NER_eng_BiLSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "NER_eng_BiLSTM.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "YsSB3CDt2NxF",
        "WP5M3GnHxY-9",
        "MRofcZfAUd9g",
        "4p6nLSAePE9U",
        "Hm9UnKBqUeZA",
        "vfswr-bAXwl5"
      ],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YsSB3CDt2NxF"
      },
      "source": [
        "## Outline\n",
        "- [Introduction](#0)\n",
        "    - [Import libraries](#0.1)\n",
        "- [Part 1:  Explore the data](#1)\n",
        "    - [1.1  Import the datasets](#1.1)\n",
        "    - [1.2  Exploratory Analysis](#1.2)\n",
        "    - [Conclusion after analysis](#1.3)\n",
        "    \n",
        "- [Part 2:  Pre-process the data](#2)\n",
        "    - [2.1  Stemming](#2.1)\n",
        "    - [2.2  Lemmatization](#2.2)\n",
        "    - [2.3  Replacement](#2.3)\n",
        "    - [2.4  Pre-processing pipeline](#2.4)\n",
        "    - [2.5  Split to train/val datasets](#2.5)\n",
        "    - [2.6  Tokenization and Padding](#2.6)\n",
        "    - [2.7  Check the Imbalance in train dataset](#2.7)\n",
        "    - [2.8  One-hot encoding](#2.8)\n",
        "\n",
        "- [Part 3:  Build the model](#3)\n",
        "    - [3.1  Glove Embedding](#3.1)\n",
        "    - [3.2  Define the model](#3.2)\n",
        "    - [3.3  Callbacks](#3.3)\n",
        "    \n",
        "\n",
        "- [Part 4:  Train the model](#4)\n",
        "- [Part 5:  Test the model](#5)\n",
        "- [Part 6:  Test with your own sentence](#6)\n",
        "\n",
        "- [Part 7:  Analyse the incorrect predictions](#7)\n",
        "    - [Potential improvement](#7.1)\n",
        "\n",
        "- [Export result to .tsv file](#8)\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Uj9t5Iav2nzR"
      },
      "source": [
        "<a name=\"0\"></a>\n",
        "# Introduction\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pZDTBRDfi6u7"
      },
      "source": [
        "<a name=\"0.1\"></a>\n",
        "## Import libraries"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ntqkdg4N3HW4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "026c8545-ec38-4d57-9ca9-e7d508510898"
      },
      "source": [
        "!python --version\n",
        "import os\n",
        "\n",
        "%tensorflow_version 2.x\n",
        "import tensorflow as tf\n",
        "print(tf.__version__)\n",
        "\n",
        "# build the tokenized sentences and tags\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "from tensorflow.keras.initializers import Constant\n",
        "from tensorflow.keras import Model\n",
        "from tensorflow.keras.layers import Input, Embedding, Bidirectional, LSTM, \\\n",
        "TimeDistributed, Dense, Dropout\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
        "\n",
        "import numpy as np # linear algebra\n",
        "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import seaborn as sns\n",
        "from collections import Counter\n",
        "import random as rnd\n",
        "from nltk.corpus import stopwords\n",
        "import nltk\n",
        "nltk.download('stopwords')\n",
        "from nltk.stem import WordNetLemmatizer \n",
        "from nltk.stem import PorterStemmer\n",
        "\n",
        "!pip install sklearn_crfsuite\n",
        "from sklearn_crfsuite.metrics import flat_classification_report\n",
        "!pip install seqeval\n",
        "from seqeval.metrics import precision_score, recall_score, f1_score, classification_report\n",
        "import csv\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Python 3.7.10\n",
            "2.5.0\n",
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n",
            "Collecting sklearn_crfsuite\n",
            "  Downloading https://files.pythonhosted.org/packages/25/74/5b7befa513482e6dee1f3dd68171a6c9dfc14c0eaa00f885ffeba54fe9b0/sklearn_crfsuite-0.3.6-py2.py3-none-any.whl\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (1.15.0)\n",
            "Requirement already satisfied: tabulate in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (0.8.9)\n",
            "Collecting python-crfsuite>=0.8.3\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/79/47/58f16c46506139f17de4630dbcfb877ce41a6355a1bbf3c443edb9708429/python_crfsuite-0.9.7-cp37-cp37m-manylinux1_x86_64.whl (743kB)\n",
            "\u001b[K     |████████████████████████████████| 747kB 12.5MB/s \n",
            "\u001b[?25hRequirement already satisfied: tqdm>=2.0 in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (4.41.1)\n",
            "Installing collected packages: python-crfsuite, sklearn-crfsuite\n",
            "Successfully installed python-crfsuite-0.9.7 sklearn-crfsuite-0.3.6\n",
            "Collecting seqeval\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/9d/2d/233c79d5b4e5ab1dbf111242299153f3caddddbb691219f363ad55ce783d/seqeval-1.2.2.tar.gz (43kB)\n",
            "\u001b[K     |████████████████████████████████| 51kB 5.0MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.14.0 in /usr/local/lib/python3.7/dist-packages (from seqeval) (1.19.5)\n",
            "Requirement already satisfied: scikit-learn>=0.21.3 in /usr/local/lib/python3.7/dist-packages (from seqeval) (0.22.2.post1)\n",
            "Requirement already satisfied: scipy>=0.17.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn>=0.21.3->seqeval) (1.4.1)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn>=0.21.3->seqeval) (1.0.1)\n",
            "Building wheels for collected packages: seqeval\n",
            "  Building wheel for seqeval (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for seqeval: filename=seqeval-1.2.2-cp37-none-any.whl size=16184 sha256=ae07dcf27893686fb86ad23dc99961e6c43db9b655a080d92441256538067ba7\n",
            "  Stored in directory: /root/.cache/pip/wheels/52/df/1b/45d75646c37428f7e626214704a0e35bd3cfc32eda37e59e5f\n",
            "Successfully built seqeval\n",
            "Installing collected packages: seqeval\n",
            "Successfully installed seqeval-1.2.2\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "T58oiLSU25Za"
      },
      "source": [
        "<a name=\"1\"></a>\n",
        "# Part 1:  Explore the data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WP5M3GnHxY-9"
      },
      "source": [
        "<a name=\"1.1\"></a>\n",
        "## 1.1 Import the datasets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iHhxVylamxCG",
        "outputId": "50607de3-273f-432a-e13b-af3344732e19"
      },
      "source": [
        "# Create new directories\n",
        "!mkdir -p /data/eng\n",
        "!mkdir -p /data/trivia10k13\n",
        "!mkdir -p /model\n",
        "\n",
        "# Download data\n",
        "!wget --no-check-certificate \\\n",
        "https://groups.csail.mit.edu/sls/downloads/movie/engtrain.bio \\\n",
        "-O /data/eng/train.tsv\n",
        "\n",
        "!wget --no-check-certificate \\\n",
        "https://groups.csail.mit.edu/sls/downloads/movie/engtest.bio \\\n",
        "-O /data/eng/test.tsv\n",
        "\n",
        "!wget --no-check-certificate \\\n",
        "https://groups.csail.mit.edu/sls/downloads/movie/trivia10k13train.bio \\\n",
        "-O /data/trivia10k13/train.tsv\n",
        "\n",
        "!wget --no-check-certificate \\\n",
        "https://groups.csail.mit.edu/sls/downloads/movie/trivia10k13test.bio \\\n",
        "-O /data/trivia10k13/test.tsv"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-06-03 08:19:10--  https://groups.csail.mit.edu/sls/downloads/movie/engtrain.bio\n",
            "Resolving groups.csail.mit.edu (groups.csail.mit.edu)... 128.30.2.44\n",
            "Connecting to groups.csail.mit.edu (groups.csail.mit.edu)|128.30.2.44|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 1013492 (990K)\n",
            "Saving to: ‘/data/eng/train.tsv’\n",
            "\n",
            "/data/eng/train.tsv 100%[===================>] 989.74K  1.85MB/s    in 0.5s    \n",
            "\n",
            "2021-06-03 08:19:11 (1.85 MB/s) - ‘/data/eng/train.tsv’ saved [1013492/1013492]\n",
            "\n",
            "--2021-06-03 08:19:11--  https://groups.csail.mit.edu/sls/downloads/movie/engtest.bio\n",
            "Resolving groups.csail.mit.edu (groups.csail.mit.edu)... 128.30.2.44\n",
            "Connecting to groups.csail.mit.edu (groups.csail.mit.edu)|128.30.2.44|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 252636 (247K)\n",
            "Saving to: ‘/data/eng/test.tsv’\n",
            "\n",
            "/data/eng/test.tsv  100%[===================>] 246.71K   714KB/s    in 0.3s    \n",
            "\n",
            "2021-06-03 08:19:12 (714 KB/s) - ‘/data/eng/test.tsv’ saved [252636/252636]\n",
            "\n",
            "--2021-06-03 08:19:12--  https://groups.csail.mit.edu/sls/downloads/movie/trivia10k13train.bio\n",
            "Resolving groups.csail.mit.edu (groups.csail.mit.edu)... 128.30.2.44\n",
            "Connecting to groups.csail.mit.edu (groups.csail.mit.edu)|128.30.2.44|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 1785558 (1.7M)\n",
            "Saving to: ‘/data/trivia10k13/train.tsv’\n",
            "\n",
            "/data/trivia10k13/t 100%[===================>]   1.70M  2.61MB/s    in 0.7s    \n",
            "\n",
            "2021-06-03 08:19:13 (2.61 MB/s) - ‘/data/trivia10k13/train.tsv’ saved [1785558/1785558]\n",
            "\n",
            "--2021-06-03 08:19:13--  https://groups.csail.mit.edu/sls/downloads/movie/trivia10k13test.bio\n",
            "Resolving groups.csail.mit.edu (groups.csail.mit.edu)... 128.30.2.44\n",
            "Connecting to groups.csail.mit.edu (groups.csail.mit.edu)|128.30.2.44|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 438729 (428K)\n",
            "Saving to: ‘/data/trivia10k13/test.tsv’\n",
            "\n",
            "/data/trivia10k13/t 100%[===================>] 428.45K   995KB/s    in 0.4s    \n",
            "\n",
            "2021-06-03 08:19:14 (995 KB/s) - ‘/data/trivia10k13/test.tsv’ saved [438729/438729]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-D8_xEVDN-dj",
        "outputId": "fe057421-cf99-47e0-cdae-d56f342434b9"
      },
      "source": [
        "def get_sentence(file_path):\n",
        "    '''\n",
        "        Input:\n",
        "            file_path - path to the tsv file\n",
        "        Output:\n",
        "            sentences - list of sentences in string format\n",
        "            tags - list associated tags in string format\n",
        "    '''\n",
        "    sentences = []\n",
        "    tags = []\n",
        "    with open(file_path) as f:\n",
        "        contents = f.read()\n",
        "        sens_tags = contents.split(\"\\n\\n\")\n",
        "        for sen_tag in sens_tags:\n",
        "            words_tags = sen_tag.split(\"\\n\")\n",
        "            while (\"\" in words_tags):\n",
        "                words_tags.remove(\"\")\n",
        "            sen = ' '.join([word_tag.split(\"\\t\")[1] for word_tag in words_tags])\n",
        "            tag = ' '.join([word_tag.split(\"\\t\")[0] for word_tag in words_tags])\n",
        "            sentences.append(sen)\n",
        "            tags.append(tag)\n",
        "\n",
        "    return sentences, tags\n",
        "\n",
        "\n",
        "train_path = \"/data/eng/train.tsv\"\n",
        "test_path = \"/data/eng/test.tsv\"\n",
        "\n",
        "sentences, tags = get_sentence(train_path)\n",
        "test_sentences, test_tags = get_sentence(test_path)\n",
        "\n",
        "print(\"The train dataset has {} sentences.\".format(len(sentences)))\n",
        "print(\"The test dataset has {} sentences.\".format(len(test_sentences)))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The train dataset has 9776 sentences.\n",
            "The test dataset has 2444 sentences.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SJyYqFkQNCoW"
      },
      "source": [
        "<a name=\"1.2\"></a>\n",
        "## 1.2 Exploratory Analysis"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 198
        },
        "id": "Q6KzO6mq8326",
        "outputId": "282abcf1-586b-4c59-87ce-dddcd4c0b66d"
      },
      "source": [
        "# Take a look at the data\n",
        "df = pd.read_csv(train_path, delimiter=\"\\t\", names=[\"Tag\", \"Word\"])\n",
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Tag</th>\n",
              "      <th>Word</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>O</td>\n",
              "      <td>what</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>O</td>\n",
              "      <td>movies</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>O</td>\n",
              "      <td>star</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>B-ACTOR</td>\n",
              "      <td>bruce</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>I-ACTOR</td>\n",
              "      <td>willis</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       Tag    Word\n",
              "0        O    what\n",
              "1        O  movies\n",
              "2        O    star\n",
              "3  B-ACTOR   bruce\n",
              "4  I-ACTOR  willis"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tBVUmUQtFRJ1"
      },
      "source": [
        "<a name=\"1.2.1\"></a>\n",
        "### 1.2.1 Sentence Length \n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 584
        },
        "id": "h2J38zVa-wkD",
        "outputId": "a8c282e7-c675-42e9-a23a-461f0b386d0b"
      },
      "source": [
        "plt.style.use(\"dark_background\")\n",
        "\n",
        "# How long are the sentences?\n",
        "def plot_sentence_length_histogram(list_sentences):\n",
        "    '''\n",
        "      Input:\n",
        "        list_sentences - a list of sentences\n",
        "      Output:\n",
        "        [print] - Min, Max, Median and Average value of sentence length\n",
        "        [plot] - Histogram plot of sentence length\n",
        "    '''\n",
        "    lengths = [len(sen.split(' ')) for sen in list_sentences]\n",
        "    a4_dims = (11.7, 8.27)\n",
        "    fig, ax = plt.subplots(figsize=a4_dims)\n",
        "    sns.histplot(lengths)\n",
        "    plt.xlabel(\"Number of tokens in a sentence\")\n",
        "    plt.ylabel(\"Number of occurrences\")\n",
        "    print(\"Min: \",np.min(lengths))\n",
        "    print(\"Max: \",np.max(lengths))\n",
        "    \n",
        "    print(\"Median: \",np.median(lengths))\n",
        "    print(\"Average: \",round(np.mean(lengths),2))\n",
        "\n",
        "plot_sentence_length_histogram(sentences)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Min:  1\n",
            "Max:  47\n",
            "Median:  9.0\n",
            "Average:  10.18\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAskAAAHuCAYAAABtdJH+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3de1RVdcL/8Q8XUbMUhjOJXAIxus00MzahFpWVt9RJrMycasTLSPXkrZoULQfXkzOJXczV8zyVhEQTDd4q8dEUvFWrEkmuBiKYIaColJKZKeD+/eGv8+QX9Vh5zkZ4v9b6rsX5nnP2/hzaKz9uv2dvL0mWAAAAADh52x0AAAAAaGkoyQAAAICBkgwAAAAYKMkAAACAgZIMAAAAGHztDuAO+/fvV2Vlpd0xAAAA0IKFh4fr0ksvPe1zrbIkV1ZWKjo62u4YAAAAaMFyc3PP+BzLLQAAAAADJRkAAAAwUJIBAAAAAyUZAAAAMFCSAQAAAAMlGQAAADBQkgEAAAADJRkAAAAwUJIBAAAAAyUZAAAAMFCSAQAAAAMlGQAAADBQkgEAAAADJRkAAAAwUJIBAAAAAyUZAAAAMFCSAQAAAIPbSnJKSor27dun4uJi59y8efNUWlqqwsJCvfPOO+rSpYvzuYSEBJWXl2v79u0aOHCgc37QoEHavn27ysvLNX36dHfFBQAAAE5huWPcfPPNVs+ePa3i4mLn3IABAywfHx9LkjV37lxr7ty5liTr6quvtgoKCiw/Pz8rIiLCqqiosLy9vS1vb2+roqLC6t69u9WuXTuroKDAuvrqq13uOzc31y2ficFgMBgMBoPResbZOqPbziR/9NFH+vrrr0+Zy87OVlNTkyRp8+bNCg0NlSTFxsYqIyNDx48f15dffqmKigr16tVLvXr1UkVFhXbt2qWGhgZlZGQoNjbWXZEBAAAASZKvXTseN26cFi9eLEkKCQnR5s2bnc9VV1crJCREklRVVXXKfO/evU+7vQkTJig+Pl6S5HA43BUbLcDC1EVyBHdrNl+3Z6/ix46zIREAAGhtbCnJM2fOVGNjo9LT08/bNpOTk5WcnCxJys3NPW/bRcvjCO6m7OqdzeYHhPawIQ0AAGiNPF6S4+Li9Kc//Un9+vVzztXU1CgsLMz5ODQ0VDU1NZJ0xnkAAADAXTx6CbhBgwZp2rRpGjZsmI4ePeqcz8zM1KhRo+Tn56eIiAhFRUVpy5Ytys3NVVRUlCIiItSuXTuNGjVKmZmZnowMAACANshtZ5Lffvtt3XrrrXI4HKqqqlJiYqJmzJih9u3bKzs7W9LJL+898sgjKikp0ZIlS1RSUqLGxkY9+uijOnHihCRp4sSJWrt2rXx8fLRo0SKVlJS4KzIAAAAgSfLSyctctCq5ubmKjo62Owbc5J21759xTfLdgwbbkAgAAFyIztYZueMeAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABg8LU7ANqOhamL5Aju1my+bs9exY8dZ0MiAACA06Mkw2Mcwd2UXb2z2fyA0B42pAEAADgzllsAAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGX7sDAO62MHWRHMHdms3X7dmr+LHjbEgEAABaOkoyWj1HcDdlV+9sNj8gtIcNaQAAwIWA5RYAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAICBkgwAAAAYKMkAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAICBkgwAAAAYKMkAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAICBkgwAAAAYKMkAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAICBkgwAAAAY3FaSU1JStG/fPhUXFzvnAgIClJWVpR07digrK0v+/v7O5xYsWKDy8nIVFhaqZ8+ezvnRo0drx44d2rFjh0aPHu2uuAAAAICT20ryG2+8oTvuuOOUuYSEBK1fv15XXHGF1q9fr4SEBEnS4MGDFRUVpaioKMXHx+uVV16RdLJUJyYmqnfv3urVq5cSExNPKdYAAACAO7itJH/00Uf6+uuvT5mLjY1VWlqaJCktLU3Dhw93zr/55puSpJycHPn7+ysoKEiDBg1Sdna2Dh48qEOHDik7O7tZ8QYAAADON19P7qxr166qra2VJNXW1qpr166SpJCQEFVVVTlfV11drZCQkDPOn86ECRMUHx8vSXI4HO76CAAAAGgDbP3inmVZ521bycnJio6OVnR0tOrq6s7bdgEAAND2eLQk79u3T0FBQZKkoKAg7d+/X5JUU1OjsLAw5+tCQ0NVU1NzxnkAAADAnTy63CIzM1NxcXFKSkpSXFycVqxY4ZyfOHGiMjIy1Lt3b9XX16u2tlZr167VP//5T+eX9QYOHKgZM2Z4MjLaoIWpi+QI7tZsvm7PXsWPHWdDIgAA4GluK8lvv/22br31VjkcDlVVVSkxMVFz587VkiVLNH78eFVWVmrkyJGSpNWrV2vIkCGqqKjQd999p7Fjx0qSDh48qGeeeUa5ubmSpP/8z//UwYMH3RUZkCQ5grspu3pns/kBoT1sSAMAAOzgtpJ8//33n3a+f//+p52fOHHiaedTU1OVmpp63nIBAAAArnDHPQAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMtpTkqVOnatu2bSouLtbbb7+t9u3bKyIiQps3b1Z5ebkyMjLUrl07SZKfn58yMjJUXl6uzZs3Kzw83I7IAAAAaEM8XpKDg4M1efJkXX/99br22mvl4+OjUaNGKSkpSfPnz1dUVJQOHjyo8ePHS5LGjx+vgwcPKioqSvPnz1dSUpKnIwMAAKCNseVMsq+vrzp27CgfHx9ddNFF2rt3r26//XYtW7ZMkpSWlqbhw4dLkmJjY5WWliZJWrZsmfr162dHZAAAALQhHi/Je/bs0fPPP6/du3dr7969qq+v19atW3Xo0CE1NTVJkqqrqxUSEiJJCgkJUVVVlSSpqalJ9fX1CgwMbLbdCRMmKDc3V7m5uXI4HJ77QAAAAGh1PF6S/f39FRsbq+7duys4OFidOnXSHXfc8Yu3m5ycrOjoaEVHR6uuru48JAUAAEBb5fGS3L9/f+3atUt1dXVqbGzUO++8o5iYGPn7+8vHx0eSFBoaqpqaGklSTU2NwsLCJEk+Pj7q0qWLvvrqK0/HBgAAQBvi6+kd7t69W3369FHHjh119OhR9evXT5999pk2btyoESNGaPHixYqLi9OKFSskSZmZmYqLi9PmzZs1YsQIbdiwwdORcQYLUxfJEdyt2Xzdnr2KHzvOhkQAAADnh8dL8pYtW7Rs2TLl5eWpsbFR+fn5WrhwoVatWqWMjAzNmTNH+fn5SklJkSSlpKToX//6l8rLy/X1119r1KhRno6MM3AEd1N29c5m8wNCe9iQBgAA4PxxWZKTkpI0Z84cHT16VGvWrNHvfvc7PfbYY0pPT//ZO509e7Zmz559ytyuXbvUu3fvZq89duyYRo4c+bP3BQAAAPxULtckDxw4UIcPH9af/vQnffnll7r88sv15JNPeiIbAAAAYAuXJdnX9+TJ5qFDh2rp0qX65ptv3B4KAAAAsJPL5Rb/+7//q9LSUh09elSPPPKIHA6Hvv/+e09kAwAAAGzhsiTPmDFD8+bNU319vU6cOKHvvvtOsbGxnsgGXBC4ygcAAK2Py5LcsWNH/cd//Icuu+wyPfTQQwoODtaVV16pVatWeSIf0OJxlQ8AAFofl2uSU1NTdfz4cd14442STt7cY86cOW4PBgAAANjFZUnu0aOHnnvuOTU0NEiSjh49Ki8vL7cHAwAAAOzisiQfP35cHTp0kGVZkqTIyEgdO3bM7cEAAAAAu7hck5yYmKg1a9YoLCxMb731lmJiYjRmzBgPRAMAAADs4bIkr1u3Tnl5eerTp4+8vLw0ZcoUffXVV57IBgAAANjC5XKL4cOHq7GxUatXr9aqVavU2NjIJeAAAADQqrksyYmJiafcZa++vl6JiYluDQUAAADYyWVJ9vZu/pIfblUNAAAAtEYuS/Jnn32mF154QZGRkYqMjNQLL7ygrVu3eiIbAAAAYAuXJXnSpEk6fvy4Fi9erMWLF+vYsWN69NFHPZENAAAAsIXLdRPfffedZsyYoRkzZngiDwAAAGA7lyU5KipKf/vb3xQREXHKWuR+/fq5NRgAAABgF5cleenSpXr11Vf1+uuvq6mpyROZAAAAAFu5LMmNjY169dVXPZEFAAAAaBFcfnFv5cqVeuSRRxQUFKSAgADnAAAAAForl2eS4+LiJElPPvmkc86yLPXo0cN9qQAAAAAbuSzJkZGRnsgBAAAAtBgul1t07NhRTz31lF577TVJ0uWXX66hQ4e6PRgAAABgF5clOTU1VcePH9eNN94oSaqpqdGcOXPcHgwAAACwi8uS3KNHDz333HNqaGiQJB09elReXl5uDwYAAADYxWVJPn78uDp06CDLsiSdXKN87NgxtwcDAAAA7OLyi3uJiYlas2aNwsLC9NZbbykmJkZjxozxQDQAAADAHmctyV5eXgoICNDdd9+tPn36yMvLS1OmTNFXX33lqXwAAACAx521JFuWpWnTpmnp0qVavXq1pzIBAAAAtnK5JnndunV64oknFBoayh33AAAA0Ca4XJN83333SZIeffRR5xx33AMAAEBr5nJNckJCgpYsWeKpPAAAAIDtzrrcwrIsPfnkk57KAgAAALQIrEkGAAAADKxJBgAAAAwuS3JkZKQncgAAAAAthsuS/Je//OW08//617/OexgAAACgJXBZkqOjo50/d+jQQf369VNeXh4lGQAAAK2Wy5I8efLkUx536dJFGRkZbgsEAAAA2M3l1S1MR44cUffu3d2RBQAAAGgRXJ5JzszMlGVZkiRvb29dc8013FwEAAAArZrLkvz88887f25sbFRlZaVqamrcGgoAAACwk8uSvHv3bu3du1fHjh2TdPLLe+Hh4aqsrHR7OAAAAMAOLtckL126VCdOnHA+bmpq0tKlS90aCgAAALCTy5Ls6+urhoYG5+OGhgb5+fm5NRQAAABgJ5cl+cCBA7rzzjudj4cNG6a6ujq3hgIAAADs5HJN8sMPP6z09HT913/9lySpurpao0ePdnswAAAAwC4uS/IXX3yhG264QZ06dZJ08jrJAAAAQGvmcrnFP/7xD3Xp0kVHjhzRkSNH5O/vr2eeecYT2QAAAABbuCzJgwcPVn19vfPxoUOHNGTIELeGAgAAAOzksiT7+PiccjWLDh06qH379m4NBQAAANjJ5Zrk9PR0rV+/XqmpqZKksWPHKi0tze3BAAAAALu4LMnz5s1TYWGh+vfvL0l65plnlJWV5fZgAAAAgF1clmRJys/PV7t27WRZlvLz892dCWjVFqYukiO4W7P5uj17FT92nA2JAACAyWVJvvfee/Xcc89p06ZN8vLy0ssvv6wnn3xSy5cv90Q+oNVxBHdTdvXOZvMDQnvYkAYAAJyOy5L81FNPKTo6WgcOHJAkORwOrVu3jpIMAACAVsvl1S28vb2dBVmSvvrqK3l7u3wbAAAAcMFyeSZ5zZo1WrNmjf79739Lku677z6tXr3a7cEAAAAAu7gsydOmTdNdd92lm266SZK0cOFCvffee24PBgAAANjlnK5u8e677+rdd991dxYAAACgRWBxMQAAAGCgJAMAAACGM5bkdevWSZLmzp3rsTAAAABAS3DGktytWzfdcMMNGjZsmP7whz+oZ8+ep4xfokuXLlq6dKlKS0tVUlKiPn36KCAgQFlZWdqxY4eysrLk7+/vfP2CBQtUXl6uwsLCX7xvAAAAwJUzfnHv73//u2bNmqXQ0FC9+OKLpzxnWZb69ev3s3e6YMECrVmzRvfee6/atWuniy66SDNnztT69euVlJSk6dOnKyEhQQkJCRo8eLCioqIUFRWl3r1765VXXlGfPn1+9r4BAAAAV85YkpcvX67ly5fr6aef1pw5c87bDjt37qxbbrlFY8aMkSQ1NDSovr5esbGxuvXWWyVJaWlp2rRpkxISEhQbG6s333xTkpSTkyN/f38FBQWptrb2vGUCAAAAfszlJeDmzJmjO++8U7fccoskadOmTVq1atXP3mH37t114MABpaam6ve//722bt2qKVOmqGvXrs7iW1tbq65du0qSQkJCVFVV5Xx/dXW1QkJCmpXkCRMmKD4+XtLJW2cDAAAAP5fLq1v885//1JQpU1RSUqKSkhJNmTJF//jHP372Dn19fXXdddfplVde0XXXXacjR44oISGh2essy/pJ201OTlZ0dLSio6NVV1f3s/MBAAAALs8kDx06VH/4wx+cpTUtLU35+fl66qmnftYOq6urVV1drS1btkiSli1bpoSEBO3bt8+5jCIoKEj79++XJNXU1CgsLMz5/tDQUNXU1PysfQMAAADn4pyuk/zjK0106dLlF+1w3759qqqq0hVXXCFJ6tevn0pKSpSZmam4uDhJUlxcnFasWCFJyszM1OjRoyVJvXv3Vn19PeuRAQAA4FYuzyQ/++yzys/P18aNG+Xl5aVbbrnltMsjfopJkyYpPT1dfn5++uKLLzR27Fh5e3tryZIlGj9+vCorKzVy5EhJ0urVqzVkyBBVVFTou+++09ixY3/RvgEAAABXXJbkjIwMbdq0SdHR0ZKk6dOna9++fb9op4WFhc7t/Vj//v1P+/qJEyf+ov0BAAAAP4XLkiydvNrEypUr3Z0FAAAAaBHOaU0yAAAA0JZQkgEAAADDWUuyt7e3SktLPZUFAAAAaBHOWpJPnDihsrKyU65TDAAAALR2Lr+4FxAQoM8//1xbtmzRkSNHnPOxsbFuDQYAAADYxWVJnjVrlidyAAAAAC2Gy5L84Ycf6rLLLlNUVJTWr1+vjh07ysfHxxPZAAAAAFu4vLrFX//6Vy1btkyvvfaaJCkkJETvvfee24MBAAAAdnFZkh999FHFxMTom2++kSRVVFTo0ksvdXswAAAAwC4uS/KxY8fU0NDgfOzj4yPLstwaCgAAALCTy5L8wQcfaMaMGerYsaP69++vpUuXcotqAAAAtGouS3JCQoIOHDig4uJiPfTQQ1q9erWefvppT2QDAAAAbOHy6haWZSktLU05OTmyLEtlZWWeyAUAAADYxmVJHjJkiF599VXt3LlTXl5e6t69ux566CGtWbPGE/kAAAAAj3NZkl944QXddttt2rlzpyQpMjJSq1atoiQDAACg1XK5Jvnw4cPOgixJX3zxhQ4fPuzWUAAAAICdzngm+a677pIkffbZZ1q1apWWLFkiy7J07733Kjc312MBAQAAAE87Y0m+8847nT/v27dPffv2lSQdOHBAHTt2dH8yAAAAwCZnLMnjxo3zZA4AAACgxXD5xb2IiAhNmjRJERER8vX9v5fHxsa6NRgAAABgF5cl+b333lNKSopWrlypEydOeCITAAAAYCuXJfn777/Xyy+/7IksAAAAQIvgsiQvWLBAf//735WVlaVjx4455/Pz890aDAAAALCLy5J87bXX6i9/+Ytuv/1253ILy7LUr18/t4cDAAAA7OCyJN97772KjIxUQ0ODJ/IAAAAAtnN5x71t27bJ39/fE1kAAACAFsHlmWR/f39t375dubm5p6xJ5hJwAAAAaK1cluTExERP5AAAAABaDJcl+cMPP/REDgAAAKDFcFmSv/nmG1mWJUny8/NTu3btdOTIEXXp0sXt4QAAAAA7uCzJnTt3PuVxbGys+vTp47ZAAAAAgN1cXt3CtGLFCg0aNMgdWQAAAIAWweWZ5Lvuusv5s7e3t66//np9//33bg0FAAAA2MllSb7zzjudPzc2NurLL7/k8m8AAABo1VyW5HHjxnkiBwAAANBinLEkz5o164xvsixLc+bMcUsgAKdamLpIjuBuzebr9uxV/Fj+EgsAgDucsSQfOXKk2VynTp00fvx4BQYGUpIBD3EEd1N29c5m8wNCe9iQBgCAtuGMJfnFF190/nzxxRdrypQpGjt2rDIyMvTCCy94JBwAAABgh7NeAi4gIEDPPPOMioqK5Ovrq+uuu04JCQk6cOCAp/IBAAAAHnfGM8nz5s3T3XffrYULF+raa6897fILAAAAoDU645nkJ554QsHBwXr66ae1Z88e1dfXq76+Xt98843q6+s9mREAAADwqDOeSfbx8fFkDgAAAKDF+Mm3pQYAAABaO0oyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYKAkAwAAAAZKMgAAAGCgJAMAAAAGSjIAAABgoCQDAAAABkoyAAAAYPC1OwBajoWpi+QI7tZsvm7PXsWPHWdDIgAAAHtQkuHkCO6m7OqdzeYHhPawIQ0AAIB9KMlAK8O/CAAA8MtRkoFWhn8RAADgl+OLewAAAIDBtpLs7e2tvLw8rVy5UpIUERGhzZs3q7y8XBkZGWrXrp0kyc/PTxkZGSovL9fmzZsVHh5uV2QAAAC0EbaV5ClTpqi0tNT5OCkpSfPnz1dUVJQOHjyo8ePHS5LGjx+vgwcPKioqSvPnz1dSUpJdkQEAANBG2FKSQ0JCNHToUL3++uvOudtvv13Lli2TJKWlpWn48OGSpNjYWKWlpUmSli1bpn79+nk+MAAAANoUW0rySy+9pGnTpunEiROSpMDAQB06dEhNTU2SpOrqaoWEhEg6WairqqokSU1NTaqvr1dgYGCzbU6YMEG5ubnKzc2Vw+Hw0CcBAABAa+Txkjx06FDt379feXl553W7ycnJio6OVnR0tOrq6s7rtgEAANC2ePwScDExMRo2bJiGDBmiDh06qHPnzlqwYIH8/f3l4+OjpqYmhYaGqqamRpJUU1OjsLAw1dTUyMfHR126dNFXX33l6dgAAABoQzx+JnnmzJkKCwtT9+7dNWrUKG3YsEEPPvigNm7cqBEjRkiS4uLitGLFCklSZmam4uLiJEkjRozQhg0bPB0ZAAAAbUyLuU7y9OnT9fjjj6u8vFyBgYFKSUmRJKWkpCgwMFDl5eV6/PHHlZCQYHNSAAAAtHa23nHvgw8+0AcffCBJ2rVrl3r37t3sNceOHdPIkSM9HQ0AAABtWIs5kwwAAAC0FJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADA4Gt3AAAt08LURXIEd2s2X7dnr+LHjrMhEQAAnkNJBnBajuBuyq7e2Wx+QGgPG9IAAOBZLLcAAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADL52B4D7LExdJEdwt2bzdXv2Kn7sOBsSAQAAXBgoya2YI7ibsqt3NpsfENrDhjQAAAAXDkoyAFvwLx0AgJaMkgzAFvxLBwCgJeOLewAAAICBM8kAzguWTwAAWhNKMoDzguUTAIDWhOUWAAAAgIGSDAAAABgoyQAAAICBkgwAAAAYPF6SQ0NDtWHDBn3++efatm2bJk+eLEkKCAhQVlaWduzYoaysLPn7+zvfs2DBApWXl6uwsFA9e/b0dGQAAAC0MR4vyY2NjXriiSf0m9/8Rn369NGjjz6qq6++WgkJCVq/fr2uuOIKrV+/XgkJCZKkwYMHKyoqSlFRUYqPj9crr7zi6cgAAABoYzxekmtra5Wfny9J+vbbb1VaWqqQkBDFxsYqLS1NkpSWlqbhw4dLkmJjY/Xmm29KknJycuTv76+goCBPxwYAAEAbYuua5PDwcPXs2VM5OTnq2rWramtrJZ0s0l27dpUkhYSEqKqqyvme6upqhYSENNvWhAkTlJubq9zcXDkcDs98AAAAALRKtpXkTp06afny5Zo6daoOHz7c7HnLsn7S9pKTkxUdHa3o6GjV1dWdr5gAAABog2wpyb6+vlq+fLnS09P17rvvSpL27dvnXEYRFBSk/fv3S5JqamoUFhbmfG9oaKhqamo8HxoAAABthi0lOSUlRaWlpZo/f75zLjMzU3FxcZKkuLg4rVixwjk/evRoSVLv3r1VX1/vXJYBAAAAuIOvp3cYExOj0aNHq6ioyPkFvpkzZwM+dLMAABBUSURBVGru3LlasmSJxo8fr8rKSo0cOVKStHr1ag0ZMkQVFRX67rvvNHbsWE9HBgAAQBvj8ZL88ccfy8vL67TP9e/f/7TzEydOdGckAAAA4BTccQ8AAAAwUJIBAAAAAyUZAAAAMFCSAQAAAAMlGQAAADBQkgEAAAADJRkAAAAwUJIBAAAAAyUZAAAAMHj8jnsA8HMsTF0kR3C3ZvN1e/Yqfuw4GxIBAFozSjKAC4IjuJuyq3c2mx8Q2sOGNACA1o7lFgAAAICBkgwAAAAYKMkAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAICBkgwAAAAYKMkAAACAgZIMAAAAGCjJAAAAgIGSDAAAABgoyQAAAIDB1+4AAOAOC1MXyRHcrdl83Z69ih87zoZEAIALCSUZQKvkCO6m7OqdzeYHhPawIQ0A4ELDcgsAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAAAMDA1S0AQFwyDgBwKkoyAIhLxgEATsVyCwAAAMBASQYAAAAMLLe4gLBmEgAAwDMoyRcQ1kwCAAB4BiUZAM4j/sUHAFoHSjIAnEf8iw8AtA58cQ8AAAAwUJIBAAAAAyUZAAAAMFCSAQAAAAMlGQAAADBwdQsAsBGXjAOAlomSDAA24pJxANAysdwCAAAAMFCSAQAAAAMlGQAAADBQkgEAAAADJRkAAAAwcHULG3HpJwAAgJaJkmwjLv0E4KfiL9cA4BmU5POIP7wAuBt/uQYAz6Akn0f84QUAANA68MU9AAAAwMCZZABoxVgGBgA/DyUZAFoxloEBwM/DcgsAAADAwJlkAIATyzMA4CRKMgDAieUZAHASyy0AAAAAA2eSAQA/209dnsFyDgAXigumJA8aNEgLFiyQj4+PXn/9dSUlJdkdCQDavJ+6PON8LeegbANwtwuiJHt7e+u///u/NWDAAFVXVys3N1eZmZkqLS21OxoAwAY/tWxzxhvAT3VBlORevXqpoqJCu3btkiRlZGQoNjaWkgwAOCfuPuPt7hJ+oZT2CyUn7HWhHCdekiy7Q7hyzz336I477tCECRMkSQ8++KB69+6tSZMmOV8zYcIExcfHS5KuvPJKlZWV/aJ9OhwO1dXV/aJt4MLGMQCOAXAMgGOgdQsPD9ell1562ucuiDPJ5yI5OVnJycnnbXu5ubmKjo4+b9vDhYdjABwD4BgAx0DbdUFcAq6mpkZhYWHOx6GhoaqpqbExEQAAAFqzC6Ik5+bmKioqShEREWrXrp1GjRqlzMxMu2MBAACglfKRNNvuEK5YlqXy8nKlp6dr0qRJeuutt/TOO++4fb95eXlu3wdaNo4BcAyAYwAcA23TBfHFPQAAAMCTLojlFgAAAIAnUZIBAAAAAyX5NAYNGqTt27ervLxc06dPtzsOPCAlJUX79u1TcXGxcy4gIEBZWVnasWOHsrKy5O/vb2NCuFtoaKg2bNigzz//XNu2bdPkyZMlcRy0Je3bt1dOTo4KCgq0bds2zZ49W5IUERGhzZs3q7y8XBkZGWrXrp29QeF23t7eysvL08qVKyVxDLRlFuP/hre3t1VRUWF1797dateunVVQUGBdffXVtudiuHfcfPPNVs+ePa3i4mLnXFJSkjV9+nRLkjV9+nRr7ty5tudkuG8EBQVZPXv2tCRZF198sVVWVmZdffXVHAdtbHTq1MmSZPn6+lqbN2+2evfubS1evNi67777LEnWK6+8Yj388MO252S4dzz22GNWenq6tXLlSksSx0DbHbYHaFGjT58+1po1a5yPExISrISEBNtzMdw/wsPDTynJ27dvt4KCgizpZIHavn277RkZnhvvvfee1b9/f46DNjo6duxobd261erVq5d14MABy8fHx5Ka/xnBaH0jJCTEWrdunXXbbbc5SzLHQNscLLcwhISEqKqqyvm4urpaISEhNiaCXbp27ara2lpJUm1trbp27WpzInhKeHi4evbsqZycHI6DNsbb21v5+fnav3+/srOztXPnTh06dEhNTU2S+DOhLXjppZc0bdo0nThxQpIUGBjIMdBGUZKBc2RZlt0R4AGdOnXS8uXLNXXqVB0+fLjZ8xwHrduJEyfUs2dPhYaGqlevXrrqqqvsjgQPGjp0qPbv3891kSFJ8rU7QEvDLbDxg3379ikoKEi1tbUKCgrS/v377Y4EN/P19dXy5cuVnp6ud999VxLHQVtVX1+vjRs36oYbbpC/v798fHzU1NTEnwmtXExMjIYNG6YhQ4aoQ4cO6ty5sxYsWMAx0EZxJtnALbDxg8zMTMXFxUmS4uLitGLFCpsTwd1SUlJUWlqq+fPnO+c4DtoOh8OhLl26SJI6dOigAQMGqLS0VBs3btSIESMkcQy0djNnzlRYWJi6d++uUaNGacOGDXrwwQc5Btow2xdGt7QxePBgq6yszKqoqLBmzpxpex6G+8fbb79t7dmzxzp+/LhVVVVljRs3zvrVr35lrVu3ztqxY4eVnZ1tBQQE2J6T4b4RExNjWZZlFRYWWvn5+VZ+fr41ePBgjoM2NK699lorLy/PKiwstIqLi61Zs2ZZkqzu3btbOTk5Vnl5ubVkyRLLz8/P9qwM94++ffs6v7jHMdA2B7elBgAAAAwstwAAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAyUZAAXFMuy9PzzzzsfP/HEE0pMTDwv205NTdU999xzXrZ1NiNGjFBJSYk2bNhwynx4eLj+/Oc/u3x/XFycXn75ZXfFO8XHH3/skf38HHFxcerWrZvdMQC0UpRkABeU77//XnfffbcCAwPtjnIKHx+fc37t+PHjNWHCBN1+++2nzEdEROj+++8/39F+kZiYGLsjnNGYMWMUHBxsdwwArRQlGcAFpbGxUQsXLtRjjz3W7DnzTPDhw4clSX379tWmTZv03nvvaefOnXr22Wd1//33KycnR0VFRYqMjHS+p3///srNzVVZWZmGDh0qSfL29ta8efO0ZcsWFRYWKj4+3rndDz/8UCtWrFBJSUmzPKNGjVJRUZGKi4s1d+5cSdKsWbN00003KSUlRfPmzTvl9XPnztXNN9+s/Px8TZ06Ve3bt9eiRYtUVFSkvLw83Xrrrc32MWTIEH3yyScKDAzUgAED9Mknn2jr1q1asmSJOnXqJEnatWuXZs+era1bt6qoqEhXXnmlJOmWW25Rfn6+8vPzlZeXp4svvrjZ9n/8O9y4caOWLl2q0tJSvfXWW6f97/PXv/5VW7ZsUUFBgZYtW6aOHTs2e82Z9vu3v/3N+TuePXu2pJNn10tKSrRw4UJt27ZNa9euVYcOHXTPPffo+uuvV3p6uvLz89WhQwddd9112rRpkz777DOtWbNGQUFBkqSNGzdq7ty5ysnJUVlZmW666Sbnf9fnnntOxcXFKiws1MSJEyXpjNsB0PbYfkcTBoPBONdx+PBh65JLLrF27dplde7c2XriiSesxMRES5KVmppq3XPPPae8Vjp556yDBw9aQUFBlp+fn1VdXW3Nnj3bkmRNnjzZmj9/vvP977//vuXl5WVdfvnlVlVVldW+fXtrwoQJ1lNPPWVJsvz8/Kzc3FwrIiLC6tu3r/Xtt99aERERzXJ269bNqqystBwOh+Xj42OtX7/eio2NtSRZGzdutP74xz82e8+P7/AlyXr88cetlJQUS5J15ZVXWpWVlVb79u2tuLg46+WXX7aGDx9uffjhh5a/v78VGBhoffDBB9ZFF11kSbKmTZvmvGPcrl27rIkTJ1qSrEceecRKTk62JFmZmZnWjTfeaEmyOnXqZPn4+Jz29/1DtkOHDlkhISGWl5eX9cknn1gxMTHNXv+rX/3K+fMzzzzj3O+Px+n2O2DAAOu1116zJFleXl7WypUrrZtvvtkKDw+3GhoarN///veWJGvx4sXWAw880Oz36Ovra3388ceWw+GwJFkjR450/u42btxoPf/885Z08o6q2dnZliTr4YcftpYuXer83AEBAWfdDoPBaFvDVwBwgTl8+LDefPNNTZ48WUePHj2n9+Tm5qq2tlaStHPnTmVlZUmSiouLddtttzlft2TJElmWpYqKCn3xxRe66qqrNHDgQP3ud7/TiBEjJEldunRRVFSUjh8/ri1btujLL79str/o6Ght2rRJdXV1kqT09HTdcsstWrFixTl/zptuusm59risrEyVlZW64oorJEm33367rr/+eg0cOFCHDx/W0KFDdc011zjXEPv5+enTTz91buudd96RJG3dulV33323pJPrjV988UWlp6frnXfeUU1NzVnzbNmyxfmagoICRURENFuz/Nvf/lZz5syRv7+/Lr74Yq1du7bZdk6334EDB2rgwIHKz8+XJF188cWKiorS7t27tWvXLhUWFjrzR0RENNvmlVdeqd/+9rfKzs6WdHL5y969e0/7+X94f//+/fXqq6+qqalJknTw4EH95je/Oet2ALQdlGQAF6SXXnpJeXl5Sk1Ndc41NjbK2/vkKjIvLy/5+fk5nzt27Jjz5xMnTjgfnzhxQr6+//e/QsuyTtmPZVny8vLSpEmTnMX6B3379tWRI0fO34f6CXbu3KnIyEhdccUV2rp1q7y8vJSdnX3GNc0/fN6mpibn501KStKqVas0ZMgQffzxxxo0aJDKysrOuM8f/w5/vJ0fe+ONNzR8+HAVFRUpLi7utEtETrdfLy8vPfvss1q4cOEprw0PD2+239Mt4fDy8tLnn3+uG2+88Zw//+m42g6AtoM1yQAuSAcPHtSSJUs0fvx459yXX36pP/7xj5KkYcOGnVKSz9W9994rLy8vRUZGKjIyUmVlZVq7dq0eeeQRZ7mKiorSRRdddNbtbNmyRX379lVgYKC8vb315z//WR988MFZ33P48GFdcsklzscfffSRHnjgAec+L7vsMmeJrays1D333KM333xT11xzjTZv3qyYmBj16NFDknTRRRcpKirqrPuLjIzUtm3bNG/ePOXm5uqqq646+y/nHFxyySXau3evfH19ndnPZb9r167VuHHjnOuog4OD9etf//qs+/rx76usrEy//vWv1adPH0mSr6+vrrnmmrO+Pzs7Ww899JDzS5cBAQE/azsAWidKMoAL1gsvvCCHw+F8nJycrL59+6qgoEA33HCDvv3225+8zd27d2vLli16//339fDDD+vYsWN6/fXXVVJSory8PBUXF+u1114769lISaqtrVVCQoI2btyowsJCbd26VZmZmWd9T1FRkZqamlRQUKCpU6fqf/7nf+Tt7a2ioiItXrxYY8aM0fHjx52vLysr0wMPPKClS5eqc+fOGjNmjP7973+rsLBQn376qcvSO3XqVOeX1hoaGvT++++f+y/qDGbNmqWcnBx9/PHH2r59+znvNzs7W2+//bY+/fRTFRUVadmyZaf8heF03njjDb366qvKz8+Xj4+PRowYoaSkJBUUFKigoMDl2eDXX39du3fvVlFRkQoKCnT//feroaHhJ28HQOvkpZOLkwEAAAD8f5xJBgAAAAyUZAAAAMBASQYAAAAMlGQAAADAQEkGAAAADJRkAAAAwEBJBgAAAAz/D36235JKkUO5AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 842.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qY8VCLDQhCbJ"
      },
      "source": [
        "### 1.2.2 Entity Length"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 584
        },
        "id": "XQrFzlrkhCm3",
        "outputId": "310ea49d-586a-4ce8-b009-2aaf519a4c9d"
      },
      "source": [
        "# How long are the entities?\n",
        "def plot_entity_length_histogram(series):\n",
        "    '''\n",
        "      Input:\n",
        "        series - a pandas series of the tags\n",
        "      Output:\n",
        "        [print] - Min, Max, Median and Average value of entity length\n",
        "        [plot] - Histogram plot of entity length\n",
        "    '''\n",
        "    tags_list=[tag for tag in series]\n",
        "    tag_length = []\n",
        "    current_length = 0\n",
        "    for tag in tags_list:\n",
        "        if tag.startswith(\"B\"):\n",
        "            tag_length.append(current_length)\n",
        "            current_length = 1\n",
        "        elif tag.startswith(\"I\"):\n",
        "            current_length += 1\n",
        "    tag_length = tag_length[1:]\n",
        "    \n",
        "    a4_dims = (11.7, 8.27)\n",
        "    fig, ax = plt.subplots(figsize=a4_dims)\n",
        "    sns.histplot(tag_length)\n",
        "    plt.xlabel(\"Number of tokens in a tag\")\n",
        "    plt.ylabel(\"Number of occurrences\")\n",
        "    print(\"Min: \",np.min(tag_length))\n",
        "    print(\"Max: \",np.max(tag_length))\n",
        "    print(\"Median: \",np.median(tag_length))\n",
        "    print(\"Average: \",round(np.mean(tag_length),2))\n",
        "\n",
        "plot_entity_length_histogram(df[\"Tag\"])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Min:  1\n",
            "Max:  16\n",
            "Median:  2.0\n",
            "Average:  1.81\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 842.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OcoUQIfwfOAe"
      },
      "source": [
        "### 1.2.3 Token frequency"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 567
        },
        "id": "b2kvMJJ3KvXu",
        "outputId": "757556df-681a-4624-c19b-94bf69835f53"
      },
      "source": [
        "def plot_top_non_stopwords_barchart(series, top=20, word=True):\n",
        "    '''\n",
        "      Input:\n",
        "        series - a pd.Series of words or tags\n",
        "        top - number of most common words to plot\n",
        "      Output:\n",
        "        [print] - No of distinct words in train dataset\n",
        "        [plot] - Barchart of most common words' occurrence\n",
        "    '''\n",
        "    stop=set(stopwords.words('english'))\n",
        "    value = 'words' if word == True else 'tags'\n",
        "    corpus=[word for word in series]\n",
        "    counter=Counter(corpus)\n",
        "    print(\"There are {} distinct {} in dataset\".format(len(dict(counter)), value))\n",
        "    print(dict(counter))\n",
        "\n",
        "    most=counter.most_common()\n",
        "    x, y=[], []\n",
        "    for word,count in most:\n",
        "        if (word not in stop):\n",
        "            x.append(count)\n",
        "            y.append(word)\n",
        "        if len(x) == top:\n",
        "          break\n",
        "    a4_dims = (11.7, 8.27)\n",
        "    fig, ax = plt.subplots(figsize=a4_dims)\n",
        "    sns.barplot(x=x,y=y)\n",
        "    plt.xlabel(\"Number of {} occurrences in a sentence\".format(value))\n",
        "    plt.ylabel(\"Most common {}s\".format(value))\n",
        "    return dict(counter)\n",
        "\n",
        "word_counter = plot_top_non_stopwords_barchart(df[\"Word\"], top=40)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "There are 6710 distinct words in dataset\n",
            "{'what': 2938, 'movies': 1702, 'star': 303, 'bruce': 37, 'willis': 24, 'show': 588, 'me': 958, 'films': 510, 'with': 1302, 'drew': 12, 'barrymore': 8, 'from': 895, 'the': 5331, '1980s': 74, 'starred': 245, 'both': 8, 'al': 23, 'pacino': 17, 'and': 777, 'robert': 86, 'deniro': 11, 'find': 810, 'all': 278, 'of': 1414, 'that': 1640, 'harold': 3, 'ramis': 2, 'bill': 33, 'murray': 14, 'a': 4224, 'movie': 3533, 'quote': 53, 'about': 1182, 'baseball': 11, 'in': 2789, 'it': 197, 'have': 335, 'mississippi': 3, 'title': 133, 'science': 163, 'fiction': 162, 'directed': 1122, 'by': 940, 'steven': 50, 'spielberg': 34, 'do': 231, 'you': 413, 'any': 738, 'thrillers': 21, 'sofia': 10, 'coppola': 26, 'leonard': 8, 'cohen': 4, 'songs': 18, 'been': 178, 'used': 16, 'elvis': 10, 'set': 39, 'hawaii': 2, 'is': 2214, 'references': 1, 'zydrate': 1, 'are': 765, 'there': 1062, 'musical': 107, 'patrick': 23, 'dempsey': 1, 'list': 795, 'westerns': 21, 'starring': 759, 'john': 155, 'wayne': 22, 'military': 56, 'related': 4, 'demi': 5, 'moore': 19, 'did': 770, 'stephen': 12, 'made': 300, 'sex': 6, 'or': 42, 'horror': 242, 'must': 90, 'be': 76, 'to': 413, 'watch': 53, 'ang': 2, 'lee': 56, 'direct': 268, 'costume': 2, 'drama': 199, '1800s': 1, 'who': 599, 'first': 100, 'james': 129, 'bond': 37, 'r': 653, 'rated': 1546, 'best': 147, 'picture': 24, 'oscar': 28, 'winners': 3, 'where': 213, 'can': 187, 'i': 433, 'buy': 10, 'sucker': 1, 'punch': 2, 'soundtrack': 64, 'british': 8, 'came': 189, 'out': 384, '1990s': 85, 'royal': 1, 'family': 97, 'camerons': 3, 'titanic': 21, 'was': 1559, 'last': 562, 'judy': 13, 'garland': 11, 'spoof': 1, 'charlie': 20, 'sheen': 10, 'id': 69, 'like': 174, 'documentary': 150, 'doctors': 1, 'chicago': 3, 'hospital': 3, 'action': 270, 'woo': 2, 'romantic': 129, 'film': 1201, 'protagonist': 1, 'doesnt': 4, 'fall': 2, 'love': 68, 'does': 66, 'original': 14, 'little': 23, 'mermaid': 4, 'hans': 5, 'christian': 23, 'anderson': 21, 'come': 127, 'on': 264, 'dvd': 8, 'disney': 48, 'beyonce': 1, 'sang': 7, 'shawn': 5, 'levy': 7, '2010s': 2, 'new': 46, 'david': 85, 'dhawan': 1, 'hannibal': 2, 'lector': 2, 'childrens': 15, 'billy': 39, 'crystal': 8, 'reviewed': 11, 'rock': 15, 'ballads': 1, 'were': 235, 'free': 1, 'willy': 9, 'filmed': 26, 'jennifer': 23, 'aniston': 9, 'main': 36, 'character': 99, 'wars': 32, 'im': 139, 'looking': 254, 'for': 612, 'song': 168, 'day': 46, 'has': 561, 'shortest': 1, 'trailer': 90, 'will': 70, 'always': 8, 'want': 121, 'see': 188, 'clint': 49, 'eastwood': 47, 'comedy': 322, 'whitney': 6, 'houstons': 1, 'u': 2, 'had': 149, 'fast': 9, 'furious': 8, 'jack': 53, 'nicholson': 11, 'an': 657, 'insane': 5, 'asylum': 1, 'dramatic': 10, 'donald': 6, 'sutherland': 9, 'rocky': 14, 'aaron': 4, 'sorkin': 1, 'trent': 1, 'reznor': 1, 'michael': 105, 'jackson': 32, '1970s': 38, 'political': 16, 'alec': 7, 'baldwin': 4, 'keaton': 13, 'play': 81, 'batman': 12, 'ewan': 7, 'macgregor': 3, 'singing': 3, 'superman': 11, 'romance': 110, 'popular': 117, '2002': 6, 'kevin': 45, 'smith': 42, 'jason': 51, 'mews': 1, 'suicide': 5, 'painless': 2, 'their': 7, 'world': 40, 'warcraft': 2, 'wes': 15, 'schwartzman': 1, 'savannah': 1, 'smiles': 1, 'owen': 4, 'wilson': 6, 'considered': 41, 'which': 345, 'features': 45, 'blaze': 1, 'glory': 4, 'maker': 1, 'planning': 2, 'produce': 2, 'farely': 1, 'brothers': 19, 'move': 21, 'frank': 36, 'sinatra': 3, 'something': 6, 'one': 44, 'armed': 2, 'man': 41, 'rick': 9, 'springfield': 2, 'steve': 41, 'martin': 52, 'cameron': 43, 'just': 24, 'two': 128, 'stars': 465, 'less': 4, 'angelina': 18, 'jolie': 16, 'released': 347, 'most': 62, '4': 6, 'winning': 22, '1998': 6, 'roadtrip': 1, 'ben': 15, 'stiller': 8, 'how': 190, 'many': 209, 'based': 79, 'texas': 7, 'chainsaw': 3, 'massacre': 4, 'jimmy': 13, 'stewart': 25, 'ever': 204, 'actor': 166, 'called': 82, 'matchstick': 2, 'men': 14, 'good': 542, 'nc': 151, '17': 152, 'not': 23, 'princess': 14, 'them': 57, 'comedies': 64, 'brad': 44, 'pitt': 33, '2000s': 35, 'godfather': 15, 'featured': 47, 'dude': 2, 'g': 240, 'war': 177, 'ii': 12, 'please': 95, 'locate': 7, 'george': 63, 'lucas': 27, '1996': 4, 'actors': 25, 'cameo': 4, 'spongebob': 1, 'year': 380, 'social': 4, 'network': 2, 'plot': 161, '1': 4, '1990': 169, '1991': 3, 'garry': 3, 'marshall': 15, '2011': 35, 'natalie': 9, 'portman': 2, 'dramas': 37, 'latest': 20, 'williams': 33, 'time': 39, 'car': 19, 'chases': 3, 'worst': 9, 'scary': 49, '2010': 184, 'thriller': 154, 'art': 2, 'heist': 3, 'late': 5, 'earl': 2, 'jones': 41, 'at': 61, 'end': 3, 'bronx': 3, 'tale': 17, 'top': 45, 'scorsesy': 1, 'moive': 2, 'pg': 791, '13': 558, 'ghosts': 13, 'dogs': 9, 'musicals': 26, 'antonio': 8, 'banderas': 7, 'hitchcock': 26, 'color': 3, 'animals': 6, 'five': 157, 'youre': 5, 'stories': 5, 'amanda': 7, 'seyfried': 3, 'ill': 11, 'back': 31, 'foreign': 8, 'blue': 8, 'rob': 23, 'reiner': 4, '1992': 2, 'drill': 1, 'team': 8, 'give': 55, 'ice': 15, 'baby': 11, 'vanilla': 1, 'atmospheric': 1, 'bambi': 1, 'teen': 29, 'got': 80, 'operas': 1, 'selma': 3, 'blair': 4, 'cars': 10, 'dragons': 5, 'clooney': 20, 'leonardo': 27, 'dicaprio': 24, 'danny': 25, 'devito': 10, 'krasinski': 1, 'look': 5, 'up': 71, 'could': 58, 'mobster': 4, 'friday': 8, '13th': 4, 'loved': 6, 'tomb': 1, 'raider': 1, 'carrel': 1, 'remake': 6, 'dirty': 8, 'dancing': 7, 'similar': 8, 'gnomeo': 2, 'juliet': 3, 'talking': 16, 'scorsese': 13, 'documentaries': 13, 'pollution': 2, 'china': 1, 'famous': 24, 'name': 486, 'cary': 17, 'grant': 22, 'eva': 3, 'marie': 1, 'saint': 4, 'billing': 1, 'backdraft': 1, 'theme': 32, '8': 3, 'mile': 4, 'america': 4, 'fererra': 1, 'susan': 11, 'sarandon': 7, 'jenna': 1, 'elfman': 4, 'julie': 15, 'andrews': 7, 'grease': 1, 'russel': 4, 'crowe': 9, 'mathematician': 1, '2001': 16, 'played': 99, 'ron': 25, 'weasley': 1, 'harry': 37, 'potter': 31, 'groups': 2, 'robbers': 1, 'robbing': 1, 'same': 6, 'bank': 9, 'whats': 111, 'coolio': 1, 'funniest': 8, 'mobsters': 2, 'recent': 31, 'sequel': 8, 'speed': 4, 'anthony': 28, 'minghella': 1, 'matt': 18, 'damon': 14, 'milos': 2, 'forman': 2, 'life': 26, 'together': 9, 'when': 97, 'hobbit': 4, 'woody': 24, 'allen': 33, 'other': 17, 'people': 62, 'think': 27, 'iii': 6, 'terrible': 2, 'highest': 31, 'kids': 26, 'release': 22, 'involving': 34, 'cartoons': 3, 'get': 52, 'biography': 104, 'real': 9, 'us': 5, 'president': 8, 'portrayed': 4, 'funny': 33, 'wedding': 5, 'toasts': 1, 'britney': 1, 'spears': 1, 'she': 8, 'satisfaction': 1, 'indiana': 19, 'judd': 7, 'apatow': 6, 'since': 5, 'dog': 9, 'chase': 25, 'statum': 1, 'nora': 1, 'ephron': 1, 'write': 5, 'spielbergs': 5, 'award': 15, 'well': 121, 'known': 16, 'julia': 26, 'roberts': 28, '1993': 1, '2012': 19, 'awful': 1, 'teenagers': 3, 'lost': 13, 'woods': 8, 'contained': 2, 'line': 19, 'phone': 3, 'home': 13, 'police': 28, 'pg13': 4, 'darren': 1, 'aranofsky': 1, 'reviews': 49, 'before': 19, '2000': 159, 'giraffe': 1, 'health': 7, 'care': 3, 'lists': 1, 'braddock': 1, 'includes': 5, 'your': 18, 'touch': 2, 'black': 26, 'keys': 1, 'critics': 23, 'non': 6, 'rating': 479, 'next': 12, 'coming': 40, 'courtney': 2, 'foxx': 4, 'looks': 2, 'great': 17, 'this': 106, 'shining': 61, 'boats': 3, 'beautiful': 3, 'mind': 9, 'score': 10, 'shierly': 1, 'temples': 2, 'greatest': 4, 'diaz': 8, 'named': 34, 'hugo': 1, 'high': 23, 'school': 17, 'sean': 35, 'penn': 7, 'horses': 4, 'sci': 114, 'fi': 114, 'laurence': 12, 'olivier': 10, '1940s': 6, 'bourne': 13, 'featuring': 66, 'alien': 26, 'invasion': 4, 'some': 288, 'lewis': 18, '3d': 4, 'animated': 87, 'monsters': 7, 'vs': 2, 'aliens': 11, 'appropriate': 5, 'toddlers': 1, 'somewhere': 6, 'tom': 74, 'hanks': 31, 'keanu': 10, 'reeves': 15, 'need': 20, 'quotes': 9, 'trailers': 4, 'lesbian': 2, 'concert': 2, '1973': 1, 'receive': 19, 'genre': 77, 'walk': 4, 'remember': 7, 'part': 13, 'beatles': 8, 'simon': 4, 'pegg': 1, 'uses': 11, 'abba': 1, 'ridley': 20, 'scott': 39, 'music': 18, 'luke': 7, 'my': 60, 'heart': 14, 'go': 26, 'won': 23, '1999': 9, 'twilight': 9, 'expected': 2, 'appear': 20, 'theatres': 1, 'four': 168, 'higher': 10, '1950s': 9, 'boxing': 8, 'hopkins': 13, 'everythings': 1, 'roses': 1, 'howard': 28, 'meryl': 8, 'streep': 9, 'mad': 13, '1969': 2, 'scarlett': 5, 'johansson': 2, 'spaceships': 2, 'candy': 5, 'helen': 8, 'hunt': 8, '1960s': 30, 'names': 10, 'books': 10, 'fifth': 2, 'freddy': 6, 'krueger': 4, 'gary': 19, 'ross': 2, 'tobey': 1, 'maguire': 1, 'peter': 44, 'than': 12, 'lord': 13, 'rings': 13, 'historical': 39, 'jeff': 10, 'daniels': 3, 'midsummer': 1, 'nights': 6, 'dream': 4, 'romatic': 2, 'johnny': 45, 'depp': 40, 'christopher': 42, 'plummer': 2, 'win': 9, 'selling': 4, 'novel': 4, 'vampires': 14, 'bay': 19, 'psychiatric': 1, 'hospitals': 1, 'pretty': 23, 'pink': 7, '1930s': 6, 'scifi': 3, 'villain': 5, 'wins': 2, 'adventure': 150, 'rolling': 2, 'stones': 2, 'lone': 2, 'ranger': 3, 'moulin': 3, 'rouge': 3, 'awards': 10, 'silent': 9, 'harpo': 1, 'marx': 1, 'scream': 6, 'skating': 2, 'roger': 12, 'mpaa': 32, 'henry': 10, 'thomas': 11, 'quentin': 14, 'tarentino': 2, 'directors': 5, 'only': 11, 'psychological': 28, 'stanly': 1, 'kubrick': 21, '1971': 2, 'ed': 4, 'norton': 12, 'recently': 14, 'babies': 2, 'around': 47, 'presley': 3, 'ma': 4, 'collection': 1, 'van': 13, 'dame': 1, 'lead': 39, 'animal': 7, 'jmes': 1, 'alfred': 25, 'suspense': 19, 'series': 18, 'spiderman': 5, 'moviews': 1, '12': 4, 'gave': 9, 'cruise': 25, 'biggs': 5, 'favorite': 14, 'things': 13, 'hilary': 2, 'duff': 1, 'tell': 88, '2004': 8, 'ad': 1, 'placement': 1, 'arnold': 14, 'schwarzenegger': 11, 'say': 9, 'mw': 1, 'nice': 1, 'enrique': 1, 'toy': 11, 'story': 31, 'robbery': 10, 'now': 14, 'director': 154, 'kung': 8, 'fu': 8, 'ashley': 4, 'olsen': 3, 'hanksmeg': 1, 'ryan': 35, 'cillian': 2, 'murphy': 16, 'role': 58, 'jonah': 7, 'hill': 10, 'suspensful': 1, 'mansell': 1, 'done': 36, 'voice': 36, 'shrek': 20, 'cant': 11, 'wait': 2, 'king': 26, 'characters': 18, '2001s': 1, 'oceans': 5, '11': 2, 'spike': 17, 'vin': 5, 