.0 (from boto3>=1.20.27->flair)\n",
            "  Downloading s3transfer-0.10.0-py3-none-any.whl (82 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m82.1/82.1 kB\u001b[0m \u001b[31m9.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from bpemb>=0.3.2->flair) (1.23.5)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.10/dist-packages (from bpemb>=0.3.2->flair) (2.31.0)\n",
            "Collecting sentencepiece (from bpemb>=0.3.2->flair)\n",
            "  Downloading sentencepiece-0.1.99-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.3/1.3 MB\u001b[0m \u001b[31m56.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: wrapt<2,>=1.10 in /usr/local/lib/python3.10/dist-packages (from deprecated>=1.2.13->flair) (1.14.1)\n",
            "Requirement already satisfied: wcwidth<0.3.0,>=0.2.12 in /usr/local/lib/python3.10/dist-packages (from ftfy>=6.1.0->flair) (0.2.12)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from gdown>=4.4.0->flair) (3.13.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from gdown>=4.4.0->flair) (1.16.0)\n",
            "Requirement already satisfied: beautifulsoup4 in /usr/local/lib/python3.10/dist-packages (from gdown>=4.4.0->flair) (4.11.2)\n",
            "Requirement already satisfied: scipy>=1.7.0 in /usr/local/lib/python3.10/dist-packages (from gensim>=4.2.0->flair) (1.11.4)\n",
            "Requirement already satisfied: smart-open>=1.8.1 in /usr/local/lib/python3.10/dist-packages (from gensim>=4.2.0->flair) (6.4.0)\n",
            "Requirement already satisfied: fsspec>=2023.5.0 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.10.0->flair) (2023.6.0)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.10.0->flair) (6.0.1)\n",
            "Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.10.0->flair) (4.5.0)\n",
            "Requirement already satisfied: packaging>=20.9 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.10.0->flair) (23.2)\n",
            "Requirement already satisfied: contourpy>=1.0.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (1.2.0)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (0.12.1)\n",
            "Requirement already satisfied: fonttools>=4.22.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (4.47.0)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (1.4.5)\n",
            "Requirement already satisfied: pillow>=6.2.0 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (9.4.0)\n",
            "Requirement already satisfied: pyparsing>=2.3.1 in /usr/local/lib/python3.10/dist-packages (from matplotlib>=2.2.3->flair) (3.1.1)\n",
            "Requirement already satisfied: jinja2 in /usr/local/lib/python3.10/dist-packages (from mpld3>=0.3->flair) (3.1.2)\n",
            "Requirement already satisfied: joblib>=1.1.1 in /usr/local/lib/python3.10/dist-packages (from scikit-learn>=1.0.2->flair) (1.3.2)\n",
            "Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.10/dist-packages (from scikit-learn>=1.0.2->flair) (3.2.0)\n",
            "Requirement already satisfied: sympy in /usr/local/lib/python3.10/dist-packages (from torch!=1.8,>=1.5.0->flair) (1.12)\n",
            "Requirement already satisfied: networkx in /usr/local/lib/python3.10/dist-packages (from torch!=1.8,>=1.5.0->flair) (3.2.1)\n",
            "Requirement already satisfied: triton==2.1.0 in /usr/local/lib/python3.10/dist-packages (from torch!=1.8,>=1.5.0->flair) (2.1.0)\n",
            "Requirement already satisfied: tokenizers<0.19,>=0.14 in /usr/local/lib/python3.10/dist-packages (from transformers[sentencepiece]<5.0.0,>=4.18.0->flair) (0.15.0)\n",
            "Requirement already satisfied: safetensors>=0.3.1 in /usr/local/lib/python3.10/dist-packages (from transformers[sentencepiece]<5.0.0,>=4.18.0->flair) (0.4.1)\n",
            "Requirement already satisfied: protobuf in /usr/local/lib/python3.10/dist-packages (from transformers[sentencepiece]<5.0.0,>=4.18.0->flair) (3.20.3)\n",
            "Collecting accelerate>=0.20.3 (from transformers[sentencepiece]<5.0.0,>=4.18.0->flair)\n",
            "  Downloading accelerate-0.26.1-py3-none-any.whl (270 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m270.9/270.9 kB\u001b[0m \u001b[31m28.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: soupsieve>1.2 in /usr/local/lib/python3.10/dist-packages (from beautifulsoup4->gdown>=4.4.0->flair) (2.5)\n",
            "Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.10/dist-packages (from jinja2->mpld3>=0.3->flair) (2.1.3)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests->bpemb>=0.3.2->flair) (3.3.2)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests->bpemb>=0.3.2->flair) (3.6)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests->bpemb>=0.3.2->flair) (2023.11.17)\n",
