notebooks/EJ_extractions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8d454419-c5b4-4196-9a60-f4ad297b63fb",
   "metadata": {},
   "source": [
    "# This notebook will provide details on metadata extraction process before and after finetuning on Environmental Justice Datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a9570085-9dcd-452b-b506-c8017c6b61ea",
   "metadata": {},
   "source": [
    "# Installing the required packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bf352e5e-5f66-4885-853a-723e2368ab14",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Obtaining file:///home/sagemaker-user/LLM-cookbook-for-open-science/notebooks\n",
      "\u001b[31mERROR: file:///home/sagemaker-user/LLM-cookbook-for-open-science/notebooks does not appear to be a Python project: neither 'setup.py' nor 'pyproject.toml' found.\u001b[0m\u001b[31m\n",
      "\u001b[0mRequirement already satisfied: spacy in /opt/conda/lib/python3.10/site-packages (3.7.2)\n",
      "Requirement already satisfied: spacy-legacy<3.1.0,>=3.0.11 in /opt/conda/lib/python3.10/site-packages (from spacy) (3.0.12)\n",
      "Requirement already satisfied: spacy-loggers<2.0.0,>=1.0.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (1.0.5)\n",
      "Requirement already satisfied: murmurhash<1.1.0,>=0.28.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (1.0.10)\n",
      "Requirement already satisfied: cymem<2.1.0,>=2.0.2 in /opt/conda/lib/python3.10/site-packages (from spacy) (2.0.8)\n",
      "Requirement already satisfied: preshed<3.1.0,>=3.0.2 in /opt/conda/lib/python3.10/site-packages (from spacy) (3.0.9)\n",
      "Requirement already satisfied: thinc<8.3.0,>=8.1.8 in /opt/conda/lib/python3.10/site-packages (from spacy) (8.2.2)\n",
      "Requirement already satisfied: wasabi<1.2.0,>=0.9.1 in /opt/conda/lib/python3.10/site-packages (from spacy) (1.1.2)\n",
      "Requirement already satisfied: srsly<3.0.0,>=2.4.3 in /opt/conda/lib/python3.10/site-packages (from spacy) (2.4.8)\n",
      "Requirement already satisfied: catalogue<2.1.0,>=2.0.6 in /opt/conda/lib/python3.10/site-packages (from spacy) (2.0.10)\n",
      "Requirement already satisfied: weasel<0.4.0,>=0.1.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (0.3.4)\n",
      "Requirement already satisfied: typer<0.10.0,>=0.3.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (0.9.0)\n",
      "Requirement already satisfied: smart-open<7.0.0,>=5.2.1 in /opt/conda/lib/python3.10/site-packages (from spacy) (5.2.1)\n",
      "Requirement already satisfied: tqdm<5.0.0,>=4.38.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (4.66.1)\n",
      "Requirement already satisfied: requests<3.0.0,>=2.13.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (2.31.0)\n",
      "Requirement already satisfied: pydantic!=1.8,!=1.8.1,<3.0.0,>=1.7.4 in /opt/conda/lib/python3.10/site-packages (from spacy) (2.6.1)\n",
      "Requirement already satisfied: jinja2 in /opt/conda/lib/python3.10/site-packages (from spacy) (3.1.3)\n",
      "Requirement already satisfied: setuptools in /opt/conda/lib/python3.10/site-packages (from spacy) (69.1.1)\n",
      "Requirement already satisfied: packaging>=20.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (23.2)\n",
      "Requirement already satisfied: langcodes<4.0.0,>=3.2.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (3.3.0)\n",
      "Requirement already satisfied: numpy>=1.19.0 in /opt/conda/lib/python3.10/site-packages (from spacy) (1.26.4)\n",
      "Requirement already satisfied: annotated-types>=0.4.0 in /opt/conda/lib/python3.10/site-packages (from pydantic!=1.8,!=1.8.1,<3.0.0,>=1.7.4->spacy) (0.6.0)\n",
      "Collecting pydantic-core==2.16.2 (from pydantic!=1.8,!=1.8.1,<3.0.0,>=1.7.4->spacy)\n",
      "  Using cached pydantic_core-2.16.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.5 kB)\n",
      "Requirement already satisfied: typing-extensions>=4.6.1 in /opt/conda/lib/python3.10/site-packages (from pydantic!=1.8,!=1.8.1,<3.0.0,>=1.7.4->spacy) (4.9.0)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0,>=2.13.0->spacy) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0,>=2.13.0->spacy) (3.6)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0,>=2.13.0->spacy) (1.26.18)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests<3.0.0,>=2.13.0->spacy) (2024.2.2)\n",
      "Requirement already satisfied: blis<0.8.0,>=0.7.8 in /opt/conda/lib/python3.10/site-packages (from thinc<8.3.0,>=8.1.8->spacy) (0.7.10)\n",