'diesel': 3, 'involve': 7, 'packed': 1, 'should': 5, 'humor': 7, 'too': 5, 'mathematicians': 1, 'three': 101, 'sylvester': 7, 'stallone': 7, 'dancers': 6, 'house': 28, 'sand': 8, 'fog': 4, 'actress': 19, 'kristen': 7, 'would': 97, 'marvel': 4, 'comics': 1, 'through': 1, 'eyes': 4, 'tomorrow': 1, 'never': 4, 'dies': 1, '2003': 1, 'wesley': 5, 'snipes': 3, 'swayze': 5, 'cinderella': 2, 'judi': 5, 'dench': 5, 'adventures': 5, 'tintin': 4, 'mermaids': 1, 'whos': 6, 'as': 127, 'chick': 8, 'flick': 49, 'lawrence': 13, 'turner': 7, 'pirates': 21, 'caribbean': 11, 'explosions': 3, 'robots': 5, 'elise': 1, 'clifton': 2, 'ward': 4, 'mockumentary': 127, 'guest': 2, 'mojave': 1, 'mon': 1, 'his': 25, 'fish': 9, 'says': 12, 'keep': 1, 'swimming': 1, '2': 32, 'years': 378, '1997': 7, 'jacksons': 2, 'mash': 9, 'clown': 4, 'posse': 2, 'mariah': 4, 'carey': 7, 'act': 29, 'jamie': 11, 'curtis': 7, 'decode': 1, 'rosie': 4, 'odonnell': 4, 'duchovnys': 1, 'edward': 18, 'jon': 11, 'bon': 2, 'jovi': 2, 'appeared': 13, 'parker': 9, 'superhero': 4, 'ghost': 7, 'rider': 5, 'la': 6, 'dolce': 2, 'vita': 2, 'wrote': 10, 'avatar': 10, 'interview': 4, 'vampire': 18, 'elizabeth': 15, 'taylor': 18, 'national': 1, 'velvet': 3, 'trying': 16, 'take': 17, 'over': 23, 'breeding': 1, 'humans': 3, 'created': 7, 'soundtrak': 1, 'nominated': 7, 'but': 19, 'havent': 1, 'going': 12, 'sin': 6, 'city': 23, 'joely': 1, 'richardson': 3, 'sharks': 6, 'lowest': 4, 'bad': 8, 'guy': 10, 'tron': 2, 'children': 42, 'ratatouille': 1, 'hunger': 17, 'games': 19, 'preview': 23, 'argonauts': 1, 'stuff': 2, 'dreams': 3, 'tim': 39, 'burtons': 1, 'newest': 15, 'ther': 1, 'gene': 24, 'wilder': 20, 'army': 16, 'darkness': 8, 'kind': 10, 'valentines': 4, 'sports': 22, '99': 1, 'basketball': 5, 'teams': 1, 'marisa': 4, 'tomei': 4, 'hattie': 1, 'mcdaniel': 1, 'avengers': 2, 'kirk': 6, 'douglas': 11, 'adam': 24, 'sandler': 21, 'horrors': 2, 'joe': 21, 'pesci': 12, 'craven': 3, 'breakfast': 1, 'club': 4, 'halle': 9, 'berry': 8, 'obi': 1, 'wan': 1, 'kenobi': 1, 'roman': 15, 'holiday': 2, 'every': 6, 'serenity': 2, 'ages': 1, 'jaws': 11, 'bruc': 1, 'ni': 1, 'battle': 11, 'algiers': 1, 'cabin': 2, 'disaster': 33, 'sandra': 20, 'bullock': 14, 'pullman': 3, 'brendan': 3, 'fehr': 1, 'its': 28, 'highly': 148, 'recommended': 55, 'summer': 7, 'allison': 1, 'hannigan': 5, 'band': 6, 'camp': 11, 'connelly': 2, 'means': 2, 'pirate': 8, 'muir': 1, 'happy': 5, 'feet': 4, 'trek': 15, '10': 12, 'friendly': 4, 'scariest': 4, 'halloween': 6, '1982': 2, 'robin': 27, 'scores': 2, 'grace': 2, 'pixar': 7, '2005': 6, 'lindsay': 3, 'lohan': 5, 'boyscout': 1, 'big': 12, 'kid': 16, 'casey': 1, 'affleck': 4, 'predator': 1, 'malcolm': 4, 'reynolds': 9, 'smurfs': 1, 'ten': 103, 'sellers': 3, 'burger': 1, 'robot': 16, 'adult': 3, '1975': 5, 'female': 13, 'leads': 3, 'wrestler': 2, 'carrie': 4, 'alone': 6, 'wall': 61, 'street': 8, 'financial': 2, 'collapse': 1, 'joss': 2, 'whedon': 2, 'make': 51, 'another': 8, '90s': 34, 'concentration': 3, 'survivor': 1, 'buddha': 1, 'contains': 3, 'bigger': 4, 'boat': 5, 'clooneys': 2, 'oldest': 1, 'ossie': 1, 'davis': 18, 'nicholas': 9, 'sparks': 2, 'jolies': 2, 'changeling': 1, 'mighty': 3, 'true': 16, 'jackie': 10, 'chan': 12, 'chris': 21, 'tucker': 7, 'religion': 4, 'stalker': 1, 'eric': 10, 'idle': 2, 'tyler': 11, 'perry': 8, 'goldfinger': 3, 'zombie': 14, 'hills': 4, 'beverly': 2, 'cop': 9, 'word': 11, 'laser': 2, 'doubt': 1, 'hulk': 5, '70s': 4, '5': 10, 'cousin': 3, 'vinney': 1, 'evil': 23, 'twins': 4, '24': 3, 'dinosaur': 1, 'daze': 1, 'safe': 3, 'noir': 103, 'crime': 115, 'anatomy': 1, 'murder': 26, 'nighy': 1, 'moonrise': 1, 'kingdom': 2, 'plays': 31, '500': 1, 'days': 13, 'spies': 2, 'noire': 1, '1986': 1, 'mystery': 121, 'third': 10, 'chim': 1, 'chimeney': 1, 'bette': 13, 'midler': 6, 'charlies': 1, 'angeles': 5, 'nicole': 14, 'kidman': 10, 'ashton': 4, 'kutcher': 4, 'napoleon': 2, 'dynamite': 1, 'anything': 5, 'leprechaun': 2, 'positive': 1, 'header': 1, 'directing': 9, 'alex': 7, 'haleys': 1, 'x': 11, 'academy': 12, 'polot': 1, 'tokyo': 1, 'france': 2, '60s': 4, 'classic': 18, 'stayin': 1, 'alive': 13, 'lethal': 2, 'weapon': 3, 'voiced': 4, 'donkey': 4, 'snowmen': 1, 'lastest': 2, 'gosling': 5, 'wash': 2, 'theaters': 10, 'currently': 5, 'travel': 3, 'isnt': 2, 'superheroes': 4, 'incredibles': 1, 'shakespearean': 1, 'branagh': 3, 'emma': 7, 'thompson': 13, 'andromeda': 1, 'strain': 1, 'josh': 8, 'duhamel': 2, 'fans': 2, 'christina': 3, 'applegate': 2, 'evening': 2, 'news': 2, 'dead': 39, '1980': 168, 'gone': 10, 'wind': 10, 'else': 4, 'blackhawk': 1, 'down': 8, 'happens': 4, 'he': 22, 'nightmare': 8, 'christmas': 17, 'hellow': 1, 'goodfellas': 10, 'celine': 1, 'dions': 1, 'really': 125, 'classical': 4, 'musicians': 6, 'boxers': 1, 'run': 38, 'super': 5, 'hero': 8, 'review': 26, 'alvin': 3, 'chipmunks': 3, 'die': 8, 'hitler': 4, 'post': 7, 'apocalyptic': 4, 'genesis': 1, 'rodriguez': 2, 'ferrell': 12, 'kirsten': 4, 'dunst': 3, 'hard': 10, 'also': 21, 'shoot': 2, 'em': 2, '21': 4, 'jump': 5, 'critically': 57, 'acclaimed': 56, 'divorce': 3, 'margaret': 1, 'thatcher': 3, 'babe': 3, 'ruth': 1, 'nutrition': 1, 'healthy': 1, 'eating': 4, 'red': 6, 'tails': 2, 'scenes': 5, 'viewers': 34, 'terry': 9, 'gilliam': 4, 'detachment': 1, 'flash': 1, '1967': 2, 'more': 26, '3': 22, 'haunted': 14, 'shark': 4, 'norman': 4, 'reedus': 1, 'mr': 17, 'mrs': 2, 'harrison': 22, 'ford': 52, 'shelby': 1, 'steel': 6, 'magnolias': 3, 'panther': 6, 'glenn': 5, 'close': 5, 'inspirational': 1, 'football': 5, 'principle': 2, 'shooting': 1, 'locations': 1, 'fistful': 1, 'dollars': 1, 'feature': 19, 'mafia': 4, 'goes': 4, 'audiences': 9, 'jane': 15, 'eyre': 2, 'carpenter': 3, 'urban': 4, 'cowboy': 14, 'pauly': 2, 'shore': 2, 'panda': 2, '1995': 10, 'puss': 2, 'n': 3, 'boots': 2, 'seven': 198, 'samurai': 5, 'amish': 1, 'beals': 1, 'dancer': 1, 'american': 23, 'reunion': 1, 'matrix': 15, 'hartnett': 1, '200s': 1, 'joseph': 12, 'gordon': 6, 'levitt': 3, 'hammer': 2, 'nemo': 5, 'finding': 5, 'spider': 9, 'max': 1, 'candace': 1, 'bergen': 3, 'minnie': 1, 'driver': 8, 'trilogy': 4, 'mirrens': 1, 'biographies': 2, 'crazy': 4, 'horse': 9, 'liz': 3, 'wayans': 3, 'castle': 3, 'sky': 4, 'vow': 4, 'macgoohan': 2, 'weird': 4, 'motion': 3, 'guns': 3, 'witches': 6, 'destroy': 1, 'cowboys': 2, 'polticial': 1, 'oliver': 17, 'stone': 19, 'voices': 4, 'gingerbread': 1, 'legal': 2, '80s': 15, 'barry': 8, 'lyndon': 2, 'metro': 1, 'goldwyn': 1, 'mayer': 2, 'become': 2, 'studio': 3, 'pie': 4, 'sparrow': 5, '1930': 3, 'average': 219, 'stare': 2, 'goats': 1, 'zombies': 9, 'lorax': 2, 'doom': 5, 'booth': 2, 'am': 75, 'maya': 2, 'rudolph': 2, 'beowulf': 1, 'drive': 11, 'killer': 21, 'birds': 5, 'casino': 3, 'royale': 1, 'aaliyah': 1, 'soundtracks': 8, 'switchfoot': 1, 'kate': 30, 'beckinsale': 4, 'rush': 5, 'hour': 4, 'eilera': 1, 'ray': 4, 'liota': 2, 'wiz': 1, 'gillian': 3, 'teenage': 5, 'wasteland': 1, 'ash': 1, 'angry': 2, 'seeking': 9, 'justice': 4, 'kirstie': 2, 'alley': 2, 'travolta': 20, 'spanish': 3, 'pattinson': 3, 'spirited': 1, 'away': 4, 'video': 1, 'denzel': 13, 'washington': 13, 'truman': 1, 'blackstreets': 1, 'times': 3, 'hamm': 1, 'groundhog': 1, 'valor': 1, 'playing': 8, 'neo': 3, 'midlers': 1, 'sta': 1, 'dash': 2, 'theremin': 1, 'docudrama': 7, 'iraq': 1, 'telepathic': 1, 'relationship': 8, 'human': 3, 'meshell': 1, 'dialog': 8, 'espiranto': 1, 'under': 2, 'sea': 5, 'kenan': 1, 'fantasy': 207, 'europe': 1, 'diesal': 1, 'road': 9, 'trips': 1, 'licensed': 1, 'bridesmaids': 2, 'ball': 3, 'apart': 1, 'surfer': 7, 'lestat': 1, 'planet': 10, 'space': 30, 'jerry': 12, 'seinfeld': 2, 'boys': 8, 'mcg': 1, 'lithgow': 4, 'detective': 5, 'casa': 1, 'de': 27, 'mi': 1, 'padre': 1, 'runaway': 7, 'jury': 2, 'hood': 6, 'nurses': 2, 'baker': 1, 'special': 2, 'effects': 2, 'lees': 1, 'lois': 1, 'maxwells': 1, 'vantage': 2, 'point': 6, 'japan': 3, 'walken': 8, 'white': 17, 'graduate': 1, 'scanners': 1, 'dee': 6, 'werewolves': 3, 'fight': 7, 'loving': 2, 'screenplay': 3, 'dark': 31, 'shadows': 2, 'walking': 1, 'tall': 1, 'heres': 1, 'penguin': 2, 'returns': 2, 'gia': 1, 'hunting': 2, 'elephants': 2, 'horrific': 1, 'thieves': 4, 'teachers': 1, 'facing': 1, 'adversity': 2, 'warwick': 2, 'heath': 8, 'ledger': 7, 'themed': 5, 'beast': 3, 'bob': 21, 'hoskins': 1, 'cooley': 1, 'riddler': 1, 'riding': 3, 'determinate': 1, 'book': 18, 'benjamin': 2, 'bratt': 1, 'campbell': 10, 'credited': 1, 'writer': 4, 'male': 4, 'abraham': 1, 'lincoln': 1, 'santa': 6, 'clause': 4, 'waht': 2, 'shaq': 2, 'shutter': 1, 'island': 7, 'nosferatu': 2, 'lion': 9, 'mona': 1, 'carvell': 1, 'maggie': 4, 'meet': 5, 'halfway': 1, 'politicians': 1, 'gus': 4, 'zant': 1, 'genres': 2, 'besides': 4, 'salt': 1, 'legolas': 1, 'franco': 4, 'theres': 3, 'no': 4, 'place': 20, 'lex': 2, 'boreanaz': 1, 'renee': 11, 'zellweger': 4, 'flicks': 8, 'forest': 3, 'fires': 1, 'said': 19, 'kings': 2, 'speech': 1, 'molly': 3, 'ringwold': 1, 'viewer': 7, '1972': 1, 'after': 12, 'skaters': 1, 'reese': 5, 'witherspoon': 4, 'large': 1, 'scale': 2, 'natural': 3, 'disasters': 1, 'casablanca': 3, 'snl': 1, 'madden': 1, 'sergio': 11, 'leone': 11, 'philip': 4, 'seymour': 8, 'hoffman': 13, 'transvestite': 1, 'mulholland': 1, 'justin': 8, 'timberlake': 6, 'jodie': 11, 'foster': 12, 'rio': 2, 'actresses': 4, 'miss': 5, 'moneypenny': 1, 'terminator': 9, 'bio': 4, 'pic': 3, 'e': 65, 'happened': 1, 'eleven': 3, 'cher': 3, 'fp': 1, 'greenaway': 1, 'aerosmith': 1, '1947': 1, 'dany': 1, 'boyles': 1, 'burton': 26, 'underworld': 6, 'streetcare': 1, 'desire': 4, 'stalag': 1, 'age': 10, 'corrupt': 1, 'policeman': 2, 'celebrity': 1, 'inc': 1, 'industry': 5, 'faye': 4, 'dunaway': 4, 'appears': 1, 'blues': 2, 'm': 12, 'quantum': 1, 'solace': 1, 'harrelson': 3, 'werewolf': 2, 'london': 12, 'during': 35, 'transformation': 1, 'scene': 8, 'channing': 3, 'tatum': 3, 'fred': 11, 'keaunu': 1, 'inner': 1, 'celebration': 1, 'richard': 27, 'gere': 6, 'amount': 2, 'sequels': 7, 'gena': 2, 'district': 1, '9': 3, 'thousand': 2, 'words': 1, '1920s': 2, 'torch': 1, 'singer': 9, 'jim': 32, 'carrey': 22, 'devil': 5, 'inside': 2, 'bottle': 1, 'shock': 1, 'farley': 2, 'critical': 2, 'future': 19, 'whoopi': 3, 'goldberg': 3, 'ip': 1, 'cube': 3, 'shadow': 2, 'fourth': 1, '1945': 1, 'transformers': 5, 'william': 22, 'shatner': 4, 'strada': 2, 'acted': 16, 'psychiatrist': 1, 'enters': 2, 'vincent': 8, 'price': 4, 'boris': 1, 'karloff': 1, 'winslet': 8, 'thats': 23, 'train': 13, 'woman': 22, 'mgm': 3, 'kaye': 1, 'betty': 3, 'yellow': 2, 'labrador': 1, 'control': 3, 'selection': 1, 'premiere': 2, 'jj': 4, 'abrams': 4, 'worth': 3, 'watching': 4, 'claire': 2, 'god': 6, 'bernard': 5, 'herman': 7, 'mewes': 1, 'identity': 15, 'chevy': 8, 'golden': 4, 'pond': 2, 'knight': 9, 'rises': 3, 'beneath': 2, 'wings': 1, 'alan': 16, 'rickman': 4, 'silence': 9, 'lambs': 7, '1988': 5, 'cherry': 1, 'pepsie': 1, 'queen': 8, 'victoria': 4, 'practical': 1, 'magic': 8, 'chasing': 1, 'amy': 8, 'starrted': 1, 'harakiri': 1, 'crown': 1, 'affair': 5, 'prison': 7, '50s': 2, 'quinton': 1, 'tarrantino': 1, 'mia': 1, 'farrow': 1, 'gun': 5, 'elmer': 4, 'bernstein': 3, 'hogan': 1, 'rope': 1, 'speilberg': 2, 'produced': 21, 'han': 1, 'solo': 2, 'co': 10, 'hula': 1, 'they': 13, 'sandlot': 2, 'tina': 5, 'fey': 4, 'screenplays': 1, 'greed': 4, 'wallstreet': 1, '1989': 3, 'cabiria': 1, 'shawshank': 11, 'redemption': 16, 'ana': 1, 'karenina': 1, 'performed': 2, 'emelio': 1, 'estevez': 3, 'hockey': 5, 'mel': 40, 'gibson': 20, 'least': 25, 'box': 8, 'office': 7, 'upon': 4, 'bacon': 5, 'off': 11, 'novels': 2, 'inspector': 4, 'clouseau': 1, 'mars': 4, 'needs': 1, 'moms': 1, 'grossing': 12, 'prequel': 3, 'roots': 1, 'pandas': 1, 'type': 17, 'rosemarys': 1, 'renoylds': 1, 'critic': 4, 'ebert': 1, 'shortbus': 2, 'adams': 8, 'enchanted': 1, '2007': 1, 'replacements': 2, 'sigourney': 2, 'weaver': 4, 'franchise': 2, 'skellington': 1, '2006': 3, 'directs': 1, 'notebook': 2, 'tourist': 1, 'linkin': 2, 'park': 6, 'leading': 6, 'actoractress': 1, 'wwii': 7, 'prince': 7, 'hell': 7, 'anymore': 2, 'mcgoohan': 1, 'diane': 3, 'pltf': 1, 'nemesis': 1, 'connery': 17, 'hope': 4, 'cuttlefish': 1, 'benny': 3, 'period': 8, 'pieces': 3, 'medieval': 2, 'superbad': 2, '1941': 2, 'rosebud': 3, 'blazing': 4, 'saddles': 5, 'supporting': 6, 'kermit': 2, 'muppets': 2, 'gael': 1, 'garcia': 4, 'bernal': 1, 'balloon': 1, 'giants': 1, 'francis': 18, 'hostel': 2, 'whose': 2, 'mouse': 4, 'staring': 32, 'j': 19, 'edgar': 4, 'hackman': 12, 'without': 6, 'gardening': 1, 'patterson': 1, 'handle': 2, 'truth': 3, 'humphrey': 13, 'bogart': 14, 'airplane': 11, 'austin': 9, 'powers': 10, 'feel': 4, 'buster': 4, 'higest': 1, 'know': 62, 'loves': 1, 'audry': 1, 'hepburn': 25, 'thr': 3, 'sixth': 2, 'zed': 1, 'lockhart': 1, 'pre': 1, '911': 3, 'trade': 1, 'center': 1, 'covered': 1, 'web': 1, 'waynes': 2, 'grit': 5, 'include': 4, 'incredible': 1, 'cry': 1, 'sister': 2, 'dillon': 4, 'drug': 4, 'use': 3, 'merican': 1, 'jessica': 10, 'alba': 3, 'marty': 4, 'mcfly': 2, '1977': 1, 'cast': 17, 'ivory': 3, 'everyone': 3, 'malcovich': 3, 'larenz': 2, 'tates': 1, 'chracter': 1, 'menace': 1, 'society': 1, 'pacimp': 1, 'houston': 6, 'luck': 7, 'lady': 5, 'comes': 8, 'bullit': 1, 'whay': 2, 'protocol': 1, 'droid': 1, 'tornadoes': 1, 'jeremy': 11, 'renner': 2, 'columbiana': 1, 'oscars': 8, 'extremely': 1, 'loud': 1, 'incredibly': 1, 'blow': 7, 'gaslight': 1, 'old': 12, '40': 1, 'ago': 5, 'league': 6, 'extraordinary': 6, 'gentlemen': 4, 'congeniality': 1, 'second': 6, 'carribean': 5, 'hudsons': 1, 'ranking': 1, 'd': 10, 'w': 1, 'griffith': 3, 'epic': 6, 'him': 4, 'offer': 3, 'refuse': 3, 'doc': 6, 'professor': 2, 'snapes': 1, 'executive': 1, 'producer': 2, 'numbers': 1, 'mohicans': 1, 'meg': 8, 'strange': 4, 'land': 3, 'eleased': 1, 'ingmar': 19, 'bergman': 23, 'abu': 1, 'calire': 1, 'danes': 2, 'blockbuster': 6, 'positively': 1, 'pitts': 2, 'barrymores': 2, 'monkees': 1, 'left': 4, 'wright': 3, 'western': 173, 'popularize': 1, 'knights': 2, 'important': 4, 'figure': 2, 'history': 61, 'india': 2, 'fighting': 6, '20s': 1, 'spain': 1, 'battlestar': 1, 'galactica': 1, 'secret': 12, 'ooze': 1, 'motorcycle': 1, 'showed': 3, 'pactick': 1, 'drag': 3, 'dave': 7, 'chappelle': 2, 'weed': 1, 'youll': 2, 'lorre': 3, 'fallon': 1, 'serious': 2, 'hits': 7, 'wonderful': 5, 'thing': 5, 'tigger': 1, 'cartoon': 5, 'scored': 5, 'tupac': 1, 'janet': 2, 'tackles': 1, 'idea': 1, 'discrimination': 1, 'b': 7, 'goonies': 8, 'major': 9, 'serves': 1, 'rate': 15, 'kissing': 2, 'deseree': 1, 'cannibal': 1, 'holocaust': 4, 'pixars': 2, 'stanley': 23, 'techno': 2, 'gets': 6, 'week': 1, 'claus': 1, 'windtalkers': 1, 'humphry': 1, 'zemeckis': 9, 'chocolates': 2, 'computer': 7, 'graphics': 1, 'encounters': 1, '1981': 4, 'jonas': 1, 'fidelity': 1, 'lebowski': 1, 'casinos': 1, 'kyle': 5, 'charlize': 12, 'therons': 1, 'gangs': 3, 'clips': 7, 'winnie': 1, 'pooh': 1, 'tarantino': 12, 'arthur': 10, 'chocolat': 1, 'giant': 7, 'cock': 1, 'blocking': 1, 'jam': 7, 'composer': 3, 'amelie': 1, 'friends': 16, 'buzz': 12, 'lightyear': 12, 'charley': 1, 'chaplin': 18, 'women': 7, 'german': 3, 'burt': 7, 'goobers': 1, 'titles': 36, 'nation': 2, 'lampoon': 1, 'depicts': 1, 'earth': 13, 'being': 24, 'destroyed': 1, 'asteroid': 1, 'officers': 1, 'capitalism': 1, 'communism': 1, 'ww': 2, 'helped': 1, 'guote': 1, 'odyssey': 8, 'lean': 16, 'agent': 7, 'crackers': 1, 'soup': 2, 'change': 4, 'habit': 1, 'johhny': 2, 'alice': 3, 'cooper': 8, 'viewed': 1, 'frankie': 2, 'avalon': 2, 'beach': 7, 'scrooged': 1, 'extramarital': 1, 'affairs': 2, 'sound': 4, 'nothing': 4, 'tv': 6, 'prostitution': 3, 'bodyguard': 2, 'johanssons': 1, 'roles': 8, 'presleymovie': 1, 'las': 4, 'vegas': 5, 'charlotte': 1, 'hall': 5, 'wallis': 1, 'madison': 4, 'dance': 6, 'composed': 2, 'arabia': 2, 'ariel': 1, 'fonda': 17, 'tonight': 6, 'matthew': 31, 'broderick': 8, 'romances': 2, 'saving': 4, 'private': 6, 'ronald': 2, 'regan': 1, 'omar': 2, 'gooding': 3, 'ratings': 190, 'above': 54, 'bays': 1, 'gaston': 1, 'nimoy': 5, 'gran': 1, 'torino': 1, 'once': 2, 'december': 2, 'biblical': 2, 'charleton': 1, 'heston': 10, 'colors': 1, 'feed': 1, 'search': 17, 'fireworks': 2, 'ricky': 5, 'gervais': 1, 'pair': 1, 'patriot': 1, 'pow': 2, 'camps': 2, 'americans': 2, 'spacey': 8, 'half': 8, 'baked': 5, 'et': 7, 'cat': 7, 'hit': 9, 'mission': 7, 'impossible': 6, 'gantry': 1, 'inception': 72, 'much': 7, 'money': 9, 'earn': 1, 'sow': 1, 'rainbow': 8, 'blacksploitation': 3, 'voodoo': 1, 'saw': 7, 'jingle': 1, 'way': 3, 'singlteon': 1, 'parody': 8, 'frankenweenie': 1, 'raja': 1, 'court': 1, 'vinny': 2, 'charlton': 10, 'hestons': 1, 'final': 4, 'died': 1, 'fancy': 1, 'gown': 1, 'blitz': 1, 'japanese': 4, 'mushrooms': 1, 'mcgregor': 2, 'six': 171, 'goodbar': 1, 'skyscraper': 1, 'dad': 2, 'ahead': 4, 'marmalade': 1, 'synopsis': 2, 'comic': 6, 'africa': 5, 'soundrack': 1, 'ferris': 4, 'buelers': 4, 'mivie': 1, 'antagonist': 1, 'invisible': 2, 'cate': 5, 'blanchett': 5, 'les': 3, 'starts': 6, 'titans': 1, 'romero': 1, 'happening': 1, 'reading': 1, 'theron': 11, 'moranis': 2, 'balls': 4, 'read': 1, 'eastbound': 1, 'threme': 1, 'futureovies': 1, 'march': 2, 'roddy': 5, 'mcdowel': 4, 'isaaks': 1, 'grisham': 1, 'spencer': 13, 'tracy': 16, '1965': 2, 'creature': 4, 'gollum': 1, 'metaphor': 1, 'environmental': 1, 'causes': 1, 'commander': 2, 'data': 1, 'folk': 1, 'bands': 2, 'extra': 4, 'terrestrial': 4, 'previews': 2, 'sandlers': 1, 'zoolander': 1, 'heavyweights': 1, 'pan': 2, 'episode': 3, 'paul': 42, 'newman': 15, 'jazz': 2, 'viwers': 1, 'why': 4, 'married': 4, 'spy': 8, 'obsessed': 1, 'sings': 4, 'blowers': 1, 'daughter': 5, 'field': 5, 'sharon': 3, 'meat': 2, 'gross': 7, 'ultimatum': 2, 'cranky': 1, 'teething': 1, 'wich': 1, 'forever': 1, 'salsa': 1, 'weekend': 3, 'bernies': 1, 'pop': 2, 'culture': 1, 'rutger': 1, 'hauer': 1, 'darrel': 2, 'hanna': 3, 'brooks': 21, 'o': 2, 'brother': 2, 'thou': 1, 'sub': 1, 'titled': 24, 'samuel': 12, 'l': 16, 'frost': 2, 'witness': 1, 'precognitive': 1, 'along': 2, 'wha': 1, 'karate': 1, 'pipi': 2, 'longstocking': 2, 'mean': 3, 'girls': 7, 'blade': 3, 'runner': 2, 'flaming': 1, 'computers': 2, 'sant': 1, 'val': 6, 'kilmer': 6, 'romeros': 1, 'direcorial': 1, 'debut': 2, 'person': 5, 'few': 2, 'disneypixar': 1, '2009': 8, 'realesed': 1, 'katy': 3, 'liquid': 1, 'terminatior': 1, 'ai': 1, 'artificial': 3, 'intelligence': 6, 'rebel': 6, 'against': 1, 'golf': 3, '1985': 4, 'watched': 3, 'benefits': 4, 'razzie': 2, 'depps': 2, 'piece': 4, 'frankly': 2, 'dear': 2, 'dont': 9, 'damn': 3, 'acedemy': 1, 'nominees': 1, 'often': 2, 'shown': 5, 'wa': 1, 'legend': 5, 'katherine': 7, '2pac': 1, 'shakur': 1, 'whar': 1, 'doing': 1, 'appearance': 4, 'commando': 1, 'valkyrie': 1, 'meyers': 2, 'fincher': 14, '2008': 7, 'rotten': 2, 'tomatoes': 2, 'ant': 1, 'bully': 3, 'yul': 2, 'brenner': 2, 'death': 29, 'forrester': 2, 'boorman': 4, 'unforgiven': 2, 'biker': 2, 'marlboro': 1, 'rod': 3, 'serling': 1, 'leslie': 5, 'gore': 7, 'sing': 5, 'sunshine': 1, 'lolipops': 1, 'rainbows': 1, '1958': 1, 're': 2, 'written': 2, 'wiig': 1, 'fletcher': 1, 'conspiracy': 1, 'theory': 1, 'rhett': 1, 'butler': 4, 'boyle': 5, 'twist': 3, 'shout': 1, 'version': 8, 'popeye': 2, 'schindlers': 1, 'colin': 6, 'higgins': 1, 'dolly': 2, 'parton': 2, 'hal': 1, 'plant': 1, 'tirantino': 1, 'supernatural': 4, 'hakuna': 1, 'matata': 1, 'destination': 2, 'olivia': 4, 'newton': 5, 'titantic': 1, 'players': 2, 'eisenhower': 1, 'power': 12, 'outer': 3, 'limits': 2, 'episodes': 1, 'included': 5, 'kidmans': 2, 'neilson': 1, 'godfellas': 1, 'howards': 1, 'production': 4, 'partner': 1, 'micky': 1, 'rooney': 1, 'own': 2, 'arc': 2, 'daniel': 22, 'jordan': 4, 'p': 4, 'tv13': 1, 'darby': 1, 'shaw': 1, 'annette': 5, 'blanket': 1, 'bingo': 1, 'sabrina': 3, 'blood': 11, 'honey': 2, 'psycho': 2, 'babysitter': 3, 'wife': 7, 'south': 7, 'african': 5, 'cage': 11, 'hollywood': 4, 'redford': 9, 'mcqueen': 13, 'poor': 1, 'west': 7, 'side': 8, 'bieber': 1, 'native': 2, 'acorns': 1, 'trainer': 1, 'working': 5, 'girl': 23, '1939': 1, 'cancer': 3, 'washingtons': 1, 'trouble': 1, 'considering': 2, 'frightening': 1, 'golfing': 1, 'tubthumping': 1, 'chased': 1, 'bear': 5, 'wood': 10, 'sheriff': 1, 'brody': 2, 'nolan': 15, 'rival': 1, 'magician': 1, 'broke': 3, 'records': 2, '1000': 2, 'corpes': 1, 'doris': 2, 'sharp': 1, 'shooter': 1, 'aragorn': 1, 'queens': 1, 'stared': 6, 'kiddman': 1, 'alices': 1, 'restaurant': 1, 'flying': 3, 'repeated': 1, 'scarface': 4, 'gangstas': 1, 'paradise': 1, 'wit': 1, 'ustinov': 3, 'pat': 1, 'benetar': 1, 'q': 1, 'rap': 1, 'mike': 12, 'tyson': 4, 'tiger': 3, 'gallo': 5, 'elton': 1, 'chocolate': 7, 'factory': 9, 'gilmore': 1, 'shakespeare': 4, 'denmark': 1, 'lamp': 1, 'nc17': 1, 'wname': 1, 'collaborates': 1, 'iv': 2, 'gods': 1, 'framed': 2, 'rabbit': 3, 'met': 4, 'sally': 2, 'crystals': 1, 'young': 11, 'frankenstein': 6, 'mental': 5, 'institution': 3, 'pacifist': 1, 'historic': 3, 'raiders': 5, 'ark': 3, 'statham': 1, 'hired': 3, 'balboa': 1, 'actual': 1, 'cult': 3, 'classics': 2, 'cloris': 1, 'leachman': 1, 'oif': 1, 'segel': 2, 'growing': 4, 'seth': 8, 'rogens': 1, 'pineapple': 1, 'express': 1, 'banned': 1, 'ship': 6, 'wwi': 1, 'middle': 4, 'east': 2, 'kelly': 15, 'debbie': 3, 'mall': 1, 'butterfinger': 1, 'nautical': 3, 'huston': 12, 'heavy': 3, 'metal': 3, 'tribute': 3, 'achilles': 1, 'endless': 1, '1940': 127, '1970': 162, 'ellen': 2, 'roark': 1, 'bars': 1, 'oldman': 6, 'rachel': 3, 'mcadams': 2, 'her': 9, 'allie': 1, 'hamilton': 8, 'britanny': 1, 'thin': 2, 'hasta': 1, 'vista': 1, 'start': 3, 'today': 3, 'prostitute': 3, 'godzilla': 3, 'pool': 1, 'monty': 5, 'pythons': 1, 'circus': 3, 'korea': 1, 'luther': 2, 'voight': 1, 'geres': 1, 'spoken': 1, 'orson': 22, 'welles': 22, 'sidney': 1, 'poitier': 1, 'fox': 13, 'v': 2, 'ringwald': 2, 'owning': 1, 'ponies': 1, 'record': 1, 'label': 1, 'rim': 1, 'english': 2, 'speaking': 1, 'salma': 2, 'hayek': 2, 'gleason': 2, 'bale': 15, 'troopers': 1, 'racing': 6, 'col': 1, 'oneill': 2, 'stargate': 1, 'continuum': 1, 'forgetting': 2, 'sarah': 8, 'britan': 1, 'myers': 3, 'moldy': 1, 'peaches': 1, 'parents': 1, 'blind': 5, 'dracula': 5, 'apolo': 1, 'multiplicity': 1, 'dealing': 11, 'disease': 1, 'roll': 1, 'scout': 1, 'diamonds': 1, 'cats': 3, 