            "Requirement already satisfied: PySocks!=1.5.7,>=1.5.6 in /usr/local/lib/python3.10/dist-packages (from requests->bpemb>=0.3.2->flair) (1.7.1)\n",
            "Requirement already satisfied: mpmath>=0.19 in /usr/local/lib/python3.10/dist-packages (from sympy->torch!=1.8,>=1.5.0->flair) (1.3.0)\n",
            "Requirement already satisfied: psutil in /usr/local/lib/python3.10/dist-packages (from accelerate>=0.20.3->transformers[sentencepiece]<5.0.0,>=4.18.0->flair) (5.9.5)\n",
            "Building wheels for collected packages: langdetect, pptree, sqlitedict\n",
            "  Building wheel for langdetect (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for langdetect: filename=langdetect-1.0.9-py3-none-any.whl size=993225 sha256=82509e9fa06df751e5932059fda88690cc3f63433676c596b3c72d4ab77d3b8f\n",
            "  Stored in directory: /root/.cache/pip/wheels/95/03/7d/59ea870c70ce4e5a370638b5462a7711ab78fba2f655d05106\n",
            "  Building wheel for pptree (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pptree: filename=pptree-3.1-py3-none-any.whl size=4608 sha256=a4002a3f8d9c05181d9ab0d17601d78e7b118578738399e21690c52a98ed4352\n",
            "  Stored in directory: /root/.cache/pip/wheels/9f/b6/0e/6f26eb9e6eb53ff2107a7888d72b5a6a597593956113037828\n",
            "  Building wheel for sqlitedict (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for sqlitedict: filename=sqlitedict-2.1.0-py3-none-any.whl size=16863 sha256=d14c436cbd977884c0556f2433b68a4e92d4b466639be1e4bc2830ad32c76de7\n",
            "  Stored in directory: /root/.cache/pip/wheels/79/d6/e7/304e0e6cb2221022c26d8161f7c23cd4f259a9e41e8bbcfabd\n",
            "Successfully built langdetect pptree sqlitedict\n",
            "Installing collected packages: sqlitedict, sentencepiece, pptree, janome, urllib3, semver, segtok, langdetect, jmespath, ftfy, deprecated, conllu, botocore, wikipedia-api, s3transfer, pytorch-revgrad, mpld3, bpemb, boto3, accelerate, transformer-smaller-training-vocab, flair\n",
            "  Attempting uninstall: urllib3\n",
            "    Found existing installation: urllib3 2.0.7\n",
            "    Uninstalling urllib3-2.0.7:\n",
            "      Successfully uninstalled urllib3-2.0.7\n",
            "Successfully installed accelerate-0.26.1 boto3-1.34.17 botocore-1.34.17 bpemb-0.3.4 conllu-4.5.3 deprecated-1.2.14 flair-0.13.1 ftfy-6.1.3 janome-0.5.0 jmespath-1.0.1 langdetect-1.0.9 mpld3-0.5.10 pptree-3.1 pytorch-revgrad-0.2.0 s3transfer-0.10.0 segtok-1.5.11 semver-3.0.2 sentencepiece-0.1.99 sqlitedict-2.1.0 transformer-smaller-training-vocab-0.3.3 urllib3-1.26.18 wikipedia-api-0.6.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from flair.data import Sentence\n",
        "from flair.models import SequenceTagger\n",
        "\n",
        "# load tagger\n",
        "tagger = SequenceTagger.load(\"flair/pos-english\")\n",
        "\n",
        "# make example sentence\n",
        "sentence = Sentence(\"I love DONALD TRUMP.\")\n",
        "\n",
        "# predict NER tags\n",
        "tagger.predict(sentence)\n",
        "\n",
        "# print sentence\n",
        "print(sentence)\n",
        "\n",
        "print(sentence.get_spans('pos'))\n",
        "\n",
        "# print predicted NER spans\n",
        "print('The following NER tags are found:')\n",
        "# iterate over entities and print\n",
        "for entity in sentence.get_spans('pos'):\n",
        "    print(entity)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ncyup18SCkRz",
        "outputId": "078f8c28-7e33-468e-9e47-ebc0b6fa7ca1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2024-01-12 14:28:46,215 SequenceTagger predicts: Dictionary with 53 tags: <unk>, O, UH, ,, VBD, PRP, VB, PRP$, NN, RB, ., DT, JJ, VBP, VBG, IN, CD, NNS, NNP, WRB, VBZ, WDT, CC, TO, MD, VBN, WP, :, RP, EX, JJR, FW, XX, HYPH, POS, RBR, JJS, PDT, NNPS, RBS, AFX, WP$, -LRB-, -RRB-, ``, '', LS, $, SYM, ADD\n",
            "Sentence[5]: \"I love DONALD TRUMP.\" → [\"I\"/PRP, \"love\"/VBP, \"DONALD\"/NNP, \"TRUMP\"/NNP, \".\"/.]\n",
            "[]\n",
            "The following NER tags are found:\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "text_examples = ['John and Mary went to the market. Steve joined them later.',\n",
        "                 \"I'm selfish, impatient and a little insecure. I make mistakes, I am out of control and at times hard to handle. But if you can't handle me at my worst, then you sure as hell don't deserve me at my best. — Marilyn Monroe\",\n",