      "Requirement already satisfied: confection<1.0.0,>=0.0.1 in /opt/conda/lib/python3.10/site-packages (from thinc<8.3.0,>=8.1.8->spacy) (0.1.4)\n",
      "Requirement already satisfied: click<9.0.0,>=7.1.1 in /opt/conda/lib/python3.10/site-packages (from typer<0.10.0,>=0.3.0->spacy) (8.1.7)\n",
      "Requirement already satisfied: cloudpathlib<0.17.0,>=0.7.0 in /opt/conda/lib/python3.10/site-packages (from weasel<0.4.0,>=0.1.0->spacy) (0.16.0)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in /opt/conda/lib/python3.10/site-packages (from jinja2->spacy) (2.1.5)\n",
      "Using cached pydantic_core-2.16.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.2 MB)\n",
      "Installing collected packages: pydantic-core\n",
      "  Attempting uninstall: pydantic-core\n",
      "    Found existing installation: pydantic_core 2.16.3\n",
      "    Uninstalling pydantic_core-2.16.3:\n",
      "      Successfully uninstalled pydantic_core-2.16.3\n",
      "Successfully installed pydantic-core-2.16.2\n",
      "Requirement already satisfied: pandas in /opt/conda/lib/python3.10/site-packages (2.1.4)\n",
      "Requirement already satisfied: numpy<2,>=1.22.4 in /opt/conda/lib/python3.10/site-packages (from pandas) (1.26.4)\n",
      "Requirement already satisfied: python-dateutil>=2.8.2 in /opt/conda/lib/python3.10/site-packages (from pandas) (2.8.2)\n",
      "Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.10/site-packages (from pandas) (2023.3)\n",
      "Requirement already satisfied: tzdata>=2022.1 in /opt/conda/lib/python3.10/site-packages (from pandas) (2024.1)\n",
      "Requirement already satisfied: six>=1.5 in /opt/conda/lib/python3.10/site-packages (from python-dateutil>=2.8.2->pandas) (1.16.0)\n",
      "Requirement already satisfied: openpyxl in /opt/conda/lib/python3.10/site-packages (3.1.2)\n",
      "Requirement already satisfied: et-xmlfile in /opt/conda/lib/python3.10/site-packages (from openpyxl) (1.1.0)\n",
      "Requirement already satisfied: beautifulsoup4 in /opt/conda/lib/python3.10/site-packages (4.12.3)\n",
      "Requirement already satisfied: soupsieve>1.2 in /opt/conda/lib/python3.10/site-packages (from beautifulsoup4) (2.5)\n",
      "Requirement already satisfied: sentence_transformers in /opt/conda/lib/python3.10/site-packages (2.6.1)\n",
      "Requirement already satisfied: transformers<5.0.0,>=4.32.0 in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (4.39.1)\n",
      "Requirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (4.66.1)\n",
      "Requirement already satisfied: torch>=1.11.0 in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (2.0.0.post101)\n",
      "Requirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (1.26.4)\n",
      "Requirement already satisfied: scikit-learn in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (1.4.1.post1)\n",
      "Requirement already satisfied: scipy in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (1.11.4)\n",
      "Requirement already satisfied: huggingface-hub>=0.15.1 in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (0.21.1)\n",
      "Requirement already satisfied: Pillow in /opt/conda/lib/python3.10/site-packages (from sentence_transformers) (9.5.0)\n",
      "Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (3.13.1)\n",
      "Requirement already satisfied: fsspec>=2023.5.0 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (2023.6.0)\n",
      "Requirement already satisfied: requests in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (2.31.0)\n",
      "Requirement already satisfied: pyyaml>=5.1 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (6.0.1)\n",
      "Requirement already satisfied: typing-extensions>=3.7.4.3 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (4.9.0)\n",
      "Requirement already satisfied: packaging>=20.9 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.15.1->sentence_transformers) (23.2)\n",
      "Requirement already satisfied: sympy in /opt/conda/lib/python3.10/site-packages (from torch>=1.11.0->sentence_transformers) (1.12)\n",
      "Requirement already satisfied: networkx in /opt/conda/lib/python3.10/site-packages (from torch>=1.11.0->sentence_transformers) (3.2.1)\n",
      "Requirement already satisfied: jinja2 in /opt/conda/lib/python3.10/site-packages (from torch>=1.11.0->sentence_transformers) (3.1.3)\n",
      "Requirement already satisfied: regex!=2019.12.17 in /opt/conda/lib/python3.10/site-packages (from transformers<5.0.0,>=4.32.0->sentence_transformers) (2023.12.25)\n",