'showcase': 1, 'acting': 6, 'talent': 1, 'pierce': 7, 'brosnan': 2, 'bedazzled': 1, 'watson': 6, 'circle': 1, 'perrys': 1, 'screen': 1, 'isaac': 3, 'biopic': 2, 'blading': 1, 'moves': 4, 'kong': 2, 'gorrilla': 1, 'h': 5, 'macy': 3, 'earned': 4, 'firth': 3, 'katniss': 2, 'success': 2, 'independent': 31, 'neil': 1, 'harris': 4, 'travelling': 2, 'astronaut': 1, 'peace': 1, 'driving': 2, 'daisy': 1, 'hercules': 2, 'york': 10, 'undertaker': 2, 'buried': 3, 'similiar': 1, 'youve': 1, 'mail': 1, 'air': 5, 'force': 2, 'pulp': 1, 'twelve': 1, 'mainstream': 2, 'sick': 2, 'killers': 5, 'saying': 8, 'shot': 7, 'stardom': 1, 'containing': 3, 'koreas': 1, 'musketeers': 2, 'theplot': 1, 'while': 1, 'sleeping': 1, 'lassie': 1, 'lucille': 2, 'provide': 15, 'cleese': 5, 'somg': 1, 'notting': 2, 'farmer': 1, 'corn': 1, 'boy': 11, 'saves': 2, 'gabor': 2, 'piper': 1, 'arent': 2, 'violent': 7, 'liberace': 1, 'titular': 1, 'track': 3, 'focus': 7, 'concept': 2, 'seeing': 4, 'ground': 1, 'leaves': 2, 'then': 3, 'animation': 70, 'braveheart': 4, 'stein': 1, 'wish': 5, 'dogfight': 1, 'dragon': 3, 'tatoo': 1, 'mob': 8, 'live': 7, 'banjo': 1, 'belong': 1, 'airport': 1, 'cloning': 1, 'morgan': 12, 'freeman': 11, 'mila': 2, 'kunis': 2, 'johnson': 7, 'dicaprios': 2, 'pregnant': 1, 'teenager': 2, '007': 1, 'moviies': 1, 'perfrmance': 1, 'stuart': 7, 'apes': 2, 'mary': 7, 'reilly': 1, 'mustangs': 1, 'portmans': 1, 'robbie': 1, 'skywalker': 1, 'waterfront': 1, 'sometime': 2, '1994': 3, 'simba': 1, 'potte': 1, 'mrovies': 1, 'gonna': 2, 'fly': 4, 'zimmer': 2, 'dinasour': 1, 'fron': 1, 'cushing': 2, 'eddie': 9, 'murphys': 2, 'paris': 3, 'right': 48, 'caddyshack': 2, 'acts': 1, 'sword': 2, 'sorcery': 2, 'madonna': 4, 'went': 3, 'undercover': 5, 'beauty': 10, 'pageant': 1, 'emmy': 2, 'rossum': 2, 'bulma': 1, 'navy': 3, 'pilot': 2, 'warden': 1, 'lola': 1, 'swahili': 1, 'gangster': 22, 'wonka': 8, 'choclate': 1, 'witch': 14, 'wizard': 7, 'oz': 4, 'jean': 11, 'dujardin': 1, 'translation': 1, 'rihanna': 1, 'current': 1, 'ends': 2, 'ester': 1, 'seabiscuit': 1, 'possible': 3, 'fun': 4, 'foxs': 1, 'berrys': 1, 'hungry': 1, 'again': 3, 'originate': 1, 'christo': 1, 'son': 3, 'kills': 2, 'mother': 5, 'leigh': 21, 'anne': 5, 'tuohy': 1, 'extreme': 5, 'violence': 17, 'work': 5, 'hughes': 3, '1984': 2, 'cg': 1, 'presleyovie': 1, 'kubricks': 1, 'johannsen': 1, 'gloria': 4, 'grahame': 2, '1960': 169, 'dolphins': 2, 'frog': 2, 'shirley': 8, 'earliest': 1, 'johnathan': 1, 'tayler': 1, '28': 2, 'wonderland': 3, 'denver': 2, 'don': 9, 'knotts': 3, 'rollerball': 1, 'pony': 1, 'entirely': 2, 'furry': 2, 'monster': 8, 'worked': 5, 'green': 7, 'eyed': 2, 'gang': 5, 'warriors': 2, 'avildsen': 1, 'karat': 1, 'night': 29, 'voldemort': 1, 'bythe': 2, 'danner': 2, '1979': 4, 'forrest': 2, 'gump': 1, 'maids': 1, 'greek': 1, 'mythology': 2, '1950': 162, 'guys': 2, 'escaped': 1, 'chain': 2, 'supermans': 1, 'bakeds': 1, 'if': 17, 'so': 7, 'ones': 1, 'redd': 2, 'babysitters': 2, 'issaiah': 1, 'kurt': 5, 'russell': 15, 'ritchies': 1, 'pets': 2, 'glen': 2, 'russia': 1, 'socery': 1, 'bugs': 2, 'regarded': 1, '1943': 1, 'spaghetti': 41, 'galahad': 1, 'videodrome': 1, 'pictures': 4, 'bollywood': 3, 'prominent': 3, 'panama': 1, 'halen': 1, 'airheads': 1, 'serial': 4, 'hughs': 1, 'mannequin': 1, 'experiments': 5, 'barber': 1, 'writtin': 1, 'rogan': 1, 'ringo': 1, 'starr': 1, 'prometheus': 2, 'aside': 2, 'student': 1, 'winfield': 1, 'sottish': 1, 'deep': 4, 'lauren': 5, 'bacall': 3, 'shogun': 1, 'sherlock': 4, 'holmes': 7, 'joins': 1, 'carnival': 1, 'sam': 14, 'worthington': 1, 'penguins': 1, 'ted': 5, 'lawyer': 1, 'hidden': 2, 'psychic': 3, 'abducts': 1, 'child': 25, 'rko': 1, 'ginger': 4, 'rogers': 6, 'ralph': 9, 'latifah': 3, 'italy': 3, 'veteran': 1, 'dinosaurs': 5, 'empire': 2, 'craig': 6, 'kindergarten': 1, 'diver': 2, 'saga': 2, 'eclipse': 2, 'doug': 3, 'glatt': 1, 'schooner': 1, 'sailboat': 1, 'unexpected': 1, 'journey': 8, '14th': 1, 'tony': 9, 'randal': 2, 'jetsons': 1, 'jet': 5, 'li': 4, 'opposite': 5, 'tights': 1, 'goldie': 4, 'hawn': 6, 'surfing': 1, 'laura': 6, 'linney': 1, 'kicking': 1, 'screaming': 2, 'hot': 3, 'gorgo': 1, '300': 4, 'rodney': 6, 'dangerfield': 3, '1968': 1, 'solved': 1, 'thier': 1, 'issues': 5, 'season': 3, 'polynesian': 1, 'bettany': 1, 'charles': 22, 'darwin': 1, 'gogh': 1, 'silberling': 1, '1974': 1, 'sid': 1, 'krofft': 1, 'lmovie': 1, 'mason': 2, 'boxer': 4, 'bamed': 1, 'flynn': 3, 'geoffrey': 2, 'marquis': 1, 'sade': 1, 'punk': 1, 'eddy': 6, 'kathy': 4, 'bates': 3, 'youth': 2, 'farce': 3, 'alli': 1, 'magraw': 1, 'addiction': 2, 'claude': 3, 'damme': 2, 'almighty': 1, 'location': 1, 'cyd': 1, 'charisse': 1, 'danced': 1, 'deborah': 1, 'foreman': 1, 'bulge': 1, 'fresno': 1, 'california': 1, 'webb': 2, 'october': 2, 'diazs': 2, 'civil': 9, 'supremacy': 1, 'freida': 1, 'mock': 1, 'jericho': 2, 'wanna': 1, 'cannonball': 1, 'thelma': 1, 'louise': 1, 'opera': 6, 'bernie': 3, 'mac': 3, 'bath': 1, 'tub': 2, 'nims': 1, 'smiths': 2, 'con': 3, 'becomes': 1, 'commodities': 1, 'trader': 1, 'parsons': 1, 'plots': 5, 'doctor': 4, 'nyc': 1, 'kill': 10, 'members': 2, 'katie': 2, 'mistaken': 2, 'scorcese': 1, 'blanc': 2, 'kennedy': 2, 'marooned': 2, 'committed': 1, 'tommy': 10, 'lights': 3, 'call': 3, '1900s': 1, 'stood': 1, 'still': 3, 'treasure': 5, 'pioneers': 2, 'fisher': 2, 'soldiers': 7, 'jolson': 1, 'littlest': 1, 'workers': 1, 'lily': 1, 'tomlin': 1, 'gambling': 1, 'marvin': 2, 'throws': 1, 'boiling': 1, 'coffee': 1, 'grahames': 1, 'face': 2, 'soul': 3, 'gandalf': 2, 'niro': 4, 'verbinski': 3, 'viggo': 5, 'mortensen': 5, 'fasten': 1, 'seatbelts': 1, 'ides': 1, 'crawford': 5, 'governor': 1, 'louisiana': 1, 'costners': 1, 'contacts': 1, 'trailerfor': 1, 'ewoks': 1, 'hays': 1, 'escape': 5, 'turkish': 1, 'kis': 1, 'meryly': 1, 'lawyers': 2, 'each': 1, 'pageants': 1, 'merchant': 1, 'antony': 2, 'zelweger': 1, 'lindsey': 2, 'feels': 1, 'gangsta': 1, 'cagney': 12, 'quaids': 1, 'keaches': 1, 'carradines': 1, 'jesse': 5, 'racial': 2, 'differences': 1, 'soon': 9, 'penelope': 2, 'cruzs': 1, 'language': 2, 'filmography': 1, 'akira': 8, 'kurusawas': 1, 'fools': 1, 'magicians': 1, '571': 1, 'mexican': 3, 'powered': 2, 'individual': 2, 'million': 3, 'dollar': 2, 'corpse': 1, 'kubrik': 1, 'barbara': 8, 'stanwyck': 4, 'mcmurray': 1, 'snider': 1, 'strangland': 1, 'belushi': 5, 'cameos': 1, 'freddie': 1, 'sonny': 1, 'steele': 2, 'citizen': 3, 'kane': 3, 'chinese': 2, 'crenna': 1, 'salesman': 2, 'swayzes': 1, 'apocalypse': 4, 'olaf': 1, 'asian': 1, 'soccer': 5, 'pryor': 6, 'anti': 1, 'erasing': 2, 'memory': 7, 'catherine': 8, 'zeta': 5, 'andersons': 1, 'window': 3, 'monkeys': 2, 'richie': 2, 'rich': 2, 'return': 4, 'jedi': 2, 'categories': 8, 'tarot': 1, 'card': 1, 'reader': 1, 'apocalpse': 1, 'swordsman': 1, 'party': 10, 'werner': 2, 'herzog': 2, 'threatened': 1, 'dimples': 1, 'provided': 1, 'himself': 1, 'kruger': 1, 'snape': 1, 'francos': 1, 'dumb': 1, 'dumber': 1, 'lightning': 2, 'found': 6, 'believe': 2, '1944': 1, 'born': 2, 'connick': 1, 'jr': 22, 'doll': 4, 'chuckie': 1, 'artist': 1, 'colonel': 1, 'fosse': 1, 'dowop': 1, 'veiwers': 1, 'ozarks': 1, 'emil': 1, 'renny': 1, 'overboard': 1, 'winter': 4, '68th': 1, 'rose': 4, 'korean': 1, 'punks': 1, 'hugh': 8, 'project': 3, 'better': 7, 'hangover': 2, 'upcoming': 5, 'lizard': 2, 'elephant': 1, 'kline': 3, 'helena': 5, 'bonham': 4, 'carter': 11, 'ancient': 1, 'indians': 4, 'mexico': 5, 'latin': 1, 'revolutions': 2, 'goldgerg': 1, 'hitchcocks': 1, 'french': 6, 'captain': 6, 'picard': 1, 'cheerleaders': 2, 'dan': 12, 'akyrod': 1, 'marlon': 12, 'brando': 8, 'we': 5, 'consider': 2, 'relied': 1, 'kindness': 1, 'strangers': 1, 'herbie': 2, 'bananas': 1, 'duvall': 9, 'country': 3, 'wrestling': 3, 'lightening': 1, 'rainmaker': 1, 'bongo': 1, 'mark': 22, 'wahlberg': 2, 'mccall': 1, 'different': 2, 'voorhees': 1, 'ex': 1, 'nba': 1, 'martins': 2, 'madam': 1, 'cullen': 4, 'moss': 1, 'dalai': 1, 'lama': 1, 'sloth': 1, 'brooke': 2, 'shields': 2, 'orleans': 1, 'prostitutes': 1, 'ladies': 2, 'oprah': 1, 'winfrey': 1, 'henson': 2, 'recommend': 30, 'australian': 1, 'pearce': 1, 'bears': 2, 'strips': 1, 'nude': 2, 'sjywalker': 1, 'lemmon': 14, 'avanti': 1, 'bag': 2, 'lions': 2, 'audrey': 9, 'nun': 1, 'belgian': 1, 'congo': 1, 'complete': 2, 'told': 2, 'kidnapped': 3, 'caine': 10, 'ireland': 1, 'atleast': 1, 'dreaming': 1, 'living': 7, 'kenny': 2, 'loggins': 1, 'akyroyd': 1, 'englends': 1, 'elm': 1, 'st': 2, 'college': 5, 'spiral': 2, 'staircase': 1, 'dorothy': 2, 'mcguire': 1, 'amnesia': 1, 'sufferer': 1, 'babylon': 2, 'adwhat': 1, 'tattooed': 1, 'diesels': 1, 'hand': 1, 'inceptioncategories': 1, 'england': 2, 'those': 1, 'brasher': 1, 'doubloon': 1, 'merlin': 2, 'kim': 5, 'basinger': 2, 'speilburg': 1, 'fortress': 2, 'ginnifer': 1, 'goodwin': 1, 'purple': 9, 'rain': 12, 'joyful': 2, 'holden': 2, 'joan': 12, 'bennett': 1, 'angela': 4, 'lansbury': 1, 'alaska': 2, 'gwyenth': 1, 'paltrow': 2, 'radio': 2, 'lounge': 1, 'annie': 1, 'racer': 1, 'whisperer': 1, 'louis': 3, 'malle': 2, 'lana': 1, 'european': 1, 'thornton': 2, 'alyson': 3, 'kidnapping': 1, 'fist': 2, 'nesbitt': 1, 'leave': 1, 'law': 2, 'tamilyn': 1, 'tomita': 1, 'heralded': 1, 'adapted': 1, 'wild': 12, 'burgess': 1, 'rabbits': 2, 'glow': 1, 'condom': 1, 'short': 49, 'megan': 3, 'actionsci': 2, 'infamous': 2, 'thorton': 2, 'ferrel': 2, 'c': 9, 'riely': 1, 'fat': 1, 'liotta': 2, 'amp': 4, 'hayak': 2, 'thunder': 1, 'superheros': 1, 'between': 6, 'battlefield': 3, 'help': 23, 'fellow': 1, 'tragedy': 4, 'sinking': 1, 'calamity': 1, 'boyer': 1, 'tries': 1, 'convince': 1, 'ingrid': 4, 'wyatt': 2, 'earp': 2, 'holliday': 2, 'fraser': 2, 'navarone': 2, 'whole': 6, 'rodrick': 1, 'estelle': 1, 'getty': 1, 'newly': 1, 'murderer': 3, 'united': 2, 'states': 1, 'fluffy': 1, 'deanna': 1, 'durbin': 1, 'francois': 7, 'truffaut': 7, 'scientist': 2, 'klingons': 1, 'fights': 2, 'oddjob': 1, 'seen': 19, 'stapler': 1, 'case': 3, 'mondays': 1, 'snakes': 3, 'welcome': 2, 'carmen': 4, 'someone': 18, 'dears': 1, 'neve': 1, 'trey': 1, 'sleepless': 1, 'seattle': 1, 'holds': 1, 'takes': 8, 'hostages': 1, 'bud': 4, 'receiver': 1, 'lupus': 1, 'skin': 2, 'revolutionary': 2, 'flight': 1, 'navigator': 1, 'bane': 1, 'cruses': 1, 'sisters': 1, 'lina': 1, 'wertmuller': 1, 'candice': 2, 'pig': 2, 'demons': 6, 'flesh': 5, 'fishes': 1, 'dori': 1, 'deneuve': 2, 'penny': 2, 'sees': 3, 'fountain': 1, 'suceessful': 1, 'unlaterally': 1, 'deliverance': 1, 'holly': 1, 'hunter': 9, 'kidnap': 1, 'uma': 2, 'thurman': 2, 'hayes': 2, 'regina': 2, 'speedway': 2, 'malcom': 3, 'radcliffe': 1, 'dances': 2, 'snake': 1, 'hudson': 5, 'race': 4, 'cash': 1, 'gwyneth': 1, 'receives': 1, 'former': 1, 'beetlejuice': 5, 'stop': 1, 'believin': 1, 'deneuvre': 1, 'mom': 2, 'wjat': 1, 'isla': 1, 'lang': 9, 'fishers': 1, 'soap': 1, 'elisabeth': 4, 'shue': 4, 'mclaine': 1, 'marriage': 5, 'jude': 1, 'kelley': 1, 'submarines': 1, 'buddy': 3, 'zanzibar': 1, 'snatch': 2, 'gennie': 1, 'pigs': 1, 'sequelsprequels': 1, 'marathon': 1, 'deal': 2, 'childhood': 2, 'memories': 1, 'jungle': 5, 'soldier': 2, 'vietnam': 3, 'montana': 1, 'pacinos': 1, 'newer': 1, 'backward': 1, 'aging': 1, 'father': 5, 'emily': 1, 'blunt': 1, 'wears': 1, 'prada': 1, 'gotta': 1, 'friend': 7, 'received': 211, 'scotts': 1, 'elliott': 3, 'gould': 2, 'teller': 1, 'raging': 2, 'bull': 2, 'shyamalan': 2, 'order': 6, 'gerald': 1, 'step': 1, 'roller': 2, 'derby': 1, 'aids': 2, 'hiv': 1, 'rosario': 1, 'dawson': 3, 'pearl': 2, 'bailey': 1, '1987': 1, 'chess': 1, 'match': 1, 'joy': 2, 'schwarzeneggers': 1, 'katharine': 10, 'cape': 1, 'fear': 11, 'iron': 2, 'catchphrase': 1, 'enjoy': 2, 'later': 4, 'elijah': 3, 'turns': 1, 'into': 7, 'chariots': 3, 'fire': 6, 'christie': 2, 'michelle': 6, 'trilogies': 2, 'angie': 1, 'dickinson': 1, 'speak': 1, 'inmvented': 1, 'depression': 3, 'listall': 1, 'ducks': 2, 'ledgers': 1, 'paranorman': 1, 'jenny': 3, 'mccarthy': 3, 'swank': 1, 'methods': 1, 'game': 3, 'python': 2, 'holy': 5, 'grail': 3, 'kiss': 10, 'ruin': 1, 'heigel': 1, 'clark': 4, 'gable': 3, 'reporter': 1, 'claudette': 1, 'colbert': 1, 'heiress': 1, 'jiminy': 1, 'cricket': 1, 'lancaster': 3, 'espianoge': 1, 'husband': 3, 'michigan': 1, 'hereafter': 1, 'orwells': 1, 'liam': 11, 'neeson': 10, 'slap': 1, 'mimi': 2, 'duchovney': 1, 'lynch': 14, 'vivien': 14, 'raiser': 2, 'cosby': 4, 'roadhouse': 1, 'ride': 2, 'surf': 1, 'jonze': 1, 'beyond': 3, 'unfortunate': 1, 'events': 1, 'crab': 2, 'grodin': 1, 'whom': 1, 'nigel': 1, 'suspicion': 1, 'highschool': 1, 'dennis': 4, 'hopper': 2, 'canadian': 1, 'leo': 1, 'gregory': 14, 'brian': 23, 'broadway': 4, 'shows': 1, 'dwayne': 1, 'nazis': 2, 'cornfield': 1, 'owl': 1, 'rocket': 1, 'waiting': 1, 'tables': 1, 'june': 2, 'laughton': 1, 'tyrone': 2, 'trial': 2, 'panic': 2, 'room': 3, 'past': 430, 'turtle': 1, 'stripes': 1, 'perlman': 2, 'assasins': 1, 'nerds': 1, 'ken': 3, 'pioneered': 1, 'cgi': 3, 'techniques': 1, 'secretariat': 1, 'cure': 1, 'ballet': 1, 'costner': 5, 'dystopian': 1, 'bomb': 2, 'longest': 1, 'yard': 1, 'decaprio': 1, 'controlling': 1, 'franki': 1, 'lane': 3, 'pet': 2, 'versions': 1, 'carol': 3, 'miracle': 1, 'juno': 1, 'ghostbusters': 2, 'angels': 5, 'duel': 1, 'burns': 3, 'danson': 1, 'sammy': 1, 'rhythm': 1, 'lifes': 1, 'worker': 1, 'coppolas': 1, 'pendleton': 1, 'heaven': 4, 'butterfly': 2, 'effect': 1, 'ws': 1, 'noyce': 1, 'moneyball': 1, 'nudity': 1, 'andy': 8, 'perfect': 3, 'storm': 2, 'bite': 1, 'racehorse': 1, 'bruckheimer': 1, 'houses': 1, 'avatars': 1, 'differ': 1, 'hasselhoff': 1, 'mutiny': 2, 'footloose': 2, 'coaches': 1, 'championship': 1, 'cross': 1, 'dressing': 1, 'masked': 1, 'minute': 1, 'lucy': 3, 'liu': 1, 'linda': 5, 'ronstadt': 1, 'here': 4, 'reynold': 1, 'plimpton': 1, 'alda': 3, 'collette': 1, 'greg': 8, 'kinnear': 1, 'contest': 1, 'scorseses': 2, 'arkin': 1, 'terrorizes': 1, 'gigi': 1, 'scottish': 1, 'schoolteacher': 1, 'veronica': 2, 'lake': 5, 'nicolas': 6, 'smashing': 1, 'pumpkins': 1, 'jake': 2, 'gyllanhall': 1, 'airlines': 1, 'plane': 4, 'brue': 1, 'walter': 5, 'mathau': 1, '3000': 1, 'kaiser': 1, 'soze': 1, 'marijuana': 1, 'ghandi': 1, 'animates': 1, 'rights': 3, 'crossdressing': 2, 'diggstown': 1, 'scanner': 1, 'darkly': 1, 'sir': 1, 'eye': 2, 'trolley': 1, 'loaf': 1, 'ovie': 1, 'baskervilles': 1, 'gutenburg': 1, 'eve': 5, 'rib': 1, 'pokemon': 1, 'brians': 1, 'caan': 1, 'omedy': 1, 'toby': 3, 'keith': 8, 'pup': 1, 'leader': 1, 'm2m': 1, 'adaptation': 2, 'jk': 1, 'rowling': 1, 'number': 2, 'rule': 1, 'aviator': 1, 'publisher': 1, 'jameson': 1, 'bird': 5, 'crimson': 1, 'tide': 2, 'caddy': 1, 'crippled': 1, 'hoodwinked': 1, 'utopia': 3, 'spiders': 1, 'biopics': 1, 'rudd': 1, 'cuba': 2, 'stealing': 1, 'befriending': 1, 'truck': 1, 'dean': 7, 'willem': 1, 'dafoe': 1, 'chopper': 1, 'frodo': 2, 'bunny': 2, 'descendants': 1, 'tin': 2, 'cup': 1, 'barbra': 1, 'streisand': 1, 'rogen': 1, 'banks': 1, 'bobby': 1, 'cruz': 3, 'era': 2, 'alcoholism': 2, 'everybody': 2, 'wants': 2, 'thanksgiving': 1, 'cooking': 1, 'debra': 8, 'messing': 5, 'mmpa': 1, 'basketnall': 1, 'diaries': 3, 'sequal': 1, 'wildcats': 1, 'hello': 1, 'nielson': 3, 'drum': 1, 'wheedon': 1, 'guess': 1, 'dinner': 1, 'kept': 1, 'fromthe': 1, 'dawn': 4, 'kiefer': 3, 'versus': 4, 'martians': 1, 'contain': 2, 'cruises': 2, 'fantacy': 1, 'cruse': 1, 'apocolypse': 1, 'lambada': 1, 'spinal': 1, 'tap': 1, 'kardashian': 1, 'contribute': 1, 'hop': 1, 'springtime': 1, 'taxi': 5, 'narnia': 1, 'adaption': 1, 'weithorn': 1, 'valet': 1, 'sells': 1, 'secrets': 5, 'eat': 1, 'worlds': 2, 'hookey': 1, 'phantom': 1, 'criticism': 1, 'robbins': 3, 'prisoner': 5, 'gelenhal': 1, 'actually': 2, 'changed': 1, 'bringing': 2, 'huge': 1, 'worms': 1, 'combine': 1, 'rookie': 1, 'liv': 2, 'took': 3, 'faces': 3, 'earnest': 1, 'borgnine': 1, 'landau': 1, 'older': 1, 'bela': 3, 'lugosi': 3, 'jodi': 1, 'forster': 1, 'running': 1, 'brave': 1, 'shower': 1, 'wanderin': 1, 'brown': 4, 'sugar': 1, 'tinman': 1, 'teresa': 1, 'opening': 2, 'bennie': 1, 'joone': 1, 'listen': 1, 'actin': 1, 'seagal': 2, 'mcdowall': 2, 'gay': 1, 'themes': 1, 'subcultures': 1, 'keenan': 1, 'bought': 1, 'wachowski': 2, 'fame': 1, 'bowie': 2, 'childs': 1, 'dakota': 1, 'fanning': 1, 'lots': 1, 'singers': 2, 'f': 4, 'gray': 4, 'chameleon': 1, 'shreck': 1, 'christine': 5, 'captian': 1, 'locked': 1, 'basement': 1, 'dana': 2, 'jfks': 1, 'funeral': 4, 'quazi': 1, 'motto': 1, 'loser': 1, 'photographer': 1, 'zealand': 1, 'mcclain': 2, 'fair': 2, 'weather': 2, 'dailey': 1, 'jaime': 2, 'clarkson': 1, 'beginning': 2, 'brokeback': 1, 'mountain': 3, 'because': 1, 'seasons': 2, 'gossip': 1, 'poehler': 1, 'rates': 1, 'hubbard': 1, 'accused': 1, 'low': 2, 'brick': 2, 'leavitts': 1, 'knowing': 1, 'whistle': 1, 'full': 5, 'yoda': 6, 'anger': 1, 'hate': 1, 'sting': 3, 'mirror': 2, 'bus': 1, 'amber': 4, 'waves': 1, 'cleef': 1, 'fully': 1, 'loaded': 1, 'fir': 1, 'reeve': 1, 'mcclure': 1, 'breath': 1, 'airplanes': 1, 'victor': 3, 'mature': 1, 'nick': 5, 'norah': 1, 'town': 8, 'station': 2, 'zebra': 1, 'crush': 1, 'farrah': 1, 'faucett': 1, 'according': 1, 'coach': 3, 'ozzy': 1, 'osbourn': 1, 'rag': 2, 'square': 1, 'wide': 1, 'shut': 1, 'gi': 1, 'prodigy': 3, 'pianist': 1, 'atlanta': 1, 'soft': 1, 'core': 1, 'porn': 1, 'hotel': 2, 'hurricane': 1, 'forced': 1, 'hide': 1, 'credits': 1, '60': 1, 'seconds': 1, 'za': 2, 'spent': 1, 'long': 3, 'centers': 36, 'zellweiger': 2, '1933': 1, 'naomi': 1, 'watts': 1, 'adrien': 1, 'hes': 3, 'tramp': 1, 'rango': 1, 'sleepy': 1, 'hollow': 1, 'curse': 4, 'having': 6, 'perkins': 3, 'sundance': 1, 'our': 3, 'macaulay': 1, 'culkin': 4, 'audi': 1, 'carel': 1, 'locally': 1, 'theater': 2, 'break': 1, 'departed': 2, 'cocoon': 1, 'group': 3, 'including': 3, 'thor': 2, 'warren': 6, 'beatty': 1, 'rosalind': 1, 'nurse': 3, 'helps': 1, 'polio': 1, 'patients': 1, 'ashore': 1, 'steiger': 1, 'attacking': 1, 'zardoz': 1, '20000': 1, 'leagues': 1, 'stooges': 5, 'courtroom': 1, 'florence': 1, 'machine': 3, 'sountracks': 1, 'zone': 2, 'kristin': 1, 'novak': 1, 'electra': 3, 'shamolan': 1, 'mockingbird': 1, 'peck': 11, 'radar': 1, 'oreilley': 1, 'kathryn': 1, 'bigelow': 2, '82nd': 1, 'll': 1, 'cool': 2, 'hilton': 1, 'victim': 1, 'sissy': 2, 'spacek': 2, 'unnatural': 1, 'brandon': 7, 'belle': 1, 'smart': 1, 'tootsie': 1, 'general': 5, 'furby': 1, 'benji': 1, 'altman': 3, 'fashion': 1, 'scarefest': 1, 'sweden': 1, 'chester': 1, 'teacher': 1, 'teaser': 1, 'prometheis': 1, 'blofeld': 1, 'enough': 5, 'attack': 3, 'ricci': 1, 'winona': 2, 'ryder': 2, 'chers': 1, 'daughters': 1, 'album': 1, 'angelica': 2, 'benning': 1, 'artists': 2, 'aladdin': 1, 'tattoo': 1, 'patricia': 3, 'arquette': 8, 'raspberry': 1, 'corman': 1, 'surreal': 1, 'ending': 2, 'sense': 1, 'futuristic': 1, 'everything': 1, 'frozen': 2, 'total': 1, 'degaul': 1, 'let': 5, 'hoosiers': 1, 'octopus': 1, 'devon': 4, 'sawa': 2, 'orlando': 1, 'bloom': 1, 'berenger': 2, 'russian': 6, 'defector': 1, 'maclachlan': 1, 'desert': 2, 'fritz': 10, 'fahrenheit': 1, 'itunes': 1, 'transgender': 1, 'allyson': 1, 'chart': 1, 'topping': 1, 'landis': 1, 'pitch': 1, 'closer': 1, 'clive': 1, 'suggest': 9, 'cort': 1, 'thinks': 1, 'whalberg': 1, 'miley': 1, 'cyrus': 2, 'erich': 1, 'von': 3, 'zipper': 1, 'martial': 1, 'arts': 1, 'chuck': 1, 'norris': 1, 'marilyn': 5, 'monroe': 6, 'farrell': 1, 'temple': 1, 'elite': 1, 'controversial': 1, 'bruno': 2, 'suspenseful': 1, 'tate': 4, 'reno': 2, 'ratinggenre': 1, 'ricki': 1, 'peper': 1, 'slater': 3, 'pheiffer': 1, 'catharine': 1, 'athletes': 1, 'dreyfuss': 1, 'ebsen': 1, 'january': 1, 'showboat': 1, 'venice': 1, 'assasin': 1, 'mathematics': 1, 'bean': 3, 'stage': 1, 'door': 4, 'bullitt': 1, 'whoopie': 1, 'waters': 2, 'modine': 5, 'ocean': 2, 'hands': 1, 'scissors': 1, 'nascar': 1, 'schneider': 5, 's': 752, 'heigl': 1, 'excellent': 57, 'held': 3, 'captive': 1, 'denise': 1, 'richards': 2, 'brent': 2, 'carver': 1, 'roy': 1, 'scheider': 1, 'lena': 3, 'horne': 1, 'juliana': 2, 'margulies': 2, 'biographical': 28, 'wildlife': 4, 'domineering': 2, 'riches': 3, 'glickenhaus': 1, 'decent': 49, 'spade': 2, 'liked': 134, 'lot': 49, 'miguel': 2, 'arteta': 1, 'fate': 2, 'tia': 4, 'mowry': 2, 'garber': 1, 'safehouse': 1, 'shemp': 2, 'demon': 2, 'body': 6, 'laughable': 3, 'nine': 172, 'decades': 335, 'jonathon': 3, 'silverman': 2, 'cadillac': 2, 'dysfunctional': 2, 'renfro': 4, 'sibling': 6, 'rivalry': 6, 'very': 104, 'harts': 1, 'rebellion': 2, 'bei': 1, 'po': 2, 'clippety': 1, 'clobbered': 1, 'cotto': 1, 'pacquiao': 1, '7': 2, '04': 1, 'chad': 4, 'fees': 4, 'abortion': 2, 'janeane': 1, 'garofolo': 1, 'petrarca': 1, 'portrait': 11, 'within': 91, 'eight': 172, 'watchable': 39, 'finnes': 2, 'fake': 2, 'research': 2, 'decade': 106, 'bridget': 4, 'wagner': 3, 'sport': 36, 'hack': 1, 'unrated': 149, 'experimentation': 1, 'wiebe': 1, 'obach': 2, 'mcneil': 3, 'haunting': 3, 'portraying': 1, 'combat': 3, 'matthews': 1, 'counter': 3, 'terrorism': 1, 'hewitt': 1, 'anchors': 1, 'aweigh': 1, 'ti': 2, 'involved': 19, 'entertaining': 2, 'schiller': 1, 'birthday': 1, 'imaginary': 2, 'interesting': 5, 'samantha': 2, 'becke': 2, 'dianne': 1, 'vassey': 2, 'ok': 70, 'using': 1, 'nathan': 2, 'frankowski': 1, 'ethan': 2, 'embry': 1, 'rakoff': 2, 'abandonment': 2, 'lessin': 1, 'struggle': 5, 'donovan': 1, 'tamera': 1, 'prowse': 2, 'fairies': 2, 'asher': 2, 'brough': 2, 'laurent': 1, 'bouzereau': 1, 'carchietta': 1, 'goldy': 3, 'ira': 2, 'sachs': 1, 'passfield': 1, 'shannen': 2, 'doherty': 2, 'thumbs': 52, 'near': 7, 'experiences': 2, 'capra': 18, 'avant': 30, 'garde': 30, 'mockumentaries': 3, 'theatre': 2, 'audience': 3, 'joachim': 1, 'schroeder': 1, 'tanks': 3, 'rejection': 3, 'younglove': 1, 'occult': 4, 'callum': 3, 'rennie': 3, 'gracen': 3, 'finance': 4, 'discovery': 5, 'missing': 5, 'rent': 3, 'satire': 2, 'melodrama': 23, 'religious': 2, 'cults': 1, 'multiple': 1, 'murders': 4, 'emotional': 14, 'smallwood': 2, 'levitation': 9, 'sheree': 1, 'mediocre': 57, 'bridges': 2, 'novocaine': 1, 'kohnen': 1, 'sorcerer': 3, 'killing': 1, 'bin': 1, 'laden': 1, 'burnette': 1, 'dutcher': 1, 'estes': 5, 'jorma': 1, 'taccone': 1, 'centered': 24, 'tokar': 1, 'available': 18, 'thought': 8, 'cindy': 2, 'lau': 2, 'development': 2, 'vincente': 1, 'minnelli': 3, 'gunfight': 5, 'appignanesi': 1, 'excited': 1, 'stratosphere': 1, '01': 1, 'horner': 1, 'bergin': 3, 