        "                 'Kurt Cobain quote from 1993\\\\n\\\\n\\\\In the end I believe my generation will surprise everyone. We already know that both political parties are playing both sides from the middle and we\\'ll elect a true outsider when we fully mature. I wouldn\\'t be surprised if it\\'s not a business tycoon who can\\'t be bought and who does what\\'s right for the people. Someone like Donald Trump as crazy as that sounds.\\\\\"',\n",
        "                 \"When you look at serial number of your AK-47 and its the same one you had in Afganistan\",\n",
        "                 \"BREAKING: Andrew Cuomo Will Be Arrested Next Week\\\\nFor Sex Crimes; Here's What We Know\",\n",
        "                 \"Statistically, you're more likely to die from living under\\\\nsocialism than by an \\\\Assault Rifle.\\\\\\\\n- Reality\",\n",
        "                 \"DONALD TRUMP MUST BE INDICTED. A FAILURE TO PROSECUTE TRUMP FOR HIS CONSPIRACY TO\\nOVERTHROW OUR DEMOCRACY GUARANTEES FUTURE ATTEMPTS TO OVERTHROW OUR DEMOCRACY.\\nGLENN KIRSCHNER\"]\n",
        "\n",
        "for text in text_examples"
      ],
      "metadata": {
        "id": "wWy2_DDNCZhA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# load tagger\n",
        "tagger = SequenceTagger.load(\"flair/ner-english-large\")"
      ],
      "metadata": {
        "id": "Wfxq0D7QFbYM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from flair.data import Sentence\n",
        "from flair.models import SequenceTagger\n",
        "\n",
        "# make example sentence\n",
        "sentence = Sentence(\"George Washington went to Washington\")\n",
        "\n",
        "# predict NER tags\n",
        "tagger.predict(sentence)\n",
        "\n",
        "# print sentence\n",
        "print(sentence)\n",
        "\n",
        "# print predicted NER spans\n",
        "print('The following NER tags are found:')\n",
        "# iterate over entities and print\n",
        "for entity in sentence.get_spans('ner'):\n",
        "    print(entity)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "apdV7rJlCj5O",
        "outputId": "992a2a9e-843c-405b-98fa-2b99be040bfd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2024-01-12 14:31:44,871 SequenceTagger predicts: Dictionary with 20 tags: <unk>, O, S-ORG, S-MISC, B-PER, E-PER, S-LOC, B-ORG, E-ORG, I-PER, S-PER, B-MISC, I-MISC, E-MISC, I-ORG, B-LOC, E-LOC, I-LOC, <START>, <STOP>\n",
            "Sentence[5]: \"George Washington went to Washington\" → [\"George Washington\"/PER, \"Washington\"/LOC]\n",
            "The following NER tags are found:\n",
            "Span[0:2]: \"George Washington\" → PER (1.0)\n",
            "Span[4:5]: \"Washington\" → LOC (1.0)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "text_examples = ['John and Mary went to the market. Steve joined them later.',\n",
        "                 \"I'm selfish, impatient and a little insecure. I make mistakes, I am out of control and at times hard to handle. But if you can't handle me at my worst, then you sure as hell don't deserve me at my best. — Marilyn Monroe\",\n",
        "                 'Kurt Cobain quote from 1993\\\\n\\\\n\\\\In the end I believe my generation will surprise everyone. We already know that both political parties are playing both sides from the middle and we\\'ll elect a true outsider when we fully mature. I wouldn\\'t be surprised if it\\'s not a business tycoon who can\\'t be bought and who does what\\'s right for the people. Someone like Donald Trump as crazy as that sounds.\\\\\"',\n",
        "                 \"When you look at serial number of your AK-47 and its the same one you had in Afganistan\",\n",
        "                 \"BREAKING: Andrew Cuomo Will Be Arrested Next Week\\\\nFor Sex Crimes; Here's What We Know\",\n",
        "                 \"Statistically, you're more likely to die from living under\\\\nsocialism than by an \\\\Assault Rifle.\\\\\\\\n- Reality\",\n",
        "                 \"DONALD TRUMP MUST BE INDICTED. A FAILURE TO PROSECUTE TRUMP FOR HIS CONSPIRACY TO\\nOVERTHROW OUR DEMOCRACY GUARANTEES FUTURE ATTEMPTS TO OVERTHROW OUR DEMOCRACY.\\nGLENN KIRSCHNER\",\n",
        "                 \"Dear America: You are waking up, as Germany once did, to the awareness that 1/3 of your people would kill\\nanother 1/3, while 1/3 watches.\\n\\nWerner Herzog\\n\",\n",
        "                 \"SHOULD PRESIDENT TRUMP SIGN AN EXECUTIVE ORDER\\n\\nTO REQUIRE VOTER I.D. NATIONALLY TO VOTE IN THE\\n2020 ELECTION?\",\n",
        "                 \"Never forget: Attacks on the working class come from those who travel by private jet, not migrant dinghy.\\\\n\\n- Zarah Sultana\"]\n",
        "\n",