      "Requirement already satisfied: tokenizers<0.19,>=0.14 in /opt/conda/lib/python3.10/site-packages (from transformers<5.0.0,>=4.32.0->sentence_transformers) (0.15.2)\n",
      "Requirement already satisfied: safetensors>=0.4.1 in /opt/conda/lib/python3.10/site-packages (from transformers<5.0.0,>=4.32.0->sentence_transformers) (0.4.2)\n",
      "Requirement already satisfied: joblib>=1.2.0 in /opt/conda/lib/python3.10/site-packages (from scikit-learn->sentence_transformers) (1.3.2)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /opt/conda/lib/python3.10/site-packages (from scikit-learn->sentence_transformers) (3.3.0)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in /opt/conda/lib/python3.10/site-packages (from jinja2->torch>=1.11.0->sentence_transformers) (2.1.5)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.15.1->sentence_transformers) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.15.1->sentence_transformers) (3.6)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.15.1->sentence_transformers) (1.26.18)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.15.1->sentence_transformers) (2024.2.2)\n",
      "Requirement already satisfied: mpmath>=0.19 in /opt/conda/lib/python3.10/site-packages (from sympy->torch>=1.11.0->sentence_transformers) (1.3.0)\n",
      "Requirement already satisfied: bert_score in /opt/conda/lib/python3.10/site-packages (0.3.13)\n",
      "Requirement already satisfied: torch>=1.0.0 in /opt/conda/lib/python3.10/site-packages (from bert_score) (2.0.0.post101)\n",
      "Requirement already satisfied: pandas>=1.0.1 in /opt/conda/lib/python3.10/site-packages (from bert_score) (2.1.4)\n",
      "Requirement already satisfied: transformers>=3.0.0 in /opt/conda/lib/python3.10/site-packages (from bert_score) (4.39.1)\n",
      "Requirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages (from bert_score) (1.26.4)\n",
      "Requirement already satisfied: requests in /opt/conda/lib/python3.10/site-packages (from bert_score) (2.31.0)\n",
      "Requirement already satisfied: tqdm>=4.31.1 in /opt/conda/lib/python3.10/site-packages (from bert_score) (4.66.1)\n",
      "Requirement already satisfied: matplotlib in /opt/conda/lib/python3.10/site-packages (from bert_score) (3.8.3)\n",
      "Requirement already satisfied: packaging>=20.9 in /opt/conda/lib/python3.10/site-packages (from bert_score) (23.2)\n",
      "Requirement already satisfied: python-dateutil>=2.8.2 in /opt/conda/lib/python3.10/site-packages (from pandas>=1.0.1->bert_score) (2.8.2)\n",
      "Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.10/site-packages (from pandas>=1.0.1->bert_score) (2023.3)\n",
      "Requirement already satisfied: tzdata>=2022.1 in /opt/conda/lib/python3.10/site-packages (from pandas>=1.0.1->bert_score) (2024.1)\n",
      "Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from torch>=1.0.0->bert_score) (3.13.1)\n",
      "Requirement already satisfied: typing-extensions in /opt/conda/lib/python3.10/site-packages (from torch>=1.0.0->bert_score) (4.9.0)\n",
      "Requirement already satisfied: sympy in /opt/conda/lib/python3.10/site-packages (from torch>=1.0.0->bert_score) (1.12)\n",
      "Requirement already satisfied: networkx in /opt/conda/lib/python3.10/site-packages (from torch>=1.0.0->bert_score) (3.2.1)\n",
      "Requirement already satisfied: jinja2 in /opt/conda/lib/python3.10/site-packages (from torch>=1.0.0->bert_score) (3.1.3)\n",
      "Requirement already satisfied: huggingface-hub<1.0,>=0.19.3 in /opt/conda/lib/python3.10/site-packages (from transformers>=3.0.0->bert_score) (0.21.1)\n",
      "Requirement already satisfied: pyyaml>=5.1 in /opt/conda/lib/python3.10/site-packages (from transformers>=3.0.0->bert_score) (6.0.1)\n",
      "Requirement already satisfied: regex!=2019.12.17 in /opt/conda/lib/python3.10/site-packages (from transformers>=3.0.0->bert_score) (2023.12.25)\n",
      "Requirement already satisfied: tokenizers<0.19,>=0.14 in /opt/conda/lib/python3.10/site-packages (from transformers>=3.0.0->bert_score) (0.15.2)\n",
      "Requirement already satisfied: safetensors>=0.4.1 in /opt/conda/lib/python3.10/site-packages (from transformers>=3.0.0->bert_score) (0.4.2)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (1.2.0)\n",
      "Requirement already satisfied: cycler>=0.10 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (0.12.1)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (4.49.0)\n",