'malkovich': 2, 'kopp': 1, 'bring': 1, 'listing': 2, 'farewell': 2, 'callan': 3, 'mulvey': 3, 'pelican': 1, 'hobbs': 1, 'vacation': 3, 'wands': 2, 'wilderness': 7, 'anyone': 4, 'gendreau': 3, 'generally': 2, 'insanity': 3, 'quattrochi': 1, 'timothy': 2, 'dalton': 1, 'lay': 1, 'cia': 4, 'cain': 2, 'sounds': 1, 'sundays': 1, 'tiffanys': 1, 'polanski': 14, 'recall': 2, 'whether': 3, 'derek': 4, 'code': 3, 'tragic': 1, 'form': 7, 'eugene': 1, 'might': 17, 'revolves': 2, 'warrior': 3, 'vulgarity': 2, 'vivica': 2, 'elves': 4, 'given': 25, 'repos': 2, 'paget': 1, 'erased': 3, 'examples': 2, 'manipulation': 2, 'lines': 2, 'megaton': 1, 'earlier': 1, 'delpy': 3, 'adler': 2, 'masterson': 4, 'larue': 1, 'speedman': 3, 'mahiro': 1, 'maeda': 1, 'dating': 2, 'jaeckel': 1, 'kuntz': 1, 'cobb': 3, 'devils': 2, 'moment': 1, 'repossessed': 1, 'officer': 1, 'homeless': 1, 'palmer': 1, 'sorvino': 10, 'tommi': 1, 'lepola': 1, 'juan': 2, 'delancer': 1, 'chaos': 3, 'direction': 2, 'carolina': 1, 'moon': 6, 'able': 9, 'picardo': 4, 'damiano': 1, 'damiani': 1, 'smuggling': 4, 'hyde': 2, 'sidoni': 2, 'describe': 5, 'terror': 6, 'tournament': 1, 'audition': 9, 'kurosawa': 7, 'autry': 5, 'pakula': 1, 'chapple': 2, 'marine': 1, 'averaged': 6, 'andrei': 5, 'tarkovsky': 5, 'andrew': 10, 'cymek': 1, 'jacobs': 1, 'leman': 1, 'bening': 2, 'haywood': 1, 'antoine': 3, 'fuqua': 1, 'marks': 1, 'genius': 1, 'gags': 4, 'benenson': 1, 'peyton': 1, 'contact': 2, 'palma': 9, 'smrz': 2, 'mysterious': 2, 'talkington': 2, 'carlos': 4, 'leon': 5, 'wedge': 1, 'falling': 2, 'category': 2, 'palko': 1, 'mosbacher': 1, 'deidre': 1, 'denis': 1, 'leary': 1, 'jacobi': 3, 'dermot': 3, 'mulroney': 3, 'dolph': 1, 'lundgren': 1, 'duncan': 3, 'mass': 3, 'eli': 1, 'roth': 2, 'federico': 14, 'fellini': 13, 'downey': 10, 'darabont': 10, 'astaire': 4, 'savage': 2, 'homicide': 1, 'hayao': 12, 'miyazaki': 12, 'narrate': 1, 'experience': 5, 'guillermo': 14, 'del': 15, 'toro': 14, 'shumlin': 1, 'opinion': 3, 'weeks': 1, 'couffer': 1, 'government': 4, 'assassin': 3, 'gold': 12, 'garner': 1, 'horan': 1, 'michelangelo': 11, 'antonioni': 11, 'patric': 1, 'lando': 1, 'slowsky': 1, 'wells': 1, 'whitesell': 1, 'lidstrom': 1, 'kenn': 1, 'navarro': 1, 'kirt': 2, 'gunn': 3, 'mario': 2, 'azzopardi': 1, 'investigations': 2, 'marlee': 2, 'matlin': 2, 'suitable': 2, 'rosen': 1, 'melanie': 2, 'biehn': 4, 'caretaker': 1, 'business': 2, 'marnos': 1, 'ziller': 1, 'painter': 4, 'mcgennis': 1, 'phil': 4, 'hartman': 3, 'siegel': 1, 'mcgowan': 2, 'phillipe': 4, 'neill': 4, 'bernhard': 3, 'sara': 1, 'sugarman': 1, 'lately': 9, 'wincer': 1, 'fugitive': 2, 'dorff': 1, 'simmons': 3, 'terrence': 10, 'malick': 9, 'entertain': 1, 'vittoria': 5, 'sica': 5, 'kramer': 1, 'marcy': 1, 'walker': 2, 'open': 4, 'bonnell': 1, 'loane': 1, 'griffiths': 1, 'laugh': 6, 'rodger': 2, 'los': 3, 'piano': 1, 'agree': 1, 'unrequited': 2, 'eden': 2, 'harvey': 3, 'concerning': 1, 'tara': 1, 'judelle': 1, 'behind': 1, 'enemy': 2, 'remote': 1, 'viewings': 1, 'pee': 2, 'wee': 2, 'hironobu': 1, 'sakaguchi': 1, 'dina': 4, 'meyer': 4, 'averaging': 1, 'grauman': 1, 'explorer': 1, 'jonathan': 8, 'glazer': 2, 'agents': 3, 'lili': 5, 'carry': 9, 'megacorporation': 3, 'happen': 8, 'memoirs': 1, 'wags': 1, 'selfishness': 1, 'across': 2, 'davidson': 4, 'assassination': 5, 'yancovic': 2, 'poltergeist': 1, 'prom': 1, 'sonja': 1, 'truckers': 2, 'giraldi': 1, 'underground': 3, 'resistance': 3, 'kris': 2, 'sherwood': 1, 'andre': 1, 'braugher': 1, 'rags': 1, 'meugniot': 1, 'politics': 2, 'nic': 2, 'izzi': 2, 'fischa': 1, 'skeet': 1, 'ulrich': 1, 'albert': 2, 'finney': 2, 'heidi': 1, 'hurt': 2, 'recommendations': 2, 'vaughan': 2, 'vengeance': 4, 'rapaport': 1, 'giancarlo': 1, 'esposito': 1, 'eaten': 6, 'intrusion': 3, 'odonell': 2, 'loretta': 1, 'alper': 1, 'elliot': 1, 'lebovitz': 1, 'doyle': 3, 'emilio': 2, 'blindness': 2, 'notorious': 1, 'scarecrow': 2, 'singles': 1, 'th': 4, 'voyage': 2, 'sinbad': 1, 'trotsky': 1, 'chosen': 2, 'dantes': 2, 'inferno': 2, 'hooker': 1, 'trunk': 1, 't': 10, 'nell': 2, 'shopgirl': 2, 'spotswood': 1, 'hustle': 1, 'toe': 2, 'raymond': 3, 'massey': 2, 'goal': 1, 'prayers': 1, 'staininger': 1, 'kenneth': 3, 'degeneres': 1, 'kunert': 2, 'salisbury': 1, 'cohn': 1, 'randolph': 2, 'mantooth': 2, 'bloody': 2, 'imposter': 2, 'heartache': 1, 'whaling': 1, 'erskine': 1, 'buscemi': 2, 'lisa': 9, 'prinze': 3, 'mistress': 1, 'clements': 1, 'bauchau': 1, 'ninja': 3, 'hopewell': 1, 'sleepwalking': 1, 'brooklyn': 4, 'gabbert': 2, 'shepard': 1, 'gabriel': 5, 'byrne': 2, 'false': 1, 'accusation': 1, 'dougray': 3, 'sight': 10, 'pumpkin': 1, 'karver': 1, 'chamitoff': 1, 'exist': 1, 'akroyd': 3, 'schizophrenia': 2, 'forget': 1, 'dolans': 1, 'eriq': 1, 'lesalle': 1, 'usually': 2, 'excuse': 1, 'connolly': 3, 'inventory': 1, 'famke': 2, 'janssen': 2, 'dealt': 4, 'deportation': 1, 'master': 3, 'kathleen': 2, 'information': 31, 'easy': 3, 'junichi': 1, 'fujisaku': 1, 'malkovic': 1, 'uth': 1, 'satterfield': 1, 'warlock': 2, 'guerrilla': 3, 'warfare': 3, 'atencio': 1, 'spirit': 3, 'dreyfus': 2, 'afghanistan': 1, 'brain': 2, 'smasher': 1, 'duck': 1, 'le': 1, 'samoura': 1, 'shaft': 1, 'static': 1, 'vanessa': 6, 'angel': 3, 'amadeus': 2, 'delirious': 1, 'insanitarium': 1, 'stand': 1, 'bulletproof': 1, 'warm': 1, 'slow': 1, 'moving': 1, 'gregg': 1, 'champion': 1, 'lapica': 1, 'cecil': 1, 'demented': 1, 'selleck': 2, 'reba': 1, 'mcentire': 1, 'moretti': 1, 'rancher': 4, 'amos': 2, 'kollek': 2, 'saulnier': 1, 'helicopter': 2, 'raid': 3, 'hiroyuki': 2, 'kitakubo': 2, 'guiness': 4, 'colm': 3, 'meaney': 2, 'joaquin': 3, 'phoenix': 3, 'theodore': 2, 'witcher': 2, 'pankov': 4, 'mitzi': 1, 'kapture': 1, 'balaban': 2, 'starvation': 1, 'takahiro': 1, 'tanaka': 1, 'mellodrama': 1, 'andrea': 3, 'destruction': 3, 'jaclyn': 1, 'norbit': 1, 'peanuts': 1, 'state': 1, 'property': 1, 'forbidden': 5, 'ocallaghan': 1, 'drake': 3, 'details': 2, 'maniac': 2, 'phoebe': 1, 'eng': 7, 'subs': 7, 'guessing': 1, 'stonerville': 1, 'getting': 4, 'stoned': 1, 'harmony': 1, 'korine': 1, 'elfont': 1, 'kwok': 2, 'alexander': 3, 'rapp': 2, 'britt': 1, 'allcroft': 1, 'cheryl': 1, 'ladd': 1, 'chrispopher': 1, 'corey': 4, 'feldman': 4, 'deforest': 2, 'erik': 2, 'macarthur': 1, 'erika': 3, 'eleniak': 3, 'small': 4, 'hayley': 2, 'mills': 2, 'jackman': 4, 'ian': 7, 'mckellen': 3, 'gedrick': 2, 'kiristine': 2, 'bravery': 5, 'kristy': 1, 'swanson': 1, 'majors': 3, 'lynn': 4, 'shelton': 4, 'hoge': 2, 'melissa': 3, 'gilbert': 2, 'feifer': 1, 'pare': 1, 'phillip': 5, 'recovering': 1, 'alcoholic': 1, 'avary': 1, 'raimi': 3, 'gellar': 2, 'teri': 2, 'polo': 2, 'heroine': 3, 'jokes': 6, 'pearson': 1, 'quality': 1, 'glover': 2, 'anybody': 1, 'yet': 5, 'sobieski': 1, 'parole': 1, 'hearing': 1, 'talisa': 1, 'soto': 1, 'cavalline': 1, 'charisma': 1, 'dacascos': 3, 'criminals': 4, 'breaking': 2, 'rescue': 5, 'nuclear': 2, 'coen': 8, 'abandoned': 4, 'infotainment': 1, 'subject': 2, 'religulous': 1, 'wallace': 1, 'heard': 6, 'meatballs': 2, 'vibes': 1, 'mm': 1, 'upwards': 1, 'swamp': 1, 'flags': 1, 'fathers': 1, 'pete': 1, 'travis': 2, 'costas': 1, 'imprisonment': 4, 'san': 3, 'giacomo': 2, 'vincie': 1, 'heartless': 1, 'barkin': 1, 'karen': 5, 'ridings': 1, 'hurtz': 1, 'due': 3, 'double': 5, 'indemnity': 1, 'documentarys': 3, 'langella': 2, 'gotten': 1, 'revolved': 6, 'campaign': 1, 'may': 10, 'revolver': 1, 'clouds': 1, 'collars': 1, 'burlesque': 1, 'compared': 1, 'shes': 4, 'duty': 1, 'perceived': 1, 'maximum': 1, 'risk': 1, 'hows': 1, 'saech': 1, 'kinear': 1, 'miranda': 1, 'nancy': 4, 'grahn': 2, 'murakami': 1, 'joel': 4, 'lloyd': 6, 'greta': 2, 'scacchi': 1, 'sedative': 2, 'kellie': 2, 'prisoners': 2, 'mine': 1, 'greenwood': 3, 'allan': 1, 'goldstein': 1, 'sands': 1, 'bosses': 1, 'edwards': 6, 'jensen': 2, 'ackles': 2, 'lovers': 2, 'raquel': 1, 'welch': 1, 'making': 1, 'dagenham': 1, 'requesting': 1, 'pam': 2, 'grier': 2, 'seem': 1, 'fishburne': 3, 'outlanders': 1, 'suggestions': 1, 'alienation': 2, 'basket': 1, 'info': 6, 'wicksboro': 1, 'incident': 1, 'moody': 1, 'bummer': 1, 'cavalry': 2, 'kuo': 1, 'ren': 1, 'wu': 2, 'public': 3, 'meins': 1, 'august': 2, 'undergrounds': 2, 'mordum': 1, 'armitage': 2, 'lange': 3, 'loneliness': 2, 'cusak': 2, 'shane': 5, 'hammond': 2, 'niall': 1, 'maccormick': 1, 'georgina': 3, 'riedel': 2, 'dimitry': 1, 'elyashkevich': 1, 'irwin': 1, 'mattson': 1, 'marcil': 2, 'addis': 1, 'wimpenny': 1, 'triumphant': 1, 'fugue': 1, 'key': 2, 'village': 1, 'mccabe': 1, 'miller': 3, 'northern': 1, 'loose': 1, 'spirits': 3, '2081': 1, 'fantastic': 1, 'thirteenth': 1, 'floor': 2, 'brittany': 2, 'snatchers': 1, 'joey': 3, 'mutilation': 2, 'driven': 1, 'masahiko': 1, 'maesawa': 1, 'definitely': 2, 'porretta': 3, 'rie': 1, 'rasmussen': 1, 'wolotzky': 2, 'franklin': 1, 'guerrero': 1, 'avery': 2, 'asner': 1, 'wonderment': 1, 'rowland': 3, 'involves': 7, 'confession': 2, 'forgiven': 1, 'lea': 8, 'salonga': 1, 'phillips': 5, 'galactic': 1, 'kendall': 1, 'prairie': 1, 'companion': 1, 'nature': 3, 'wicked': 1, 'vince': 3, 'vieluf': 1, 'leick': 2, 'masciantonio': 1, 'berkley': 1, 'larry': 6, 'peerce': 1, 'lyde': 1, 'dementia': 1, 'tribe': 1, 'levin': 1, 'kenya': 4, 'montgomery': 5, 'clift': 4, 'ann': 4, 'unconventional': 4, 'appreciate': 1, 'theresa': 4, 'randall': 3, 'gangsters': 2, 'beth': 1, 'evans': 4, 'mcdowell': 3, 'donna': 2, 'derrico': 2, 'karl': 1, 'hirsch': 2, 'sherman': 1, 'curly': 2, 'miner': 5, 'warsaw': 1, 'ghetto': 1, 'dudikoff': 2, 'whiteman': 1, 'mandy': 1, 'patinkin': 1, 'foxes': 1, 'omalley': 1, 'trejo': 1, 'morse': 2, 'dunn': 4, 'malevolence': 1, 'dwells': 1, 'wing': 1, 'chaser': 1, 'supposedly': 1, 'unicorn': 1, 'asking': 2, 'inquiring': 1, 'peterson': 3, 'interested': 3, 'immortality': 1, 'vaughn': 3, 'parallel': 2, 'universes': 1, 'balloons': 3, 'rommel': 1, 'lustig': 2, 'imagination': 2, 'pritts': 1, 'danielle': 2, 'fishel': 2, 'mulgrew': 1, 'catholicism': 1, 'osamu': 1, 'dezaki': 1, 'quaid': 1, 'ubaldo': 1, 'ragona': 1, 'basannavar': 1, 'mackenzie': 2, 'alicia': 2, 'silverstone': 2, 'rel': 1, 'filming': 1, 'style': 2, 'passion': 1, 'career': 1, 'mcclellan': 1, 'garett': 2, 'maggart': 2, 'suffering': 3, 'buzzell': 1, 'notarile': 1, 'setting': 2, 'survival': 4, 'mathew': 3, 'lillard': 3, 'julianna': 2, 'lavin': 2, 'specific': 2, 'treachery': 1, 'skelding': 2, 'wise': 4, 'vino': 1, 'salame': 1, 'joshua': 2, 'carroll': 4, 'oconner': 2, 'priestley': 2, 'ziering': 3, 'friendship': 2, 'alvarez': 1, 'peggy': 1, 'whaley': 3, 'durante': 1, 'mastrantonio': 1, 'moira': 1, 'lander': 1, 'marra': 1, 'casper': 1, 'dien': 1, 'naim': 1, 'sematary': 1, 'galaxy': 2, 'kelber': 1, 'warthog': 2, 'kazuo': 1, 'terada': 1, 'sure': 1, 'wen': 1, 'jiang': 1, 'searching': 12, 'buchanan': 1, 'akhurst': 1, 'bounty': 6, 'hunters': 1, 'mann': 2, 'xavier': 2, 'puslowski': 2, 'dome': 1, 'dogville': 1, 'robinson': 2, 'crusoe': 1, 'sharpes': 1, 'stabbing': 2, 'infantolino': 1, 'colman': 1, 'kalin': 2, 'paradox': 1, 'pilgrim': 1, 'englishman': 1, 'brandis': 1, 'disturbed': 1, 'zacharias': 1, 'proyas': 1, 'survivors': 1, 'paolo': 3, 'montalban': 3, 'wondering': 1, 'karaoke': 2, 'isitt': 1, 'entitled': 6, 'electric': 1, 'mist': 1, 'replace': 1, 'clinton': 3, 'choose': 1, 'abbess': 1, 'wizards': 2, 'moran': 1, 'forbes': 1, 'cook': 2, 'yancy': 1, '1800': 1, 'bologna': 1, 'schwimmer': 3, 'summit': 1, 'dustin': 5, 'diamond': 2, 'devitos': 1, 'downtown': 2, 'adrian': 2, 'vitoria': 1, 'steiman': 1, 'pena': 1, 'barbie': 2, 'fabrice': 2, 'du': 2, 'welz': 1, 'schrader': 1, 'leonetti': 1, 'stevan': 1, 'mena': 1, 'deals': 4, 'infantry': 1, 'vohrer': 1, 'deception': 1, 'zaphiratos': 1, 'karbelnikoff': 1, 'brett': 2, 'ricardo': 2, 'montalbon': 2, 'beghe': 3, 'gavin': 1, 'budd': 1, 'emmett': 2, 'alston': 2, 'collins': 2, 'lynda': 2, 'jail': 4, 'betrayal': 2, 'ernest': 1, 'shari': 1, 'lies': 2, 'marciano': 1, 'jada': 3, 'pinkett': 3, 'sollett': 1, 'nazi': 7, 'occupation': 1, 'activism': 4, 'isabella': 3, 'rosselini': 3, 'bryan': 1, 'sivertson': 1, 'hindman': 2, 'spacecrafts': 2, 'marcus': 3, 'raboy': 1, 'memorial': 1, 'ransom': 3, 'kearsley': 1, 'steyermark': 1, 'yasmine': 2, 'bleeth': 2, 'bayne': 1, '18': 2, 'century': 2, 'wilkins': 1, 'caruso': 1, 'tatnya': 1, 'ali': 1, 'walters': 2, 'optimism': 1, 'anarchy': 2, 'brickman': 1, 'chong': 3, 'paddy': 1, 'breathnach': 1, 'parasites': 4, 'awarded': 1, 'tori': 1, 'spelling': 1, 'cory': 1, 'brenden': 1, 'sexton': 1, 'pantiliano': 1, 'stephenson': 1, 'lara': 3, 'fabian': 1, 'joven': 1, 'tan': 1, 'cynthia': 1, 'occupied': 2, 'poland': 2, 'lederman': 1, 'diana': 2, 'riggs': 1, 'illness': 1, 'zane': 2, 'planes': 1, 'water': 2, 'cert': 1, 'mayor': 1, 'these': 2, 'drop': 1, 'gorgeous': 1, 'earthlings': 1, 'edwin': 1, 'mccain': 1, 'grandmas': 1, 'jealousy': 5, 'hoodrats': 1, 'hoodrat': 1, 'slated': 1, 'platoon': 5, 'sorority': 3, 'row': 2, 'nicholsons': 1, 'anywhere': 2, 'rosalie': 1, 'bunch': 2, 'fastest': 1, 'indian': 1, 'obvious': 1, 'marsters': 2, 'morey': 2, 'darjeeling': 1, 'limited': 1, 'necropolis': 1, 'mckenzie': 2, 'astin': 3, 'rydell': 1, 'michell': 2, 'entertainment': 3, 'thir': 1, 'en': 1, 'adults': 1, 'rip': 1, 'circuit': 1, 'remaking': 1, 'gathering': 1, 'eagles': 1, 'check': 1, 'altitude': 1, 'donnell': 1, 'rawlings': 1, 'ashy': 1, 'classy': 1, 'toons': 1, 'hells': 1, 'kitchen': 1, 'television': 1, 'ring': 2, 'shaun': 1, 'sheep': 1, 'leap': 1, 'lambkind': 1, 'siren': 1, 'slaughtered': 1, 'vomit': 1, 'dolls': 2, 'museum': 1, 'even': 1, 'alden': 1, 'redgrave': 1, 'tracie': 4, 'lords': 4, 'somehow': 2, 'ginsburg': 2, 'hooks': 2, 'widen': 1, 'mccaulay': 3, 'tales': 1, 'execution': 6, 'jordanna': 3, 'brewster': 3, 'foley': 1, 'dylan': 2, 'neal': 2, 'flintstones': 1, 'honor': 4, 'mayfield': 1, 'gina': 2, 'gershon': 2, 'heather': 2, 'graham': 1, 'ruzickova': 1, 'bonamy': 1, 'jaye': 2, 'erick': 1, 'dowdle': 1, 'taking': 2, 'stevenson': 2, 'morio': 1, 'asaka': 1, 'invisibility': 3, 'sebastian': 2, 'panneck': 2, 'amitri': 1, 'espejo': 1, 'hartle': 1, 'ned': 1, 'farr': 1, 'market': 3, 'gluck': 1, 'daniela': 2, 'pestova': 2, 'jag': 1, 'mundhra': 1, 'nikki': 2, 'cox': 3, 'vonda': 3, 'shepherd': 4, 'rosson': 1, 'scoggins': 3, 'soderbergh': 6, 'sage': 1, 'date': 3, 'duchovny': 2, 'illegal': 1, 'activity': 1, 'drenner': 1, 'patient': 1, 'disguised': 1, 'mira': 7, 'furlan': 1, 'quod': 2, 'drugs': 2, 'sexuality': 4, 'dandridge': 1, 'contend': 1, 'revolving': 7, 'stan': 4, 'kirsch': 4, 'beltran': 1, 'sticks': 1, 'starrcade': 1, 'ultimate': 1, 'mirrormask': 1, 'preston': 1, 'cruelty': 1, 'fine': 2, 'decision': 2, 'gunpoint': 1, 'focuses': 2, 'torture': 6, 'detectives': 3, 'explores': 1, 'focusing': 2, 'neel': 1, 'tetsuro': 1, 'amino': 1, 'teens': 3, 'societal': 2, 'girly': 1, 'hardship': 1, 'everlasting': 3, 'base': 2, 'bandits': 3, 'hilarious': 1, 'couple': 1, 'twisters': 1, 'snipers': 3, 'neretva': 1, 'shack': 1, 'joneses': 1, '976': 1, 'bangkok': 1, 'dr': 10, 'caparulo': 1, 'cap': 1, 'malibu': 1, 'motocrossed': 1, 'paper': 1, 'woodstock': 1, 'tenten': 1, 'diemens': 1, 'cannabis': 1, 'cured': 1, 'irving': 3, 'lerner': 1, 'revenge': 5, 'tx': 1, 'ado': 2, 'kalangis': 1, 'church': 1, 'gorney': 3, 'baron': 2, 'munchausen': 1, 'sunset': 2, 'briefcase': 1, 'oregon': 1, 'jennie': 1, 'garth': 2, 'amateurs': 1, 'silver': 1, 'haim': 3, 'gero': 1, 'bancroft': 2, 'investigation': 3, 'gosselar': 2, 'roxann': 2, 'browne': 1, 'unhappy': 2, 'sexual': 2, 'fischer': 1, 'content': 1, 'kentis': 1, 'harvest': 3, 'waitt': 1, 'mathilde': 1, 'bittner': 1, 'self': 3, 'saffa': 1, 'fever': 3, 'head': 9, 'plan': 1, 'anticipated': 1, 'jeri': 3, 'sweetheart': 1, 'gasaway': 1, 'cattle': 1, 'marc': 2, 'rocco': 2, 'biel': 2, 'townsend': 3, 'criminal': 1, 'mastermind': 1, 'sherry': 1, 'stringfield': 1, 'dual': 2, 'identities': 1, 'boaz': 1, 'yakin': 1, 'appeal': 1, 'jeffery': 1, 'friedman': 1, 'fickman': 1, 'maddocks': 1, 'freiburger': 1, 'surgery': 1, 'shalmar': 1, 'joke': 3, 'kimberly': 1, 'mcculough': 1, 'merit': 1, 'badge': 1, 'ari': 1, 'taub': 1, 'robins': 1, 'shakes': 1, 'jennings': 1, 'shoehorned': 1, 'subplot': 1, 'hitchhikers': 1, 'guide': 1, 'toys': 3, 'chance': 1, 'dissociative': 1, 'disorder': 1, 'jfk': 2, 'gomez': 1, 'bleckner': 1, 'fighter': 1, 'jarmusch': 1, 'krista': 2, 'morrit': 2, 'fuest': 1, 'julian': 3, 'kemp': 1, 'files': 1, 'hoover': 1, 'levar': 3, 'blackmail': 1, 'donuts': 2, 'carrot': 1, 'keri': 1, 'maid': 1, 'cole': 1, 'scar': 2, 'alison': 1, 'raimund': 1, 'huber': 1, 'shunji': 1, 'iwai': 1, 'transylvania': 1, 'micheal': 1, 'lester': 2, 'calista': 1, 'flockhart': 1, 'mccrudden': 1, 'haggerty': 4, 'lucci': 1, 'turteltaub': 1, 'lozano': 1, 'powell': 2, 'whitworth': 2, 'jan': 4, 'rachman': 1, 'crimes': 4, 'velcrow': 1, 'ripper': 1, 'garrett': 2, 'wang': 2, 'johan': 1, 'grimonprez': 1, 'peopled': 1, 'laughed': 1, 'gian': 2, 'keth': 1, 'szarabajka': 1, 'wolochatiuk': 1, 'huntington': 2, 'seiji': 1, 'chiba': 1, 'stolz': 3, 'comical': 1, 'nibbelink': 1, 'technology': 1, 'delaney': 2, 'manger': 1, 'aidan': 4, 'quinn': 4, 'maurice': 4, 'chair': 1, 'mellisa': 3, 'hart': 3, 'carrere': 2, 'polson': 1, 'josetxo': 1, 'mateo': 1, 'bell': 3, 'ferland': 1, 'bassett': 3, 'dorn': 2, 'storage': 1, 'felicias': 1, 'fusion': 1, 'trip': 2, 'fro': 1, 'tanya': 4, 'cyborg': 1, 'ileana': 2, 'winkler': 2, 'tiffani': 3, 'thiessen': 3, 'byington': 1, 'loss': 1, 'mitchum': 1, 'takashi': 1, 'ishii': 1, 'reconnaissance': 1, 'cafiero': 1, 'broken': 2, 'engagement': 1, 'hagman': 1, 'sugg': 1, 'rowlands': 1, 'herbert': 2, 'coleman': 1, 'gerard': 2, 'depardieu': 2, 'pushed': 1, 'lew': 1, 'attempted': 2, 'milestone': 1, 'relating': 1, 'gino': 1, 'nichele': 1, 'natascha': 1, 'mcelhone': 1, 'jeffrey': 1, 'katzenberg': 1, 'kathie': 1, 'gifford': 1, 'kriv': 1, 'stenders': 1, 'burgi': 1, 'midlife': 1, 'crisis': 1, 'geno': 1, 'mcgahee': 1, 'zach': 1, 'hofmeyr': 1, 'brock': 3, 'basil': 1, 'cloke': 1, 'minutes': 1, 'graffiti': 1, 'tennis': 1, 'pursuit': 2, 'sortie': 1, 'des': 1, 'ateliers': 1, 'vibert': 1, 'bride': 3, 'toms': 1, 'midnight': 1, 'garden': 1, 'valmont': 1, 'melman': 1, 'shanley': 1, 'strike': 2, 'schlesinger': 1, 'lambert': 4, 'lon': 1, 'chaney': 1, 'attempt': 3, 'orphans': 1, 'wrye': 2, 'fernando': 1, 'colunga': 1, 'sher': 1, 'anna': 2, 'galvin': 1, 'einstein': 2, 'incorporate': 1, 'shapeshifting': 3, 'ryans': 1, 'rupert': 3, 'everet': 3, 'poet': 2, 'irish': 2, 'friedkin': 1, 'hamburg': 1, 'winger': 2, 'lebrock': 2, 'margret': 1, 'places': 3, 'basic': 1, 'instinct': 1, 'strong': 1, 'gornick': 2, 'melodramas': 2, 'several': 12, 'jeffs': 1, 'kieran': 1, 'carney': 2, 'twenty': 1, 'reprieve': 1, 'pale': 1, 'gates': 3, 'mcfadden': 2, 'central': 2, 'believers': 1, 'dangerous': 2, 'saturn': 1, 'territories': 1, 'cooler': 1, 'rimshop': 1, 'delbert': 1, 'lawless': 1, 'listed': 1, 'dishonor': 2, 'receiving': 2, 'kazuaki': 1, 'kiriya': 1, 'skill': 1, 'stevens': 2, 'infidelity': 1, 'fraunces': 1, 'premise': 2, 'exciting': 1, 'maniacts': 1, 'chuang': 1, 'mellinda': 2, 'mention': 1, 'haussman': 1, 'michel': 2, 'orion': 1, 'suede': 1, 'ide': 1, 'shootout': 1, 'brokedown': 1, 'palace': 1, 'fiddler': 1, 'roof': 1, 'sarahs': 1, 'choice': 1, 'maria': 3, 'bello': 4, 'favorites': 1, 'pinson': 3, 'cukor': 1, 'woodward': 1, 'debuted': 1, 'frankel': 1, 'osteen': 1, 'lachman': 1, 'heroes': 2, 'crash': 2, 'polito': 1, 'addresses': 1, 'consequences': 2, 'deforestation': 1, 'skirmishes': 1, 'border': 2, 'reiser': 2, 'meehl': 1, 'womens': 1, 'der': 1, 'beek': 1, 'ultimately': 1, 'ikea': 1, 'rise': 1, 'byrd': 1, 'mcdonald': 1, 'archer': 1, 'luner': 2, 'disfigured': 1, 'supper': 1, 'doomsday': 1, 'device': 1, 'directory': 1, 'bam': 1, 'margera': 1, 'global': 3, 'climate': 1, 'issue': 1, 'boase': 1, 'madsen': 2, 'gains': 1, 'guilietta': 1, 'masina': 1, 'haley': 2, 'flatland': 1, 'okay': 1, 'cky': 1, 'k': 1, 'enjoyed': 1, 'retirement': 3, 'sullivan': 1, 'kirby': 1, 'dick': 1, 'example': 1, 'security': 2, 'kell': 1, 'evan': 1, 'detten': 1, 'willard': 2, 'ivan': 2, 'mitov': 1, 'melinda': 1, 'clarke': 1, 'pregnancy': 2, 'jo': 2, 'chandler': 2, 'wilcox': 1, 'ric': 1, 'monte': 1, 'snatching': 1, 'salley': 1, 'dennehy': 3, 'kari': 2, 'wuhrer': 2, 'mccormack': 1, 'marcos': 1, 'efron': 1, 'fina': 1, 'torres': 1, 'duhame': 1, 'jac': 1, 'schaeffer': 1, 'kinney': 1, 'loverboy': 1, 'mnemonic': 1, 'soapdish': 1, 'bradley': 1, 'stockholm': 1, 'syndrome': 1, 'summarize': 2, 'serpent': 1, 'keegan': 1, 'raising': 1, 'arizona': 1, 'panorama': 1, 'hardware': 1, 'cover': 1, 'zerophilia': 1, 'acceptance': 1, 'exit': 1, 'milla': 1, 'jovovich': 1, 'rafelson': 1, 'oh': 1, 'thunderbolt': 1, 'lightfoot': 1, 'sweatshop': 1, 'elektra': 1, 'actions': 1, 'phedon': 1, 'papamichael': 1, 'cess': 1, 'silvera': 1, 'terrorist': 3, 'travelers': 1, 'boondock': 1, 'saints': 1, 'corky': 1, 'romano': 1, 'bunraku': 1, 'jay': 3, 'duplass': 1, 'calvert': 1, 'rigg': 1, 'hybrid': 1, 'unsettling': 1, 'alaimo': 1, 'sackheim': 2, 'cheetah': 1, 'gonzalez': 1, 'paxton': 1, 'tatopoulos': 1, 'killed': 5, 'laughs': 1, 'fugitives': 2, 'lamont': 1, 'starman': 1, 'offered': 7, 'such': 3, 'damato': 2, 'promises': 2, 'assayas': 1, 'shou': 1, 'wanted': 2, 'wanting': 4, 'lasted': 1, 'martyn': 1, 'pick': 1, 'alberta': 1, 'company': 2, 'cassandra': 2, 'alyssa': 2, 'moy': 1, 'job': 5, 'claudia': 3, 'debrah': 1, 'farentino': 1, 'donnie': 1, 'walberg': 1, 'ridgemont': 1, 'geena': 1, 'geri': 2, 'halliwell': 2, 'poole': 1, 'chen': 1, 'madeline': 1, 'stowe': 1, 'follows': 1, 'shanks': 3, 'valen': 1, 'boer': 1, 'instrumental': 1, 'appearing': 1, 'stacey': 1, 'ugly': 1, 'guard': 1, 'fridel': 1, 'oconnor': 4, 'sitch': 2, 'cambodia': 2, 'blaine': 1, 'patton': 2, 'budreau': 1, 'pistol': 1, 'whip': 1, 'leah': 1, 'sturgis': 1, 'lou': 2, 'sims': 1, 'friz': 1, 'freleng': 1, 'outcast': 1, 'food': 2, 'approved': 1, 'meerkat': 1, 'ri': 1, 'chard': 1, 'though': 1, 'kinji': 1, 'fukasaku': 1, 'morrison': 1, 'dominique': 1, 'milano': 2, 'parise': 1, 'corporate': 2, 'ernst': 1, 'gossner': 1, 'doran': 1, 'dale': 3, 'steffanino': 1, 'barnz': 1, 'doucette': 1, 'largely': 1, 'ulmer': 1, 'florian': 1, 'henckel': 1, 'donnersmarck': 1, 'averages': 1, 'murdered': 1, 'marek': 1, 'losey': 1, 'solitary': 2, 'confinement': 2, 'clancy': 1, 'roos': 1, 'germany': 1, 'alphonso': 2, 'tze': 1, 'chun': 1, 'deemed': 3, 'larabe': 2, 'kolton': 1, 'focused': 3, 'kilner': 1, 'norda': 1, 