        "for text in text_examples:\n",
        "  # make example sentence\n",
        "  sentence = Sentence(text)\n",
        "\n",
        "  # predict NER tags\n",
        "  tagger.predict(sentence)\n",
        "\n",
        "  # print sentence\n",
        "  print(sentence)\n",
        "\n",
        "  nb_persons = 0\n",
        "\n",
        "  # print predicted NER spans\n",
        "  print('The following NER tags are found:')\n",
        "  # iterate over entities and print\n",
        "  for entity in sentence.get_spans('ner'):\n",
        "      print(entity)\n",
        "      # print(type(entity))\n",
        "      # print(dir(entity))\n",
        "      print(int(entity.tag == 'PER'))\n",
        "      # print(entity.text)\n",
        "      # print(entity.labels)\n",
        "      # print(entity.score)\n",
        "      nb_persons += 1\n",
        "  print()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 245
        },
        "id": "IZa1u1pyFiSq",
        "outputId": "92ca4f24-c7dc-4861-f127-f13bfc56648c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'Sentence' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-1-7fa11e48e242>\u001b[0m in \u001b[0;36m<cell line: 12>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mtext\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtext_examples\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m   \u001b[0;31m# make example sentence\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m   \u001b[0msentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mSentence\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtext\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m   \u001b[0;31m# predict NER tags\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'Sentence' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "\n",
        "def generate_new_feature(df, custom_function):\n",
        "    \"\"\"\n",
        "    Generate a new feature in the DataFrame based on the provided function.\n",
        "\n",
        "    Parameters:\n",
        "    - df: pandas DataFrame\n",
        "        The input DataFrame containing the training set with features.\n",
        "    - custom_function: function\n",
        "        The custom function to apply for generating the new feature.\n",
        "\n",
        "    Returns:\n",
        "    - df: pandas DataFrame\n",
        "        The DataFrame with the new feature added.\n",
        "    \"\"\"\n",
        "    # Apply the custom function to each row in the DataFrame\n",
        "    new_feature = df.apply(lambda row: custom_function(row), axis=1)\n",
        "\n",
        "    # Add the new feature to the DataFrame\n",
        "    df['new_feature'] = new_feature\n",
        "\n",
        "    return df\n",
        "\n",
        "# Example of a custom function\n",
        "def custom_function(row):\n",
        "    # Example: Sum all values in the row\n",
        "    return row.sum()\n",
        "\n",
        "# Example usage\n",
        "# Assuming you have a DataFrame named 'train_df'\n",
        "train_df = pd.DataFrame({'feature1': [1, 2, 3], 'feature2': [4, 5, 6]})\n",
        "\n",
        "print(train_df)\n",
        "\n",
        "result_df = generate_new_feature(train_df, custom_function)\n",
        "\n",
        "# Display the DataFrame with the new feature\n",
        "print(result_df)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4UorsZaA619s",
        "outputId": "375f5799-dfc4-4797-8857-2d216bed3b5d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "   feature1  feature2\n",
            "0         1         4\n",
            "1         2         5\n",
            "2         3         6\n",
            "   feature1  feature2  new_feature\n",
            "0         1         4            5\n",
            "1         2         5            7\n",
            "2         3         6            9\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "    print(\"Loading features array files\")\n",
        "    train_ft, test_ft, dev_ft = map(\n",
        "        load_features_array, [\n",
        "            train_ft_info[\"features\"], test_ft_info[\"features\"], dev_ft_info[\"features\"]])\n",
        "    print(\"Features Lengths\", len(train_ft), len(test_ft), len(dev_ft))\n",
        "\n",
        "    print(type(train_ft))\n",
        "    print(train_ft.shape)"
      ],
      "metadata": {
        "id": "a_Ukz4AYCJ_1"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "bambas",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.12.1"
    },
    "colab": {
      "provenance": []
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}