      "Requirement already satisfied: kiwisolver>=1.3.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (1.4.5)\n",
      "Requirement already satisfied: pillow>=8 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (9.5.0)\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert_score) (3.1.1)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests->bert_score) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests->bert_score) (3.6)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests->bert_score) (1.26.18)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests->bert_score) (2024.2.2)\n"
     ]
    }
   ],
   "source": [
    "!pip install -e .\n",
    "!pip install spacy\n",
    "!pip install pandas\n",
    "!pip install openpyxl\n",
    "!pip install beautifulsoup4\n",
    "!pip install sentence_transformers\n",
    "!pip install bert_score\n",
    "!pip install python-dotenv\n",
    "# !pip install instructor\n",
    "!pip install loguru\n",
    "!pip install nltk\n",
    "!pip install unstructured\n",
    "!pip install python-docx \n",
    "!pip install exceptions\n",
    "!pip install pypdf\n",
    "!pip install sqlalchemy==2.0.28\n",
    "# !pip install langchain\n",
    "!python3 -m spacy download en_core_web_lg\n",
    "!pip install git+https://git@github.com/NASA-IMPACT/larch.git --no-cache-dir"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63435f43-e01f-499b-adcc-39fa8ca086cd",
   "metadata": {},
   "source": [
    "## Importing necessary modules and packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "210d6f90-3864-4e0a-8dca-5f40d299b1e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import os\n",
    "import re\n",
    "import ssl\n",
    "import requests\n",
    "from bs4 import BeautifulSoup\n",
    "from dotenv import load_dotenv\n",
    "import instructor\n",
    "from loguru import logger\n",
    "from openai import OpenAI\n",
    "from pydantic import BaseModel,Field\n",
    "from typing import Literal\n",
    "import pandas as pd\n",
    "import spacy\n",
    "import nltk\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.corpus import stopwords\n",
    "from functools import cache\n",
    "from typing import Callable, List, Optional, Type\n",
    "from abc import ABC, abstractmethod\n",
    "from sklearn.metrics import accuracy_score\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "# Download NLTK resources\n",
    "try:\n",
    "    _create_unverified_https_context = ssl._create_unverified_context\n",
    "except AttributeError:\n",
    "    pass\n",
    "else:\n",
    "    ssl._create_default_https_context = _create_unverified_https_context\n",
    "nltk.download('punkt')\n",
    "nltk.download('stopwords')\n",
    "# Load environment variables from .env file\n",
    "load_dotenv()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e239714a-b310-4f0c-9b9e-14b521d84dff",
   "metadata": {},
   "source": [
    "# Importing necessary modules from larch:\n",
    "Larch is a LLM toolbox. The tool tentatively has the following components to create any downstream llm-based search engine.\n",
    "1. DocumentIndexer\n",
    "2. MetadataExtractor\n",
    "3. SearchEngine\n",
    "4. MetadataValidator\n",
    "5. MetadataEvaluator\n",
    "6. TextProcessor\n",
    "\n",
    "...and additionally it has more utilities for RAG pipeline.\n",
    "\n",
    "For Environmental Justice metadata extraction, the components we will use are MetadataExtractor and TextProcessor\n",
    "1. **MetadataExtractor**:\n",
    "   - **larch.metadata._base.AbstractMetadataExtractor** allows for implementing any downstream metadata extractor.\n",
    "   - **larch.metadata.extractors_openai.SimpleOpenAIMetadataExtractor**: It is barebone extractor that just hits OpenAI API based on instructor patching. No function calling.\n",
    "   - **larch.metadata.extractors_openai.InstructorBasedOpenAIMetadataExtractor** is the standard extractor that is recommended to use. This uses function-calling. However, since InstructorBasedOpenAIMetadataExtractor has validation check for schema response, we will manually create a class InstructorBasedOpenAIMetadataExtractor and implement the same method which doesnot require validation\n",
    "2. **TextProcessor**: larch.processors.TextProcessor allows for processing text. All the text processors takes in a text and egest out processed text.\n",
    "    - **larch.processors.PIIRemover** uses spacy to identify Personal Identification Information (name, email, phone number) and mask them out.\n",