'aronoff': 1, 'seller': 1, 'nia': 1, 'vardalos': 1, 'technically': 1, 'prendergast': 1, 'mazin': 1, 'holechek': 1, 'anjelica': 1, 'russ': 1, 'todd': 4, 'verow': 1, 'alcohol': 4, 'fairuza': 1, 'balk': 1, 'kristine': 1, 'mendes': 1, 'huck': 1, 'botko': 1, 'jalmari': 1, 'helander': 1, 'kevan': 1, 'rage': 2, 'kostas': 1, 'karagiannis': 1, 'genocide': 1, 'nair': 1, 'gurland': 1, 'mohr': 1, 'lance': 3, 'weiler': 1, 'pepper': 2, 'argott': 1, 'standoff': 3, 'toni': 1, 'harman': 1, 'liza': 2, 'int': 2, 'decerchio': 1, 'anita': 1, 'laselva': 1, 'vehicle': 1, 'demme': 1, 'psychopath': 2, 'northam': 1, 'isacsson': 1, 'pays': 1, 'melski': 1, 'accidental': 1, 'alberto': 1, 'cavalcanti': 1, 'embezzlement': 2, 'menell': 1, 'hypocrisy': 1, 'deblois': 1, 'reflection': 1, 'levien': 1, 'pascal': 1, 'franchot': 1, 'mcconaughey': 2, 'fiennes': 1, 'smits': 1, 'patty': 2, 'jenkins': 3, 'scribner': 1, 'revolution': 1, 'bangalter': 1, 'rowell': 1, 'warner': 1, 'spiner': 1, 'till': 2, 'barker': 1, 'axelgaard': 1, 'teddy': 4, 'tylo': 1, 'christensen': 1, 'oliveira': 1, 'ruben': 1, 'preuss': 1, 'brownrigg': 1, 'ritter': 1, 'betz': 1, 'solunga': 2, 'despair': 1, 'sangiuliano': 1, 'natasha': 3, 'pavlovich': 2, 'benard': 3, 'annakin': 1, 'aldas': 1, 'reiners': 1, 'myles': 2, 'fergusons': 2, 'mccarthys': 1, 'biographys': 1, 'thora': 1, 'birch': 2, 'choices': 1, 'hank': 1, 'braxtan': 1, 'jerami': 1, 'asquith': 1, 'gariazzo': 1, 'shyu': 1, 'greenfield': 1, 'whale': 1, 'facts': 1, 'malloy': 1, 'lissa': 1, 'rinna': 1, 'thief': 2, 'henstridge': 1, 'wainwright': 1, 'gabby': 1, 'peters': 1, 'tuukka': 1, 'tiensuu': 1, 'matthau': 1, 'chu': 1, 'amusing': 1, 'annoying': 1, 'roommate': 1, 'simple': 1, 'platt': 1, 'cybill': 1, 'tea': 2, 'leoni': 2, 'afterlife': 1, 'portillo': 2, 'speers': 1, 'binder': 1, 'kent': 1, 'jared': 2, 'leto': 2, 'stern': 2, 'wolfinger': 1, 'sadofsky': 1, 'axel': 1, 'rebecca': 4, 'cammisa': 1, 'sophia': 1, 'loren': 1, 'river': 4, 'doremus': 2, 'rosanna': 2, 'munro': 1, 'blutman': 1, 'desi': 1, 'arnaz': 1, 'cheech': 1, 'marin': 1, 'engstrom': 1, 'tomas': 1, 'sandquist': 1, 'baird': 1, 'wiley': 1, 'fishman': 1, 'classified': 1, 'dover': 1, 'koshashvili': 1, 'henriksen': 1, 'kiersch': 1, 'humber': 1, 'balderstone': 1, 'placid': 1, 'monkey': 1, 'presidents': 1, 'sleepover': 1, 'freebie': 1, 'capote': 1, 'elf': 1, 'hustler': 1, 'pasdar': 1, 'denk': 1, 'motor': 1, 'hurley': 1, 'rentzel': 1, 'thueson': 1, 'squeakquel': 1, 'boogeyman': 1, 'turin': 1, 'webber': 1, 'orgy': 1, 'becoming': 1, 'darklands': 1, 'def': 1, 'diggers': 1, 'dysfunktional': 2, 'electroma': 1, 'gung': 1, 'ho': 1, 'carlsons': 1, 'makin': 1, 'corpses': 1, 'confusing': 1, 'internal': 1, 'jakes': 1, 'corner': 1, 'luster': 1, 'brandos': 1, 'bones': 2, 'blueberry': 1, 'lai': 2, 'prefontaine': 1, 'niros': 1, 'satans': 2, 'helper': 2, 'simpatico': 1, 'figures': 1, 'slingshot': 1, 'someones': 1, 'knocking': 1, 'splintered': 1, 'hawks': 1, 'hounds': 1, 'adele': 1, 'sec': 1, 'grey': 1, 'hire': 1, 'ambush': 1, 'others': 2, 'signal': 1, 'trading': 1, 'trick': 1, 'twin': 1, 'peaks': 1, 'unrivaled': 1, 'wisegal': 1, 'zenon': 1, 'zequel': 1, 'molina': 1, 'maggi': 1, 'sykes': 1, 'tiffany': 1, 'kilbourne': 1, 'raged': 1, 'katsuhiro': 1, 'ohtomo': 1, 'gibney': 1, 'duffy': 1, 'freudenthal': 1, 'concerns': 1, 'barnick': 1, 'ving': 1, 'rhames': 1, 'monika': 1, 'treut': 1, 'injustice': 1, 'crowder': 1, 'february': 1, 'christophe': 1, 'gans': 1, 'conway': 1, 'jessy': 1, 'terrero': 1, 'organized': 2, 'troupe': 1, 'description': 1, 'carvey': 1, 'scolari': 1, 'abilities': 1, 'lien': 1, 'bobbie': 1, 'ellis': 1, 'brutality': 1, 'gayheart': 2, 'boogie': 1, 'woogie': 1, 'laughing': 1, 'twisted': 2, 'tasers': 1, 'isaacs': 1, 'lorentzon': 1, 'darlene': 1, 'vogel': 1, 'rene': 3, 'russo': 4, 'vaungh': 1, 'heady': 2, 'anat': 1, 'seftel': 1, 'polish': 1, 'williamson': 1, 'abuse': 1, 'felitta': 1, 'callahan': 3, 'mining': 3, 'minkoff': 1, 'wincat': 1, 'alcala': 1, 'bellware': 1, 'lieutenant': 1, 'chelsom': 1, 'dom': 1, 'rotheroe': 1, 'champions': 1, 'gale': 1, 'harding': 1, 'regarding': 2, 'hacker': 1, 'mikey': 1, 'hilb': 1, 'strock': 1, 'posey': 1, 'ferguson': 1, 'altieri': 1, 'langer': 1, 'burny': 1, 'mattinson': 1, 'delta': 2, 'burke': 3, 'rowan': 1, 'bednarski': 1, 'stephens': 1, 'cassidy': 3, 'jewish': 2, 'paudge': 1, 'behan': 1, 'yudis': 2, 'jorge': 1, 'solis': 1, 'sweeney': 1, 'sorbo': 1, 'foot': 1, 'wayons': 2, 'stellan': 2, 'olsson': 2, 'cerasolis': 1, 'damian': 1, 'roland': 1, 'emmerich': 1, 'bo': 1, 'zenga': 1, 'dogma': 1, 'headless': 1, 'horseman': 1, 'darwell': 1, 'longis': 1, 'homo': 1, 'erectus': 1, 'rollercoaster': 1, 'mouseketeers': 1, 'breakin': 1, 'everett': 1, 'harmon': 1, 'overton': 1, 'courtship': 1, 'eddies': 1, 'basis': 1, 'believer': 1, 'adventurer': 1, 'narrated': 1, 'monkeybone': 1, 'edge': 1, 'annapolis': 1, 'blackboard': 1, 'teds': 1, 'bullet': 1, 'cherrybomb': 1, 'americas': 1, 'cyber': 1, 'threat': 1, 'divine': 1, 'ya': 2, 'sisterhood': 1, 'dumping': 2, 'gekijouban': 1, 'stay': 1, 'unlimited': 1, 'works': 1, 'gymkata': 1, 'normal': 1, 'newsmakers': 1, 'majestys': 1, 'service': 1, 'silicon': 1, 'valley': 3, 'rancid': 1, 'rockaway': 1, 'sanctimony': 1, 'taboo': 1, 'botany': 1, 'confidant': 1, 'tremors': 1, 'triangle': 2, 'waking': 1, 'wilde': 1, 'marina': 1, 'gavrilova': 1, 'exotic': 1, 'mask': 1, 'trikonis': 1, 'cream': 1, 'coscarellis': 1, 'tapping': 2, 'bairstow': 1, 'garfield': 1, 'gabrielle': 1, 'anwar': 1, 'barely': 1, 'preparatory': 1, 'goldblum': 2, 'rater': 1, 'loop': 1, 'ally': 1, 'beverley': 1, 'mitchell': 2, 'schatzberg': 1, 'ingo': 1, 'rademacher': 1, 'darrell': 1, 'mapson': 1, 'jumbo': 1, 'roadkill': 1, 'bonifacio': 1, 'frakes': 1, 'cocktail': 1, 'assemblage': 1, 'sphere': 1, 'koehler': 1, 'zuccon': 1, 'quadrophenia': 1, 'tides': 4, 'fit': 1, 'uninvited': 1, 'unloved': 1, 'camping': 1, 'dusk': 1, 'destiny': 1, 'hedge': 1, 'spinout': 1, 'roscoe': 1, 'penance': 1, 'metamorphosis': 1, 'tigerland': 1, 'bongwater': 1, 'cow': 1, 'target': 2, 'patti': 1, 'labelle': 1, 'sargent': 1, 'kopple': 1, 'segal': 2, 'resnikoff': 1, 'abell': 1, 'storke': 1, 'tells': 2, 'locklear': 1, 'fairy': 1, 'far': 1, 'elliotte': 1, 'bankruptcy': 1, 'hamid': 1, 'gurkha': 1, 'grief': 1, 'wim': 1, 'wenders': 1, 'dannelly': 1, 'rory': 1, 'mchenry': 1, 'bangs': 1, 'grossman': 1, 'heisler': 1, 'egleson': 1, 'mandelbaum': 1, 'fields': 1, 'zieff': 1, 'demoniacs': 1, 'tile': 1, 'simcha': 1, 'jacobovici': 1, 'create': 2, 'fellinis': 1, 'zeitgeist': 1, 'griffin': 1, 'reflections': 1, 'groomsmen': 1, 'radiation': 1, 'fated': 2, 'hurst': 1, 'judging': 1, 'digital': 1, 'evolution': 2, 'houchins': 1, 'witchcraft': 1, 'renoire': 1, 'settings': 1, 'spader': 1, 'naked': 2, 'lighter': 1, 'lower': 2, 'fahey': 1, 'salli': 1, 'reputations': 1, 'viewing': 1, 'jann': 1, 'butch': 1, 'saul': 1, 'dibb': 1, 'lockheart': 1, 'switzer': 1, 'mccallum': 1, 'gondry': 1, 'hartfield': 1, 'qualifies': 7, 'pitof': 1, 'fairchild': 1, 'stouffer': 1, 'dani': 1, 'menkin': 1, 'colleges': 1, 'bouzaglo': 1, 'sheridan': 1, 'hollis': 1, 'chamberlain': 1, 'lacey': 1, 'chabert': 1, 'thaw': 1, 'brady': 1, 'barrett': 1, 'iren': 1, 'koster': 1, 'pytka': 1, 'ciarn': 1, 'hinds': 1, 'shorts': 1, 'ipson': 1, 'donal': 1, 'mosher': 1, 'batchelor': 1, 'courtois': 1, 'galland': 1, 'brings': 1, 'forth': 1, 'totally': 1, 'paint': 1, 'wagon': 1, 'leitch': 1, 'colleen': 1, 'dewhurst': 1, 'rosenberg': 1, 'kosick': 1, 'tillman': 1, 'valentine': 1, 'pelka': 1, 'coneybeare': 1, 'hillcoat': 1, 'carl': 1, '964': 2, 'pinocchio': 2, 'neighbours': 1, 'cages': 1, 'rudo': 1, 'y': 1, 'cursi': 1, 'shrink': 2, 'musician': 1, 'unwed': 1, 'gunslinger': 1, 'alderton': 1, 'chih': 1, 'leong': 1, 'hallucination': 1, 'signorelli': 1, 'lowe': 1, 'cypher': 1, 'cipolla': 1, 'cicely': 1, 'dammes': 1, 'atwell': 1, 'passenger': 1, '57': 1, 'wad': 1, 'coal': 1, 'rufus': 1, 'sewell': 1, 'rosas': 1, 'terence': 1, 'daw': 1, 'crasher': 1, 'bend': 1, 'cheyenne': 1, 'cruel': 1, 'necessary': 1, 'creation': 1, 'dressed': 1, 'straight': 1, 'magoo': 1, 'magnificent': 2, 'sleep': 2, 'drunks': 1, 'boxleitner': 1, 'whatever': 1, 'wire': 1, 'brotherhood': 1, 'tapes': 1, 'grimble': 1, 'bender': 1, 'enjoyable': 1, 'trandem': 1, 'mate': 1, 'gullivers': 2, 'travels': 2, 'weights': 1, 'kinsey': 1, 'joffe': 1, 'necromentia': 1, 'overnight': 1, 'delivery': 1, 'pounds': 1, 'subtle': 1, 'seduction': 1, 'saved': 1, 'pain': 1, 'proudly': 1, 'hail': 2, 'whaledreamers': 1, 'chucky': 1, 'fruit': 1, 'metropia': 1, 'judgment': 2, 'walls': 1, 'cinema': 2, 'afraid': 1, 'senior': 1, 'gamers': 1, 'dorkness': 1, 'rising': 2, 'billion': 1, 'scooby': 2, 'doo': 3, 'abracadabra': 1, 'heredity': 1, 'valdemar': 1, 'bittersweet': 1, 'backyards': 1, 'fierlinger': 1, 'vadim': 1, 'sheng': 1, 'ding': 1, 'gummo': 1, 'grizzly': 1, 'loch': 1, 'ness': 1, 'lawn': 1, 'principal': 1, 'purchase': 5, 'online': 2, 'sandor': 1, 'rank': 2, 'among': 2, 'nicola': 1, 'bounce': 1, 'showing': 1, 'sunday': 1, 'dagger': 1, 'okuribito': 1, 'ran': 1, 'rear': 2, 'graveyard': 1, 'shift': 1, 'hours': 1, 'pepin': 1, 'markowitz': 1, 'storytelling': 1, 'lovelace': 1, 'conrad': 1, 'rooks': 1, 'centering': 1, 'hutchison': 1, 'palance': 1, 'cutler': 1, 'stacy': 1, 'zinn': 1, 'bakshi': 1, 'dealers': 1, 'fleming': 1, 'aquarium': 1, 'sabato': 1, 'berkelys': 1, 'deniros': 1, 'mcnee': 1, 'joness': 1, 'diary': 1, 'wolf': 1, 'assassins': 1, 'merry': 1, 'bridge': 2, 'camouflaged': 1, 'blackie': 1, 'wheel': 1, 'candyman': 1, 'cleo': 1, 'pimpernel': 1, 'cello': 1, 'aykroyd': 1, 'unplugged': 1, 'ufos': 1, 'bone': 2, 'hat': 1, 'chamber': 1, 'hey': 2, 'esther': 1, 'blueburger': 1, 'player': 1, 'abby': 1, 'kick': 1, 'ass': 1, 'laputa': 1, 'lets': 2, 'lornas': 1, 'neshoba': 1, 'northfork': 1, 'orange': 1, 'county': 1, 'perestroika': 1, 'primal': 1, 'cut': 1, 'dirt': 1, 'seraphim': 1, 'falls': 1, 'silkwood': 1, 'slam': 1, 'bang': 1, 'splatter': 1, 'disco': 1, 'blvd': 1, 'sweet': 1, 'sweetbacks': 1, 'baadasssss': 1, 'bells': 1, 'ipcress': 1, 'file': 1, '1900': 1, 'knew': 1, 'bees': 1, 'sweetest': 1, 'telling': 1, 'unknown': 1, 'wanderers': 1, 'wicker': 1, 'thunderpants': 1, 'mini': 1, 'midkiff': 1, 'environmentalism': 1, 'gallipoli': 1, 'nineteen': 1, 'eighty': 1, 'bricktown': 1, 'flypaper': 1, 'longshots': 1, 'wolves': 1, 'buck': 1, 'expedition': 1, 'invaders': 1, 'tortured': 1, 'irishman': 1, 'zoom': 1, 'cu': 1, 'mama': 1, 'rugrats': 1, 'sitting': 1, 'waterfall': 1, 'almost': 1, 'meth': 1, 'legion': 1, 'macbeth': 1, 'masters': 1, 'universe': 1, 'terrifying': 1, 'girlfriends': 1, 'lunch': 1, 'babes': 1, 'thon': 1, 'tenshi': 1, 'tamago': 1, 'lovely': 2, 'winters': 1, 'curfew': 1, 'disclosure': 1, 'salvation': 1, 'chips': 1, 'lifetime': 1, 'cosmos': 1, 'ace': 1, 'hole': 1, 'candle': 1, 'wine': 1, 'cemetery': 1, 'junction': 1, 'chains': 1, 'worry': 1, 'italian': 1, 'fubar': 1, 'gargoyles': 1, 'conquering': 1, 'changi': 1, 'keeping': 1, 'mum': 1, 'alphabet': 1, 'charge': 1, 'feather': 1, 'souled': 1, 'watermelon': 1, 'chicken': 1, 'paranormal': 1, 'penomena': 1, 'relic': 1, 'matilda': 1, 'miles': 1, 'profondo': 1, 'rosso': 1, 'smash': 1, 'camera': 1, 'grinch': 1, 'batteries': 1, 'wives': 1, 'lefay': 1, 'crossover': 1, 'wax': 1, 'passage': 1, 'marseille': 1, 'jawbreaker': 1, 'highway': 1, 'runnery': 1, 'eagle': 1, 'easel': 1, 'robocop': 1, 'gear': 1, 'solid': 1, 'orchid': 1, 'vicious': 1, 'priest': 1, 'cedric': 2, 'gibbons': 1, 'apartment': 1, 'sorted': 1, 'hutton': 1, 'hardwicke': 1, 'traci': 1, 'bingham': 1, 'movei': 1, 'movement': 2, 'fictional': 1, 'espionage': 1, 'fuminori': 1, 'kizaki': 1, 'trevor': 1, 'saloon': 1, 'thrown': 1, 'situations': 1, 'crossed': 1, 'axe': 1, 'horde': 1, 'withing': 1, 'geraint': 1, 'wyne': 1, 'davies': 1, 'website': 5, 'detailed': 2, 'opinions': 1, 'common': 1, 'already': 1, 'users': 1, 'evita': 5, 'comments': 1, 'boring': 1, 'sentiment': 1, 'site': 1, 'bit': 1, 'awesome': 1, 'glimpse': 1, 'clip': 1, 'extended': 1, 'highlights': 1, 'spoilers': 2, 'snippets': 1, 'cinemas': 1, '97': 2, 'uhhh': 1, 'darth': 6, 'vader': 6, 'chapter': 1, 'muppet': 1, 'isn': 1, 'potato': 8, 'favourite': 1, 'probably': 1, 'quiet': 1, 'front': 1, 'sunrise': 1, 'intolerance': 1, 'schindler': 1, 'north': 1, 'northwest': 1, 'hur': 1, 'mystic': 1, 'paradisio': 1, 'sierra': 1, 'madre': 1, 'heat': 3, 'connection': 1, 'breathless': 1, 'avventura': 1, 'saturday': 1, 'battleship': 1, 'potemkin': 1, 'un': 1, 'chien': 1, 'andalou': 1, 'strawberries': 1, 'ugestu': 1, 'monogatari': 1, 'ambersons': 1, 'zhivago': 1, 'exorcist': 1, 'towers': 1, 'dictator': 1, 'deadly': 1, 'darling': 1, 'clementine': 1, 'yojimbo': 1, 'cabinet': 1, 'caligari': 1, 'discreet': 1, 'charm': 1, 'bourgeoisie': 1, 'peeping': 1, 'marienbad': 1, 'noon': 1, 'wages': 1, 'splendor': 1, 'grass': 1, 'fargo': 1, 'ikiru': 1, 'singin': 2, 'aguirre': 1, ',': 1, 'wrath': 2, 'aparajito': 1, 'philadelphia': 1, 'manchurian': 1, 'candidate': 1, 'afternoon': 1, 'diabolique': 1, 'traffic': 1, 'cries': 1, 'whispers': 1, 'closely': 1, 'trains': 1, 'strangelove': 1, 'maltese': 1, 'falcon': 1, 'chinatown': 1, 'rules': 1, 'scrooge': 1, 'eraserhead': 1, 'apu': 1, 'grapes': 1, 'rosemary': 1, 'das': 1, 'boot': 1, 'stagecoach': 1, 'yankee': 1, 'doodle': 1, 'dandy': 1, 'kwai': 1, 'asphalt': 1, 'stranger': 3, 'alicein': 1, 'hara': 1}\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 842.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d7pIJmZCgRMA"
      },
      "source": [
        "### 1.2.4 Cumulative token frequency"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 529
        },
        "id": "1mi4Vw5ABFXg",
        "outputId": "0bb72afe-75e2-4f9c-d64c-cc8694710a71"
      },
      "source": [
        "# Plot the cumulative distribution of token frequency\n",
        "def cumulative_token_frequency(series, limit=20):\n",
        "    '''\n",
        "      Input:\n",
        "        series - pd.Series of words\n",
        "      Output:\n",
        "        [plot] - cumulative distribution of token frequency\n",
        "    '''\n",
        "    corpus=[word for word in series]\n",
        "    counter=Counter(corpus)\n",
        "    tokens_count = dict(counter).items()\n",
        "\n",
        "    prop_list = []\n",
        "    print(\"Vocabulary Size: \", len(tokens_count))\n",
        "    for i in range(limit):\n",
        "        tokens_filtered = len(list(filter(lambda x: x[1]<=i, tokens_count)))\n",
        "        prop_list.append(round(tokens_filtered*100/len(tokens_count),2))\n",
        "    a4_dims = (11.7, 8.27)\n",
        "    fig, ax = plt.subplots(figsize=a4_dims)\n",
        "    plt.plot(prop_list)\n",
        "    plt.grid()\n",
        "    plt.xlabel(\"Counts\")\n",
        "    plt.ylabel(\"Proportion of Vocabulary (%)\")\n",
        "    # print(\"Proportion of unique words less than\",limit,\": \", round(tokens_filtered*100/len(tokens_dict),2),\"%\")\n",
        "\n",
        "cumulative_token_frequency(df[\"Word\"])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Vocabulary Size:  6710\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 842.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KE4VV7X4fUwj"
      },
      "source": [
        "### 1.2.5 Entity Frequency"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 546
        },
        "id": "dB_9MvjOAoqq",
        "outputId": "d29bc77c-a1a2-490c-c03f-154021e36148"
      },
      "source": [
        "tag_counter = plot_top_non_stopwords_barchart(df[\"Tag\"], top=25, word=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "There are 25 distinct tags in dataset\n",
            "{'O': 61008, 'B-ACTOR': 3220, 'I-ACTOR': 3474, 'B-YEAR': 2858, 'B-TITLE': 2376, 'B-GENRE': 4354, 'I-GENRE': 786, 'B-DIRECTOR': 1720, 'I-DIRECTOR': 1850, 'B-SONG': 245, 'I-SONG': 446, 'B-PLOT': 1927, 'I-PLOT': 1687, 'B-REVIEW': 221, 'B-CHARACTER': 385, 'I-CHARACTER': 342, 'B-RATING': 2007, 'B-RATINGS_AVERAGE': 1869, 'I-RATINGS_AVERAGE': 1673, 'I-TITLE': 3495, 'I-RATING': 840, 'B-TRAILER': 113, 'I-TRAILER': 7, 'I-REVIEW': 132, 'I-YEAR': 2456}\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 842.4x595.44 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YPCRkkm768sx"
      },
      "source": [
        "<a name=\"1.3\"></a>\n",
        "## Conclusion after analysis"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BhDBB8847Dev"
      },
      "source": [
        "### Vocab\n",
        "1. The vocab size is 6710. This is not a large dataset.\n",
        "2. Our dataset it comprised of short sentences, with average length of 10 and median length of 9, and the length ranges from 1 to 47. Only a small amount of sentences have the length greater than 30 => set the max length equal to 30 => need a lot of padding tokens.\n",
        "3. 50% of the vocabulary only occur once but they could be person's names so let's keep them.\n",
        "4. Year: can be replaced by a common\n",
        "5. Number in text: can be replaced by a common\n",
        "6. Lemmatization: *films* to *film*\n",
        "7. All words are in lowercase.\n",
        "8. No punctuations.\n",
        "9. No informal text.\n",
        "10. No abbreviation (like *'ll, can't, 2morrow*)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fRaEnZIe8UWy"
      },
      "source": [
        "### Tags\n",
        "1. Most of tags are short, with median length and average length of 2 words. The longest tag has 16 words. This could be a movie name.\n",
        "2. Most sentences are about movies and ratings.\n",
        "3. There are 25 classes of entities, divided to 3 categories: B Tags (Beginning of an entity), I Tag (Intermediate Entity), Or None Tag (O). Proportions of B,I,O are about 21.40%, 17.28%, 61.32% respectively.\n",
        "4. Most of the tags are in the minority and O is the most common entity => need to over-sample the tags from the minority groups.\n",
        "5. As under the section 2.7 Check the dataset imbalance, the percentage of sentences that only contain O tags -> 0.59% => It's small amount so we don't need to delete those sentences.\n",
        "6. ALso under the section 2.7 Check the dataset imbalance, the percentage of OOV tokens in test set -> 3.53%\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dtr2P3ZE3sE6"
      },
      "source": [
        "<a name=\"2\"></a>\n",
        "# Part 2:  Pre-process the data\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MRofcZfAUd9g"
      },
      "source": [
        "<a name=\"2.1\"></a>\n",
        "## 2.1 Stemming"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1J6BsedjUeOf"
      },
      "source": [
        "def stem_sentence(sentence):\n",
        "    sentence = sentence.split(' ')\n",
        "    stemmer = PorterStemmer()\n",
        "    result = [stemmer.stem(word) for word in sentence]\n",
        "    stemmed_sentence = ' '.join(result)\n",
        "    return stemmed_sentence"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4p6nLSAePE9U"
      },
      "source": [
        "<a name=\"2.2\"></a>\n",
        "## 2.2 Lemmatization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "reh2ym9rO2PM",
        "outputId": "e58500ab-a58e-4abe-ab2c-48e905ed18d5"
      },
      "source": [
        "nltk.download('punkt')\n",
        "nltk.download('wordnet')\n",
        "def lemmatize_sentence(sentence):\n",
        "    tokenization = nltk.word_tokenize(sentence)\n",
        "    lemmatizer = WordNetLemmatizer()\n",
        "    result = [lemmatizer.lemmatize(word) for word in tokenization]\n",
        "    lemmatized_sentence = ' '.join(result)\n",
        "    return lemmatized_sentence"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Unzipping tokenizers/punkt.zip.\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/wordnet.zip.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hm9UnKBqUeZA"
      },
      "source": [
        "<a name=\"2.3\"></a>\n",
        "## 2.3 Replacement"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EZzhc1zBUeiv"
      },
      "source": [
        "import re\n",
        "\n",
        "def replace(sentence, to_replace, replace_by):\n",
        "    replaced = sentence.replace(to_replace, replace_by)\n",
        "    return replaced\n",
        "\n",
        "def replace_num(sentence):\n",
        "    replaced = re.sub(r'^\\d{1,2}$', \"NUM\", sentence) # replace 1, 2 digits\n",
        "    replaced = re.sub(r'^\\d{4}$', \"YEAR\", replaced) # replace year\n",
        "    replaced = re.sub(r'^\\d{4}s$', \"YEAR\", replaced) # replace year\n",
        "    return replaced\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vfswr-bAXwl5"
      },
      "source": [
        "<a name=\"2.4\"></a>\n",
        "## 2.4 Pre-processing pipeline"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uvGcbsdOXw3N"
      },
      "source": [
        "def apply_preproc(data_generator):\n",
        "    data_generator = list(map(lambda x: replace(x, \"ca n t\",\"cannot\"), data_generator))\n",
        "    data_generator = list(map(lambda x: replace(x, \"ll\",\"will\"), data_generator))\n",
        "    data_generator = list(map(lambda x: replace_num(x), data_generator))\n",
        "    data_generator = list(map(lambda x: lemmatize_sentence(x), data_generator))\n",
        "    return data_generator\n",
        "\n",
        "processed_sentences = apply_preproc(sentences)\n",
        "processed_test_sentences = apply_preproc(test_sentences)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WMK0gaxoOi97"
      },
      "source": [
        "<a name=\"2.5\"></a>\n",
        "## 2.5 Split to train/val datasets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yjPJLq1I0_CX"
      },
      "source": [
        "split_ratio = 0.8\n",
        "\n",
        "def train_val_split(data, label, ratio, shuffle=True, random_seed=33):\n",
        "    length = len(data)\n",
        "    lines_index = [*range(length)]  \n",
        "    # shuffle the indexes if shuffle is set to True\n",
        "    rnd.seed(random_seed)\n",
        "    if shuffle:\n",
        "        rnd.shuffle(lines_index)\n",
        "    split_point = int(length * ratio)\n",
        "\n",
        "    train_data = []\n",
        "    train_label = []\n",
        "    val_data = []\n",
        "    val_label = []\n",
        "    for i in range(length):\n",