    "    - **larch.processors.NonAlphaNumericRemover** removes non-alpha-numeric characters from the text\n",
    "    - **larch.processors.TextProcessingPipeline** is a container to hold all the text processors and run them sequentially.\n",
    "\n",
    "\n",
    "Further documentation can be found at : https://github.com/NASA-IMPACT/larch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0bbda35c-1dc5-4102-8b86-b6dae809f83e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import larch\n",
    "larch.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "b8b27f57-96c8-4286-bf52-384feca46319",
   "metadata": {},
   "outputs": [],
   "source": [
    "from larch.metadata import InstructorBasedOpenAIMetadataExtractor\n",
    "from larch.metadata.extractors_openai import SimpleOpenAIMetadataExtractor\n",
    "from larch.processors import PIIRemover, TextProcessingPipeline"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3de7233-d812-4906-a433-8104235e4351",
   "metadata": {},
   "source": [
    "# Further details on Metadata Extraction:\n",
    "*Pydantic*: It is a data validation and settings management library for Python. Pydantic models are Python classes that are used to define the structure and validation rules for data. These models are designed to enforce data integrity by defining the types of data expected, as well as any constraints or validation rules that should be applied.\n",
    "\n",
    "**How are we using Pydantic in our extraction process??**\n",
    "*Solution: We will define a schema based on pydantic BaseModel called Environmental Justice and define the variables based on the features we are interested in. We set up potential values of each features either by giving list of options, or some regex patterns that the field value relates to.* \n",
    "\n",
    "**Note** *Our model will give output in the designed schema format. Eventhough the format is the same as that of schema, we can't guarrantee that it will hold the regex pattern and field values defined in pydantic schema. For this, we will carryout prompt engineering to make model aware on potential patterns for each variable* \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e4363e03-6ac2-4b6f-ae31-04e93c15493b",
   "metadata": {},
   "source": [
    "# Schema Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "031fdd5d-1cb4-4ced-a314-aa8825270d77",
   "metadata": {},
   "outputs": [],
   "source": [
    "class EnvironmentalJustice(BaseModel):\n",
    "    Indicators: Literal[\"Disasters\",\"Human Dimensions\",\"Food Availability\",\"Health & Air Quality\",\"Water Availability\", \"Extreme Heat\",\"Urban Flooding\",\"Climate Change\"]\n",
    "    Geographic_Coverage: str\n",
    "    Format: str\n",
    "    Spatial_Resolution: str = Field(pattern=r'^(\\d+(\\.\\d+)? [a-zA-Z]+|varies|N/A)$')\n",
    "    Temporal_Resolution: str = Field(pattern=r'^(\\d+(\\.\\d+)? [a-zA-Z]+|N/A|varies|weekly|monthly|daily|yearly|varies-multiple datasets included|Daily < Weekly|Hourly < Daily|Weekly < Monthly|Monthly < Yearly|1 minute)$')\n",
    "    Temporal_Extent: str = Field(pattern=r'^(\\d{4}-\\d{2}-\\d{2} to present|present|\\d{4}-\\d{2}-\\d{2} (?:to|until) present|\\d{4}-\\d{2}-\\d{2} to \\d{4}-\\d{2}-\\d{2}|varies(?:- multiple datasets (?:included|available))?)$')\n",
    "    \n",
    "schema = EnvironmentalJustice"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "3399a668-3e66-4597-8343-d0a2f3d584f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "class InstructorBasedOpenAIMetadataExtractor(SimpleOpenAIMetadataExtractor):\n",
    "    \"\"\"\n",
    "    This uses `instructor` to hit the OpenAI function calling api.\n",
    "    Note: The schema for metadata should be of `Type[OpenAISchema]`.\n",
    "    \"\"\"\n",
    "\n",
    "    def _extract(self, text: str):\n",
    "        if self.debug:\n",
    "            logger.debug(f\"nchars={len(text)}\\nText :: {text}\")\n",
    "        schema = instructor.openai_schema(self.schema)\n",
    "        response = self.openai_client.chat.completions.create(\n",
    "            model=self.model,\n",
    "            temperature=0,\n",
    "            functions=[schema.openai_schema],\n",
    "            function_call={\"name\": schema.openai_schema[\"name\"]},\n",
    "            messages=self._get_messages(text),\n",
    "        )\n",
    "        return response  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0de27700",
   "metadata": {},
   "source": [
    "# Defining the text_processor_pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "d2f3f373",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Access the API key from the environment\n",
    "openai_api_key = os.getenv(\"OPENAI_API_KEY\")\n",
    "assert openai_api_key\n",
    "text_processor = TextProcessingPipeline(\n",
    "    lambda x: re.sub(r\"\\$(?=\\w|\\n|\\()\", \" \", x).strip(),\n",
    "    lambda x: re.sub(r\"\\)(?=\\w|\\n|\\()\", \" \", x).strip(),\n",
    "    lambda x: re.sub(r\"\\#(?=\\w|\\n|\\()\", \" \", x).strip(),\n",
    "    lambda x: x.replace(\"\\t\", \" \").replace(\"!\", \" \").strip(),\n",
    "    PIIRemover()\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "44bd9a98-0d83-4543-9ede-c5ed8bfe9ed5",
   "metadata": {},
   "source": [
    "# Metadata Extractions begins from here"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d111a1b8",
   "metadata": {},
   "source": [
    "# To carry out metadata extractions, we need to do the following steps:\n",
    "1. Download validation data\n",
    "2. Preprocess validation data in the desired format\n",
    "3. Metadata Extractions"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4155c0d0-0582-4e79-8aff-c1240d260e7b",
   "metadata": {},
   "source": [
    "# Downloading Validation data\n",
    "## Exploring contents of validation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "b83e694d-2d42-4a4c-ac79-e1fcf74b263d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Dataset</th>\n",
       "      <th>Indicators</th>\n",
       "      <th>Description</th>\n",
       "      <th>Description Simplified</th>\n",
       "      <th>Geographic Coverage</th>\n",
       "      <th>Format</th>\n",
       "      <th>Spatial Resolution</th>\n",
       "      <th>Spatial Resolution (Standard)</th>\n",
       "      <th>Temporal Resolution</th>\n",
       "      <th>Temporal Extent</th>\n",
       "      <th>Latency</th>\n",
       "      <th>Source/Link</th>\n",
       "      <th>Project</th>\n",
       "      <th>Strengths</th>\n",
       "      <th>Limitations</th>\n",
       "      <th>Data Visualization</th>\n",
       "      <th>Intended Use</th>\n",
       "      <th>Tab Name Indicator</th>\n",
       "      <th>Indicators                     (Select from drop-down list)</th>\n",
       "      <th>sde_link</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ABoVE: Burn Severity of Soil Organic Matter, N...</td>\n",
       "      <td>Disasters</td>\n",
       "      <td>This dataset provides maps at 30-m resolution ...</td>\n",
       "      <td>The ABoVE: Burn Severity of Soil Organic Matte...</td>\n",
       "      <td>Northwest Territories and Northern Alberta, Ca...</td>\n",
       "      <td>GeoTIFF</td>\n",
       "      <td>30 meters</td>\n",
       "      <td>30 meters</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2014-05-01 to 2015-10-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.3334/ORNLDAAC/1694</td>\n",
       "      <td>ABoVE - Arctic-Boreal Vulnerability Experiment</td>\n",
       "      <td>Pre-classified burn severity for ease of use</td>\n",
       "      <td>Lacks recent data</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Path C</td>\n",
       "      <td>Disasters</td>\n",
       "      <td>Disasters</td>\n",
       "      <td>https://sciencediscoveryengine.nasa.gov/app/na...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Daily and Annual PM2.5 Concentrations for the ...</td>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>The Daily and Annual PM2.5 Concentrations for ...</td>\n",
       "      <td>The Daily and Annual PM2.5 Concentrations for ...</td>\n",
       "      <td>United States</td>\n",
       "      <td>raster, tabular, vector</td>\n",
       "      <td>0.00833 degrees</td>\n",
       "      <td>1 kilometer</td>\n",
       "      <td>Daily; Annual</td>\n",
       "      <td>2000-01-01 to 2016-12-31</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://doi.org/10.7927/0rvr-4538</td>\n",
       "      <td>AQDH - Air Quality Data for Health-Related App...</td>\n",
       "      <td>16-year temporal extent</td>\n",
       "      <td>NaN</td>\n",
       "      <td>SEDAC Map widget</td>\n",
       "      <td>Path B</td>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>https://sciencediscoveryengine.nasa.gov/app/na...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ECCO Global Mean Sea Level - Daily Mean (Versi...</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>This dataset provides daily-averaged global me...</td>\n",