        "        if i <= split_point:\n",
        "            train_data.append(data[lines_index[i]])\n",
        "            train_label.append(label[lines_index[i]])\n",
        "        else:\n",
        "            val_data.append(data[lines_index[i]])\n",
        "            val_label.append(label[lines_index[i]])\n",
        "    return train_data, train_label, val_data, val_label\n",
        "\n",
        "\n",
        "train_sentences, train_tags, val_sentences, val_tags = \\\n",
        "      train_val_split(processed_sentences, tags, split_ratio)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3Sot8kCJltNu"
      },
      "source": [
        "<a name=\"2.6\"></a>\n",
        "## 2.6 Tokenization and Padding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GDfsVDq9KxgV"
      },
      "source": [
        "oov_tok = \"<OOV>\"\n",
        "trunc_type='post'\n",
        "pad_type='post'\n",
        "max_length = 50\n",
        "\n",
        "vocab_tokenizer = Tokenizer(oov_token=oov_tok)\n",
        "vocab_tokenizer.fit_on_texts(train_sentences)\n",
        "\n",
        "vocab = vocab_tokenizer.word_index\n",
        "reverse_vocab = dict([(value, key) for (key, value) in vocab.items()])\n",
        "vocab_size = len(vocab)\n",
        "\n",
        "\n",
        "train_sequences = vocab_tokenizer.texts_to_sequences(train_sentences)\n",
        "val_sequences = vocab_tokenizer.texts_to_sequences(val_sentences)\n",
        "test_sequences = vocab_tokenizer.texts_to_sequences(processed_test_sentences)\n",
        "\n",
        "train_padded_sequences = pad_sequences(train_sequences,\n",
        "                                        maxlen=max_length, \n",
        "                                        truncating=trunc_type, \n",
        "                                        padding=pad_type)\n",
        "\n",
        "val_padded_sequences = pad_sequences(val_sequences,\n",
        "                                      maxlen=max_length, \n",
        "                                      truncating=trunc_type, \n",
        "                                      padding=pad_type)\n",
        "\n",
        "test_padded_sequences = pad_sequences(test_sequences,\n",
        "                                      maxlen=max_length, \n",
        "                                      truncating=trunc_type, \n",
        "                                      padding=pad_type)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CP_DmX44K24X",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c581b751-a8a5-438a-cdde-53305a7d09a9"
      },
      "source": [
        "tag_tokenizer = Tokenizer(filters=\".\", lower=False, oov_token=oov_tok)\n",
        "tag_tokenizer.fit_on_texts(train_tags)\n",
        "\n",
        "tag_map = tag_tokenizer.word_index\n",
        "reverse_tag_map = dict([(value, key) for (key, value) in tag_map.items()])\n",
        "tag_size = len(tag_map)\n",
        "\n",
        "train_tag_sequences = tag_tokenizer.texts_to_sequences(train_tags)\n",
        "val_tag_sequences = tag_tokenizer.texts_to_sequences(val_tags)\n",
        "test_tag_sequences = tag_tokenizer.texts_to_sequences(test_tags)\n",
        "\n",
        "\n",
        "train_padded_tags = pad_sequences(train_tag_sequences,\n",
        "                                        maxlen=max_length, \n",
        "                                        truncating=trunc_type, \n",
        "                                        padding=pad_type)\n",
        "\n",
        "val_padded_tags = pad_sequences(val_tag_sequences,\n",
        "                                      maxlen=max_length, \n",
        "                                      truncating=trunc_type, \n",
        "                                      padding=pad_type)\n",
        "\n",
        "test_padded_tags = pad_sequences(test_tag_sequences,\n",
        "                                      maxlen=max_length, \n",
        "                                      truncating=trunc_type, \n",
        "                                      padding=pad_type)\n",
        "\n",
        "print(\"\\nExample of a a sentence and its tokenized, padded version\")\n",
        "print(train_sentences[0])\n",
        "print(train_padded_sequences[0])\n",
        "print(\"\\nExample of a list of tags in a sentence and its tokenized, padded version\")\n",
        "print(train_tags[0])\n",
        "print(train_padded_tags[0])\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "Example of a a sentence and its tokenized, padded version\n",
            "what is the movie triangle\n",
            "[   5    7    2    3 1866    0    0    0    0    0    0    0    0    0\n",
            "    0    0    0    0    0    0    0    0    0    0    0    0    0    0\n",
            "    0    0    0    0    0    0    0    0    0    0    0    0    0    0\n",
            "    0    0    0    0    0    0    0    0]\n",
            "\n",
            "Example of a list of tags in a sentence and its tokenized, padded version\n",
            "O O O O B-TITLE\n",
            "[2 2 2 2 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
            " 0 0 0 0 0 0 0 0 0 0 0 0 0]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VmNBER6vh8fZ"
      },
      "source": [
        "<a name=\"2.7\"></a>\n",
        "## 2.7 Check the Imbalance in train/test dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1JkHts8fDElj",
        "outputId": "594fd0fd-2ea0-4688-b005-be211a32076f"
      },
      "source": [
        "#Percentage of B, I and O Tags in train dataset\n",
        "def get_tag_proportion(series_tags):\n",
        "    '''\n",
        "      Input:\n",
        "        series_tags - pd.Series of tags\n",
        "      Output:\n",
        "        [print] - B, I and O tags' proportion\n",
        "    '''\n",
        "    tags_list=[tag for tag in series_tags]\n",
        "    counter=dict(Counter(tags_list))\n",
        "    beg = 0\n",
        "    inter = 0\n",
        "    out = 0\n",
        "    for key, value in counter.items():\n",
        "        if key.startswith(\"B\"):\n",
        "            beg += value\n",
        "        elif key.startswith(\"I\"):\n",
        "            inter += value\n",
        "        else:\n",
        "            out += value\n",
        "    total = len(tags_list)\n",
        "    print(\"B tags proportion = {0:.2%}\".format(round(beg/total,4)))\n",
        "    print(\"I tags proportion = {0:.2%}\".format(round(inter/total,4)))\n",
        "    print(\"O tags proportion = {0:.2%}\".format(round(out/total,4)))\n",
        "\n",
        "get_tag_proportion(df[\"Tag\"])\n",
        "\n",
        "# Percentage of sentences that only contain O tags\n",
        "# If this percentage > 50% => the dataset is imbalanced => drop empty sentences\n",
        "def get_empty_tag_sentence_proportion(list_tag_sequence):\n",
        "    '''\n",
        "      Input:\n",
        "        list_tag_sequence - list of tag sequences in train/test set\n",
        "      Output:\n",
        "        [print] - Percentage of sentences that only contain O tags\n",
        "    '''\n",
        "    count = 0\n",
        "    for seq in list_tag_sequence:\n",
        "        if sum(seq) == 2 * len(seq): # if seq contains only 2 (token for O tag)\n",
        "            count += 1\n",
        "\n",
        "    \n",
        "    print(\"\\nPercentage of sentences that only contain O tags -> {0:.2%}\".\\\n",
        "          format(round(count/len(list_tag_sequence),4)))\n",
        "    \n",
        "get_empty_tag_sentence_proportion(train_tag_sequences)\n",
        "\n",
        "def get_OOV_density(list_token_sequence):\n",
        "    '''\n",
        "      Input:\n",
        "        list_token_sequence - list of token sequences in test set\n",
        "      Output:\n",
        "        [print] - Percentage of OOV token in the test set\n",
        "    '''\n",
        "    list_token_sequence = [token for seq in list_token_sequence for token in seq]\n",
        "    counter=dict(Counter(list_token_sequence))\n",
        "    print(\"\\nPercentage of OOV tokens in test set -> {0:.2%}\".\\\n",
        "          format(round(counter[1]/len(list_token_sequence),4)))\n",
        "\n",
        "get_OOV_density(test_sequences)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "B tags proportion = 21.40%\n",
            "I tags proportion = 17.28%\n",
            "O tags proportion = 61.32%\n",
            "\n",
            "Percentage of sentences that only contain O tags -> 0.59%\n",
            "\n",
            "Percentage of OOV tokens in test set -> 3.54%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0OZsbuw2iDAp"
      },
      "source": [
        "<a name=\"2.8\"></a>\n",
        "## 2.8 One-hot encoding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qdsnDiIZwGaC"
      },
      "source": [
        "# Ont hot encoding\n",
        "train_padded_tags = np.array([to_categorical(tags, num_classes = tag_size+1) \\\n",
        "                              for tags in train_padded_tags])\n",
        "val_padded_tags = np.array([to_categorical(tags, num_classes = tag_size+1) \\\n",
        "                            for tags in val_padded_tags])\n",
        "test_padded_tags = np.array([to_categorical(tags, num_classes = tag_size+1) \\\n",
        "                             for tags in test_padded_tags])\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "84RwGepu37jO"
      },
      "source": [
        "<a name=\"3\"></a>\n",
        "# Part 3:  Building the model\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yX9PMBUtuJmJ"
      },
      "source": [
        "<a name=\"3.1\"></a>\n",
        "## 3.1 Glove Embedding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3QAO1_GehxQ1",
        "outputId": "58d403c6-9566-4f90-d5e2-6e882c568a51"
      },
      "source": [
        "!mkdir -p /glove_embedding\n",
        "# Download data\n",
        "!wget --no-check-certificate \\\n",
        "http://nlp.stanford.edu/data/glove.6B.zip -O /glove_embedding/glove.6B.zip\n",
        "!unzip /glove_embedding/glove.6B.zip -d /glove_embedding\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-06-03 08:19:24--  http://nlp.stanford.edu/data/glove.6B.zip\n",
            "Resolving nlp.stanford.edu (nlp.stanford.edu)... 171.64.67.140\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://nlp.stanford.edu/data/glove.6B.zip [following]\n",
            "--2021-06-03 08:19:24--  https://nlp.stanford.edu/data/glove.6B.zip\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:443... connected.\n",
            "HTTP request sent, awaiting response... 301 Moved Permanently\n",
            "Location: http://downloads.cs.stanford.edu/nlp/data/glove.6B.zip [following]\n",
            "--2021-06-03 08:19:25--  http://downloads.cs.stanford.edu/nlp/data/glove.6B.zip\n",
            "Resolving downloads.cs.stanford.edu (downloads.cs.stanford.edu)... 171.64.64.22\n",
            "Connecting to downloads.cs.stanford.edu (downloads.cs.stanford.edu)|171.64.64.22|:80... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 862182613 (822M) [application/zip]\n",
            "Saving to: ‘/glove_embedding/glove.6B.zip’\n",
            "\n",
            "/glove_embedding/gl 100%[===================>] 822.24M  5.16MB/s    in 2m 42s  \n",
            "\n",
            "2021-06-03 08:22:07 (5.09 MB/s) - ‘/glove_embedding/glove.6B.zip’ saved [862182613/862182613]\n",
            "\n",
            "Archive:  /glove_embedding/glove.6B.zip\n",
            "  inflating: /glove_embedding/glove.6B.50d.txt  \n",
            "  inflating: /glove_embedding/glove.6B.100d.txt  \n",
            "  inflating: /glove_embedding/glove.6B.200d.txt  \n",
            "  inflating: /glove_embedding/glove.6B.300d.txt  \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "u2SIlF-NeR_o",
        "outputId": "b3d92d12-4e8c-470c-e694-33e6fc5c6fbc"
      },
      "source": [
        "GLOVE_DIR = \"/glove_embedding\"\n",
        "embedding_dim = 300\n",
        "hits = 0\n",
        "misses = 0\n",
        "embeddings_index = {}\n",
        "\n",
        "with open(os.path.join(GLOVE_DIR, 'glove.6B.300d.txt')) as f:\n",
        "    for line in f:\n",
        "        values = line.split()\n",
        "        word = values[0]\n",
        "        coefs = np.asarray(values[1:], dtype='float32')\n",
        "        embeddings_index[word] = coefs\n",
        "\n",
        "print('Found %s word vectors.' % len(embeddings_index))\n",
        "\n",
        "embedding_matrix = np.zeros((len(vocab) + 1, embedding_dim))\n",
        "for word, i in vocab.items():\n",
        "    embedding_vector = embeddings_index.get(word)\n",
        "    if embedding_vector is not None:\n",
        "        # words not found in embedding index will be all-zeros.\n",
        "        embedding_matrix[i] = embedding_vector\n",
        "        hits += 1\n",
        "    else:\n",
        "        misses += 1\n",
        "print(\"Converted %d words (%d misses)\" % (hits, misses))\n",
        "\n",
        "        "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 400000 word vectors.\n",
            "Converted 5008 words (574 misses)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3h-sT21kjV2X"
      },
      "source": [
        "<a name=\"3.2\"></a>\n",
        "## 3.2 Define the model "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tshH4jK03oWM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "67b6f8ca-5ce7-4a8f-d677-003aa2a2a847"
      },
      "source": [
        "# Model architecture\n",
        "batch_size = 32\n",
        "embedding_dim = 300\n",
        "max_length = 50\n",
        "\n",
        "def BiLSTM(vocab_size=vocab_size, tag_size=tag_size, hidden_size = 32, \n",
        "                embedding_dim=embedding_dim):\n",
        "    sequence_input = Input(shape = (max_length,))\n",
        "\n",
        "    model = Embedding(input_dim = vocab_size+1, \n",
        "                      output_dim = embedding_dim, \n",
        "                      input_length = max_length, \n",
        "                      embeddings_initializer=Constant(embedding_matrix),\n",
        "                      trainable=False,\n",
        "                      mask_zero = False)(sequence_input)\n",
        "                            \n",
        "    model = Bidirectional(LSTM(units = hidden_size,return_sequences=True,\n",
        "                              recurrent_dropout=0.1))(model)\n",
        "                  \n",
        "    model = TimeDistributed(Dense(hidden_size, activation=\"relu\"))(model)\n",
        "    outputs = Dense(tag_size+1, activation='softmax')(model)\n",
        "    #crf = CRF(tag_size+1) # CRF layer\n",
        "    #outputs = crf(model)\n",
        "\n",
        "    model = Model(inputs=sequence_input, outputs=outputs)\n",
        "\n",
        "    model.compile(optimizer=\"RMSprop\", \n",
        "                  loss = tf.keras.losses.categorical_crossentropy, \n",
        "                  metrics=['accuracy'])\n",
        "    return model\n",
        "\n",
        "model = BiLSTM(vocab_size=vocab_size, tag_size=tag_size, hidden_size = 32, \\\n",
        "                embedding_dim=embedding_dim)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
            "WARNING:tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sXWo98J3jgLQ"
      },
      "source": [
        "<a name=\"3.3\"></a>\n",
        "## 3.3 Callbacks"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d2hDC-w7DGvu"
      },
      "source": [
        "# Callback\n",
        "class myCallback(tf.keras.callbacks.Callback):\n",
        "  def on_epoch_end(self, epoch, logs={}):\n",
        "    if(logs.get('val_accuracy')>0.95):\n",
        "      print(\"\\nReached 95% accuracy so cancelling training!\")\n",
        "      self.model.stop_training = True\n",
        "\n",
        "checkpointer = ModelCheckpoint(filepath = 'NER_BiLSTM.h5',\n",
        "                       verbose = 0,\n",
        "                       mode = 'auto',\n",
        "                       save_best_only = True,\n",
        "                       monitor='val_loss')\n",
        "\n",
        "earlystopper = EarlyStopping(monitor='val_loss', min_delta=0, patience=3, \n",
        "                             verbose=0, mode='auto', \n",
        "                             baseline=None, restore_best_weights=True)\n",
        "\n",
        "initial_learning_rate = 0.001\n",
        "epochs = 10\n",
        "decay = initial_learning_rate / epochs\n",
        "def lr_time_based_decay(epoch, lr):\n",
        "    return lr * 1 / (1 + decay * epoch)\n",
        "\n",
        "lr_scheduler = tf.keras.callbacks.LearningRateScheduler(lr_time_based_decay, verbose=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gKnkkiSS4EwI"
      },
      "source": [
        "<a name=\"4\"></a>\n",
        "# Part 4:  Train the Model \n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vQ7uIYmgX3SC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "10c23a4d-c31e-45d3-9dbb-51b322b3ef9d"
      },
      "source": [
        "num_epochs = 15\n",
        "history = model.fit(train_padded_sequences, train_padded_tags, \n",
        "          batch_size=batch_size, epochs=num_epochs, \n",
        "          validation_data= (val_padded_sequences, val_padded_tags),\n",
        "          callbacks=[checkpointer, earlystopper, lr_scheduler])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/15\n",
            "\n",
            "Epoch 00001: LearningRateScheduler reducing learning rate to 0.0010000000474974513.\n",
            "245/245 [==============================] - 118s 452ms/step - loss: 0.4174 - accuracy: 0.9329 - val_loss: 0.1493 - val_accuracy: 0.9618\n",
            "Epoch 2/15\n",
            "\n",
            "Epoch 00002: LearningRateScheduler reducing learning rate to 0.0009999000574917021.\n",
            "245/245 [==============================] - 110s 449ms/step - loss: 0.1187 - accuracy: 0.9688 - val_loss: 0.1074 - val_accuracy: 0.9719\n",
            "Epoch 3/15\n",
            "\n",
            "Epoch 00003: LearningRateScheduler reducing learning rate to 0.000999700106714659.\n",
            "245/245 [==============================] - 110s 448ms/step - loss: 0.0902 - accuracy: 0.9760 - val_loss: 0.0922 - val_accuracy: 0.9759\n",
            "Epoch 4/15\n",
            "\n",
            "Epoch 00004: LearningRateScheduler reducing learning rate to 0.0009994003415259673.\n",
            "245/245 [==============================] - 110s 447ms/step - loss: 0.0762 - accuracy: 0.9794 - val_loss: 0.0825 - val_accuracy: 0.9782\n",
            "Epoch 5/15\n",
            "\n",
            "Epoch 00005: LearningRateScheduler reducing learning rate to 0.0009990007918579775.\n",
            "245/245 [==============================] - 109s 445ms/step - loss: 0.0676 - accuracy: 0.9818 - val_loss: 0.0756 - val_accuracy: 0.9802\n",
            "Epoch 6/15\n",
            "\n",
            "Epoch 00006: LearningRateScheduler reducing learning rate to 0.0009985014876310735.\n",
            "245/245 [==============================] - 109s 443ms/step - loss: 0.0614 - accuracy: 0.9836 - val_loss: 0.0713 - val_accuracy: 0.9816\n",
            "Epoch 7/15\n",
            "\n",
            "Epoch 00007: LearningRateScheduler reducing learning rate to 0.0009979026914447063.\n",
            "245/245 [==============================] - 110s 449ms/step - loss: 0.0564 - accuracy: 0.9849 - val_loss: 0.0708 - val_accuracy: 0.9817\n",
            "Epoch 8/15\n",
            "\n",
            "Epoch 00008: LearningRateScheduler reducing learning rate to 0.0009972046657933619.\n",
            "245/245 [==============================] - 110s 449ms/step - loss: 0.0523 - accuracy: 0.9861 - val_loss: 0.0684 - val_accuracy: 0.9821\n",
            "Epoch 9/15\n",
            "\n",
            "Epoch 00009: LearningRateScheduler reducing learning rate to 0.0009964075567443476.\n",
            "245/245 [==============================] - 109s 444ms/step - loss: 0.0491 - accuracy: 0.9869 - val_loss: 0.0671 - val_accuracy: 0.9825\n",
            "Epoch 10/15\n",
            "\n",
            "Epoch 00010: LearningRateScheduler reducing learning rate to 0.0009955116266172385.\n",
            "245/245 [==============================] - 108s 440ms/step - loss: 0.0461 - accuracy: 0.9876 - val_loss: 0.0676 - val_accuracy: 0.9823\n",
            "Epoch 11/15\n",
            "\n",
            "Epoch 00011: LearningRateScheduler reducing learning rate to 0.0009945171376267872.\n",
            "245/245 [==============================] - 107s 436ms/step - loss: 0.0433 - accuracy: 0.9884 - val_loss: 0.0704 - val_accuracy: 0.9814\n",
            "Epoch 12/15\n",
            "\n",
            "Epoch 00012: LearningRateScheduler reducing learning rate to 0.000993424351882975.\n",
            "245/245 [==============================] - 107s 437ms/step - loss: 0.0412 - accuracy: 0.9889 - val_loss: 0.0664 - val_accuracy: 0.9832\n",
            "Epoch 13/15\n",
            "\n",
            "Epoch 00013: LearningRateScheduler reducing learning rate to 0.0009922336476668565.\n",
            "245/245 [==============================] - 107s 435ms/step - loss: 0.0389 - accuracy: 0.9896 - val_loss: 0.0649 - val_accuracy: 0.9835\n",
            "Epoch 14/15\n",
            "\n",
            "Epoch 00014: LearningRateScheduler reducing learning rate to 0.00099094540310837.\n",
            "245/245 [==============================] - 108s 440ms/step - loss: 0.0373 - accuracy: 0.9900 - val_loss: 0.0642 - val_accuracy: 0.9838\n",
            "Epoch 15/15\n",
            "\n",
            "Epoch 00015: LearningRateScheduler reducing learning rate to 0.0009895599961864132.\n",
            "245/245 [==============================] - 109s 444ms/step - loss: 0.0352 - accuracy: 0.9905 - val_loss: 0.0668 - val_accuracy: 0.9838\n",
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 50)]              0         \n",
            "_________________________________________________________________\n",