       "      <td>The ECCO (Estimating the Circulation and Clima...</td>\n",
       "      <td>Global</td>\n",
       "      <td>netCDF-4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1992-01-01 to 2018-01-01</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.5067/ECTSD-MSL44</td>\n",
       "      <td>ECCO - Estimating the Circulation and Climate ...</td>\n",
       "      <td>20-year temporal extent</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Path C</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>https://sciencediscoveryengine.nasa.gov/app/na...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ECOSTRESS Land Surface Temperature and Emissiv...</td>\n",
       "      <td>Extreme Heat</td>\n",
       "      <td>The ECOsystem Spaceborne Thermal Radiometer Ex...</td>\n",
       "      <td>The ECOSTRESS Land Surface Temperature dataset...</td>\n",
       "      <td>Global</td>\n",
       "      <td>HDF5</td>\n",
       "      <td>70 m</td>\n",
       "      <td>70 meters</td>\n",
       "      <td>Varies</td>\n",
       "      <td>2018-07-09 to Present</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.5067/ECOSTRESS/ECO2LSTE.001</td>\n",
       "      <td>ECOSTRESS - ECOsystem Spaceborne Thermal Radio...</td>\n",
       "      <td>Recent data available</td>\n",
       "      <td>Latency unclear</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Path B</td>\n",
       "      <td>Extreme Heat</td>\n",
       "      <td>Extreme Heat</td>\n",
       "      <td>https://sciencediscoveryengine.nasa.gov/app/na...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Food Insecurity Hotspots Data Set, v1 (2009 â...</td>\n",
       "      <td>Food Availability,Human Dimensions</td>\n",
       "      <td>The Food Insecurity Hotspots Data Set consists...</td>\n",
       "      <td>The Food Insecurity Hotspots Dataset shows the...</td>\n",
       "      <td>Central America, Caribbean Islands, Central As...</td>\n",
       "      <td>raster, vector, map</td>\n",
       "      <td>250 m</td>\n",
       "      <td>250 m</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2009-01-01 to 2019-12-31</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://doi.org/10.7927/cx02-2587</td>\n",
       "      <td>FOOD - Food Security Data Collection</td>\n",
       "      <td>10-year temporal extent</td>\n",
       "      <td>Limited geographic coverage</td>\n",
       "      <td>https://sedac.ciesin.columbia.edu/data/set/foo...</td>\n",
       "      <td>Path B</td>\n",
       "      <td>Human Dimensions</td>\n",
       "      <td>Human Dimensions</td>\n",
       "      <td>https://sciencediscoveryengine.nasa.gov/app/na...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             Dataset  \\\n",
       "0  ABoVE: Burn Severity of Soil Organic Matter, N...   \n",
       "1  Daily and Annual PM2.5 Concentrations for the ...   \n",
       "2  ECCO Global Mean Sea Level - Daily Mean (Versi...   \n",
       "3  ECOSTRESS Land Surface Temperature and Emissiv...   \n",
       "4  Food Insecurity Hotspots Data Set, v1 (2009 â...   \n",
       "\n",
       "                           Indicators  \\\n",
       "0                           Disasters   \n",
       "1                Health & Air Quality   \n",
       "2                      Climate Change   \n",
       "3                        Extreme Heat   \n",
       "4  Food Availability,Human Dimensions   \n",
       "\n",
       "                                         Description  \\\n",
       "0  This dataset provides maps at 30-m resolution ...   \n",
       "1  The Daily and Annual PM2.5 Concentrations for ...   \n",
       "2  This dataset provides daily-averaged global me...   \n",
       "3  The ECOsystem Spaceborne Thermal Radiometer Ex...   \n",
       "4  The Food Insecurity Hotspots Data Set consists...   \n",
       "\n",
       "                              Description Simplified  \\\n",
       "0  The ABoVE: Burn Severity of Soil Organic Matte...   \n",
       "1  The Daily and Annual PM2.5 Concentrations for ...   \n",
       "2  The ECCO (Estimating the Circulation and Clima...   \n",
       "3  The ECOSTRESS Land Surface Temperature dataset...   \n",
       "4  The Food Insecurity Hotspots Dataset shows the...   \n",
       "\n",
       "                                 Geographic Coverage                   Format  \\\n",