            "embedding (Embedding)        (None, 50, 300)           1674900   \n",
            "_________________________________________________________________\n",
            "bidirectional (Bidirectional (None, 50, 64)            85248     \n",
            "_________________________________________________________________\n",
            "time_distributed (TimeDistri (None, 50, 32)            2080      \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 50, 27)            891       \n",
            "=================================================================\n",
            "Total params: 1,763,119\n",
            "Trainable params: 88,219\n",
            "Non-trainable params: 1,674,900\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M3vHe1vaC-fG",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6a982b63-5732-4ef2-a3b2-db581ec329d8"
      },
      "source": [
        "history.history.keys()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy', 'lr'])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K-LuVEjYZ1qg",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 73
        },
        "outputId": "8c62eb5c-3ef0-4630-829d-8b52aaea2086"
      },
      "source": [
        "acc = history.history['accuracy']\n",
        "val_acc = history.history['val_accuracy']\n",
        "loss = history.history['loss']\n",
        "val_loss = history.history['val_loss']\n",
        "plt.figure(figsize = (8,8))\n",
        "epochs = range(1, len(acc) + 1)\n",
        "plt.plot(epochs, acc, 'wo', label='Training acc')\n",
        "plt.plot(epochs, val_acc, 'w', label='Validation acc')\n",
        "plt.title('Training and validation accuracy')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f8ffe0d5a50>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_orawE5havtv",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 499
        },
        "outputId": "fc17f13c-d589-45a5-aa37-17dab1091f93"
      },
      "source": [
        "plt.figure(figsize = (8,8))\n",
        "plt.plot(epochs, loss, 'wo', label='Training loss')\n",
        "plt.plot(epochs, val_loss, 'w', label='Validation loss')\n",
        "plt.title('Training and validation loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-ygvHcHU4Nkj"
      },
      "source": [
        "<a name=\"5\"></a>\n",
        "# Part 5:  Test the model\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SYIAQDjhsHNQ"
      },
      "source": [
        "# Convert from one-hot encoding (3D array) to 2D array \n",
        "test_padded_tags_pred = model.predict(test_padded_sequences)\n",
        "test_padded_tags_pred = np.argmax(test_padded_tags_pred, axis=-1)\n",
        "test_padded_tags_true = np.argmax(test_padded_tags, axis=-1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aS4oaqkBE4XX"
      },
      "source": [
        "# Convert the index to tag\n",
        "test_tags_pred =[0]*len(test_padded_tags_pred)\n",
        "for idx, row in enumerate(test_padded_tags_pred):\n",
        "    add = []\n",
        "    for i in row:\n",
        "        add.append(reverse_tag_map[i]) if i != 0 else add.append(\"PAD\")\n",
        "    test_tags_pred[idx] = add\n",
        "\n",
        "test_tags_true =[0]*len(test_padded_tags_true)\n",
        "for idx, row in enumerate(test_padded_tags_true):\n",
        "    add = []\n",
        "    for i in row:\n",
        "        add.append(reverse_tag_map[i]) if i != 0 else add.append(\"PAD\")\n",
        "    test_tags_true[idx] = add"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OO5Pg9r2MmV8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5b9d58d6-77dc-4381-8a74-3e26c8dea358"
      },
      "source": [
        "print(\"Micro F1-score is : {:.1%}\".format(f1_score(test_tags_true, test_tags_pred)))\n",
        "print(\"Micro Precision-score is : {:.1%}\".format(precision_score(test_tags_true, test_tags_pred)))\n",
        "print(\"Micro Recall-score is : {:.1%}\".format(recall_score(test_tags_true, test_tags_pred)))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: PAD seems not to be NE tag.\n",
            "  warnings.warn('{} seems not to be NE tag.'.format(chunk))\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Micro F1-score is : 82.8%\n",
            "Micro Precision-score is : 82.0%\n",
            "Micro Recall-score is : 83.7%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "znKBJcadMr9l",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6952c03e-ac24-44e3-8cba-752c94c6f27d"
      },
      "source": [
        "!pip install sklearn_crfsuite\n",
        "from sklearn_crfsuite.metrics import flat_classification_report\n",
        "\n",
        "report = flat_classification_report(y_pred=test_tags_pred, y_true=test_tags_true)\n",
        "print(report)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: sklearn_crfsuite in /usr/local/lib/python3.7/dist-packages (0.3.6)\n",
            "Requirement already satisfied: python-crfsuite>=0.8.3 in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (0.9.7)\n",
            "Requirement already satisfied: tqdm>=2.0 in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (4.41.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (1.15.0)\n",
            "Requirement already satisfied: tabulate in /usr/local/lib/python3.7/dist-packages (from sklearn_crfsuite) (0.8.9)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/metrics/_classification.py:1272: 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"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "                   precision    recall  f1-score   support\n",
            "\n",
            "          B-ACTOR       0.89      0.92      0.91       812\n",
            "      B-CHARACTER       0.70      0.49      0.58        90\n",
            "       B-DIRECTOR       0.90      0.83      0.87       456\n",
            "          B-GENRE       0.91      0.94      0.92      1117\n",
            "           B-PLOT       0.74      0.58      0.65       491\n",
            "         B-RATING       0.97      0.97      0.97       500\n",
            "B-RATINGS_AVERAGE       0.91      0.88      0.90       451\n",
            "         B-REVIEW       0.67      0.04      0.07        56\n",
            "           B-SONG       0.85      0.52      0.64        54\n",
            "          B-TITLE       0.79      0.75      0.77       562\n",
            "        B-TRAILER       0.84      0.90      0.87        30\n",
            "           B-YEAR       0.95      0.94      0.94       720\n",
            "          I-ACTOR       0.91      0.89      0.90       862\n",
            "      I-CHARACTER       0.57      0.39      0.46        75\n",
            "       I-DIRECTOR       0.89      0.85      0.87       496\n",
            "          I-GENRE       0.86      0.72      0.78       222\n",
            "           I-PLOT       0.75      0.40      0.52       496\n",
            "         I-RATING       0.97      0.87      0.92       226\n",
            "I-RATINGS_AVERAGE       0.86      0.81      0.84       403\n",
            "         I-REVIEW       0.25      0.02      0.04        45\n",
            "           I-SONG       0.76      0.67      0.71       119\n",
            "          I-TITLE       0.74      0.81      0.77       856\n",
            "        I-TRAILER       0.00      0.00      0.00         8\n",
            "           I-YEAR       0.96      0.97      0.96       610\n",
            "                O       0.95      0.97      0.96     14929\n",
            "              PAD       1.00      1.00      1.00     97514\n",
            "\n",
            "         accuracy                           0.98    122200\n",
            "        macro avg       0.79      0.70      0.72    122200\n",
            "     weighted avg       0.98      0.98      0.98    122200\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HE7Q2oD2M9rL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d2c6e203-d751-4b46-c631-b5b4c165fdc3"
      },
      "source": [
        "# At every execution model picks some random test sample from test set.\n",
        "i = np.random.randint(0,test_padded_sequences.shape[0]) # choose a random number between 0 and len(X_te)b\n",
        "p = model.predict(np.array([test_padded_sequences[i]]))\n",
        "p = np.argmax(p, axis=-1)\n",
        "true = np.argmax(test_padded_tags[i], -1)\n",
        "\n",
        "print(\"Sample number {} of {} (Test Set)\".format(i, test_padded_sequences.shape[0]))\n",
        "# Visualization\n",
        "print(\"{:20}||{:20}||{}\".format(\"Word\", \"True\", \"Pred\"))\n",
        "print(60 * \"=\")\n",
        "for word, tag, pred in zip(test_padded_sequences[i], true, p[0]):\n",
        "    if word != 0:\n",
        "        print(\"{:20}: {:20} {}\".format(reverse_vocab[word], reverse_tag_map[tag], reverse_tag_map[pred]))\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sample number 1175 of 2444 (Test Set)\n",
            "Word                ||True                ||Pred\n",
            "============================================================\n",
            "did                 : O                    O\n",
            "steven              : B-DIRECTOR           B-DIRECTOR\n",
            "spielberg           : I-DIRECTOR           I-DIRECTOR\n",
            "direct              : O                    O\n",
            "any                 : O                    O\n",
            "horror              : B-GENRE              B-GENRE\n",
            "movie               : O                    O\n",
            "in                  : O                    O\n",
            "the                 : O                    O\n",
            "1980                : B-YEAR               B-YEAR\n",
            "s                   : I-YEAR               I-YEAR\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sNtJLetx4Sb_"
      },
      "source": [
        "<a name=\"6\"></a>\n",
        "# Part 6:  Test with your own sentence"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V0niUrOnVskX",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 722
        },
        "outputId": "51453fb1-958b-4b99-c405-6b8e4a705b57"
      },
      "source": [
        "# if this cell fails => run the 2nd time, it will work\n",
        "\n",
        "original_to_test = [\"is michael scofield the protagonist in prison break\", \n",
        "                    \"what is the highest rated romantic movie in all time\"]\n",
        "\n",
        "to_test = apply_preproc(original_to_test)\n",
        "\n",
        "vocab_tokenizer.fit_on_texts(to_test)\n",
        "to_test = vocab_tokenizer.texts_to_sequences(to_test)\n",
        "\n",
        "to_test = pad_sequences(to_test,\n",
        "                        maxlen=max_length, \n",
        "                        truncating=trunc_type, \n",
        "                        padding=pad_type)\n",
        "\n",
        "to_test_tag_pred = model.predict(to_test)\n",
        "to_test_tag_pred = np.argmax(to_test_tag_pred, axis=-1)\n",
        "\n",
        "for i, row in enumerate(to_test_tag_pred):\n",
        "    print(\"\\n{:20}||{}\".format(\"Word\", \"Pred\"))\n",
        "    print(40 * \"=\")\n",
        "    for j, pred in enumerate(row):\n",
        "        words = original_to_test[i].split(' ')\n",
        "        length = len(words)\n",
        "        if pred != 0 and j < length:\n",
        "            print(\"{:20}: {}\".format(words[j], reverse_tag_map[pred]))\n",
        "            "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "InvalidArgumentError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mInvalidArgumentError\u001b[0m                      Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-32-c4e2d3b9d23e>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     14\u001b[0m                         padding=pad_type)\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m \u001b[0mto_test_tag_pred\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mto_test\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     17\u001b[0m \u001b[0mto_test_tag_pred\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0margmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mto_test_tag_pred\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maxis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/keras/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1725\u001b[0m           \u001b[0;32mfor\u001b[0m \u001b[0mstep\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msteps\u001b[0m\u001b[0;34m(\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[1;32m   1726\u001b[0m             \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_predict_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1727\u001b[0;31m             \u001b[0mtmp_batch_outputs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0miterator\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   1728\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshould_sync\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1729\u001b[0m               \u001b[0mcontext\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masync_wait\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;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    887\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    888\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mOptionalXlaContext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_jit_compile\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;32m--> 889\u001b[0;31m         \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\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    890\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    891\u001b[0m       \u001b[0mnew_tracing_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexperimental_get_tracing_count\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;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\u001b[0m in \u001b[0;36m_call\u001b[0;34m(self, *args, **kwds)\u001b[0m\n\u001b[1;32m    922\u001b[0m       \u001b[0;31m# In this case we have not created variables on the first call. So we can\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    923\u001b[0m       \u001b[0;31m# run the first trace but we should fail if variables are created.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 924\u001b[0;31m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_stateful_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwds\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    925\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_created_variables\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    926\u001b[0m         raise ValueError(\"Creating variables on a non-first call to a function\"\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   3022\u001b[0m        filtered_flat_args) = self._maybe_define_function(args, kwargs)\n\u001b[1;32m   3023\u001b[0m     return graph_function._call_flat(\n\u001b[0;32m-> 3024\u001b[0;31m         filtered_flat_args, captured_inputs=graph_function.captured_inputs)  # pylint: disable=protected-access\n\u001b[0m\u001b[1;32m   3025\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   3026\u001b[0m   \u001b[0;34m@\u001b[0m\u001b[0mproperty\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36m_call_flat\u001b[0;34m(self, args, captured_inputs, cancellation_manager)\u001b[0m\n\u001b[1;32m   1959\u001b[0m       \u001b[0;31m# No tape is watching; skip to running the function.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1960\u001b[0m       return self._build_call_outputs(self._inference_function.call(\n\u001b[0;32m-> 1961\u001b[0;31m           ctx, args, cancellation_manager=cancellation_manager))\n\u001b[0m\u001b[1;32m   1962\u001b[0m     forward_backward = self._select_forward_and_backward_functions(\n\u001b[1;32m   1963\u001b[0m         \u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\u001b[0m in \u001b[0;36mcall\u001b[0;34m(self, ctx, args, cancellation_manager)\u001b[0m\n\u001b[1;32m    594\u001b[0m               \u001b[0minputs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    595\u001b[0m               \u001b[0mattrs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mattrs\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 596\u001b[0;31m               ctx=ctx)\n\u001b[0m\u001b[1;32m    597\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    598\u001b[0m           outputs = execute.execute_with_cancellation(\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/execute.py\u001b[0m in \u001b[0;36mquick_execute\u001b[0;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[1;32m     58\u001b[0m     \u001b[0mctx\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mensure_initialized\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[1;32m     59\u001b[0m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0;32m---> 60\u001b[0;31m                                         inputs, attrs, num_outputs)\n\u001b[0m\u001b[1;32m     61\u001b[0m   \u001b[0;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     62\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mname\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mInvalidArgumentError\u001b[0m: 2 root error(s) found.\n  (0) Invalid argument:  indices[0,2] = 5583 is not in [0, 5583)\n\t [[node model/embedding/embedding_lookup (defined at <ipython-input-27-733853abd5d6>:2) ]]\n  (1) Invalid argument:  indices[0,2] = 5583 is not in [0, 5583)\n\t [[node model/embedding/embedding_lookup (defined at <ipython-input-27-733853abd5d6>:2) ]]\n\t [[model/embedding/embedding_lookup/_6]]\n0 successful operations.\n0 derived errors ignored. [Op:__inference_predict_function_17969]\n\nErrors may have originated from an input operation.\nInput Source operations connected to node model/embedding/embedding_lookup:\n model/embedding/embedding_lookup/17449 (defined at /usr/lib/python3.7/contextlib.py:112)\n\nInput Source operations connected to node model/embedding/embedding_lookup:\n model/embedding/embedding_lookup/17449 (defined at /usr/lib/python3.7/contextlib.py:112)\n\nFunction call stack:\npredict_function -> predict_function\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WZJApvzvaJcy"
      },
      "source": [
        "<a name=\"7\"></a>\n",
        "# Part 7: Analyse the incorrect predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fb8q_rD4aNRH"
      },
      "source": [
        "def get_incorrect(y_pred, y_true, X_test):\n",
        "    y_pred.flatten()\n",
        "    y_true.flatten()\n",
        "    X_test.flatten()\n",
        "    where_incorrect = y_true != y_pred\n",
        "    incorrect_idxes = np.where(where_incorrect==1)[0]\n",
        "    incorrect_tokens = X_test[incorrect_idxes]\n",
        "    incorrect_tokens = dict(Counter(incorrect_tokens.flatten()))\n",
        "    incorrect_tags = y_true[incorrect_idxes]\n",
        "    incorrect_tags = dict(Counter(incorrect_tags.flatten()))\n",
        "    return incorrect_tokens, incorrect_tags\n",
        "\n",
        "\n",
        "incorrect_tokens, incorrect_tags = get_incorrect(test_padded_tags_pred, \n",
        "                                                 test_padded_tags_true, \n",
        "                                                 test_padded_sequences) \n",
        "\n",
        "incorrect_tokens = sorted(incorrect_tokens.items(), key=lambda x:x[1], reverse=True)\n",
        "incorrect_tags = sorted(incorrect_tags.items(), key=lambda x:x[1], reverse=True)\n",
        "\n",
        "print(\"{:^20}||{:^15}\".format(\"Incorrect word\", \"Frequency\"))\n",
        "print(37 * \"=\")\n",
        "for idx, count in incorrect_tokens[:20]:\n",
        "    if idx != 0:\n",
        "        print(\"{:20}: {:15}\".format(reverse_vocab[idx], count))\n",
        "\n",
        "print(\"\\n{:^20}||{:^15}\".format(\"Incorrect tag\", \"Frequency\"))\n",
        "print(37 * \"=\")\n",
        "for idx, count in incorrect_tags[:20]:\n",
        "    if idx != 0:\n",
        "        print(\"{:20}: {:15}\".format(reverse_tag_map[idx], count))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5rOw_59Ztzy7"
      },
      "source": [
        "## Conclusion after analysis "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NiAN2Zy--j-_"
      },
      "source": [
        "1. A lot of incorrect words are stopwords, due to their occurence in the movie titles/plots sometimes. This is the problem of token label inconsistency.\n",
        "2. The values of macro-average are much lower than micro-average of Precision, Recall and F1-score, as a result of imbalanced classes.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N2J2LDiBRkyq"
      },
      "source": [
        "## Potential improvements"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9aBUF7SIHVhd"
      },
      "source": [
        "1. To tackle imbalanced classes problem, use over-sample to gain more examples of tags from the minority groups.\n",
        "2. To tackle label inconsistency, there're 3 solutions:\n",
        "\n",
        "  *   Use larger context. For example, use longer sentences, or combine 2 or more sentences that have similar/corelated meaning.\n",
        "  *   Use CRF decoder layer.\n",
        "  *   Use Character/Subword-level encoders like ELMO, Flair, CNN and BERT. \n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NkJ8Go7BVVgz"
      },
      "source": [
        "<a name=\"8\"></a>\n",
        "# Export result to .tsv file"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3EQxLZgzRdL9"
      },
      "source": [
        "# write result to a new txt file\n",
        "with open('/tmp/pred.tsv', 'wt') as out_file:\n",
        "    tsv_writer = csv.writer(out_file, delimiter='\\t')\n",
        "    test_size = len(test_padded_sequences)\n",
        "    for i in range(test_size):\n",
        "        for pred, word in zip(test_padded_tags_pred[i], test_padded_sequences[i]):\n",
        "            if pred != 0 and word != 0:\n",
        "                tsv_writer.writerow([reverse_tag_map[pred], reverse_vocab[word]])\n",
        "        tsv_writer.writerow([])"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}