       "0  Northwest Territories and Northern Alberta, Ca...                  GeoTIFF   \n",
       "1                                      United States  raster, tabular, vector   \n",
       "2                                             Global                 netCDF-4   \n",
       "3                                             Global                     HDF5   \n",
       "4  Central America, Caribbean Islands, Central As...      raster, vector, map   \n",
       "\n",
       "  Spatial Resolution Spatial Resolution (Standard) Temporal Resolution  \\\n",
       "0          30 meters                     30 meters                 NaN   \n",
       "1    0.00833 degrees                   1 kilometer       Daily; Annual   \n",
       "2                NaN                           NaN                 NaN   \n",
       "3               70 m                     70 meters              Varies   \n",
       "4              250 m                         250 m                 NaN   \n",
       "\n",
       "            Temporal Extent Latency  \\\n",
       "0  2014-05-01 to 2015-10-01     NaN   \n",
       "1  2000-01-01 to 2016-12-31     NaN   \n",
       "2  1992-01-01 to 2018-01-01     NaN   \n",
       "3     2018-07-09 to Present     NaN   \n",
       "4  2009-01-01 to 2019-12-31     NaN   \n",
       "\n",
       "                                         Source/Link  \\\n",
       "0           https://dx.doi.org/10.3334/ORNLDAAC/1694   \n",
       "1                  https://doi.org/10.7927/0rvr-4538   \n",
       "2             https://dx.doi.org/10.5067/ECTSD-MSL44   \n",
       "3  https://dx.doi.org/10.5067/ECOSTRESS/ECO2LSTE.001   \n",
       "4                  https://doi.org/10.7927/cx02-2587   \n",
       "\n",
       "                                             Project  \\\n",
       "0     ABoVE - Arctic-Boreal Vulnerability Experiment   \n",
       "1  AQDH - Air Quality Data for Health-Related App...   \n",
       "2  ECCO - Estimating the Circulation and Climate ...   \n",
       "3  ECOSTRESS - ECOsystem Spaceborne Thermal Radio...   \n",
       "4               FOOD - Food Security Data Collection   \n",
       "\n",
       "                                      Strengths                  Limitations  \\\n",
       "0  Pre-classified burn severity for ease of use            Lacks recent data   \n",
       "1                       16-year temporal extent                          NaN   \n",
       "2                       20-year temporal extent                          NaN   \n",
       "3                         Recent data available              Latency unclear   \n",
       "4                       10-year temporal extent  Limited geographic coverage   \n",
       "\n",
       "                                  Data Visualization Intended Use  \\\n",
       "0                                                NaN       Path C   \n",
       "1                                   SEDAC Map widget       Path B   \n",
       "2                                                NaN       Path C   \n",
       "3                                                NaN       Path B   \n",
       "4  https://sedac.ciesin.columbia.edu/data/set/foo...       Path B   \n",
       "\n",
       "     Tab Name Indicator  \\\n",
       "0             Disasters   \n",
       "1  Health & Air Quality   \n",
       "2        Climate Change   \n",
       "3          Extreme Heat   \n",
       "4      Human Dimensions   \n",
       "\n",
       "  Indicators                     (Select from drop-down list)  \\\n",
       "0                                          Disasters            \n",
       "1                               Health & Air Quality            \n",
       "2                                     Climate Change            \n",
       "3                                       Extreme Heat            \n",
       "4                                   Human Dimensions            \n",
       "\n",
       "                                            sde_link  \n",
       "0  https://sciencediscoveryengine.nasa.gov/app/na...  \n",
       "1  https://sciencediscoveryengine.nasa.gov/app/na...  \n",
       "2  https://sciencediscoveryengine.nasa.gov/app/na...  \n",
       "3  https://sciencediscoveryengine.nasa.gov/app/na...  \n",
       "4  https://sciencediscoveryengine.nasa.gov/app/na...  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "validation_data=pd.read_csv(\"../data/EJ_validation_data.csv\",encoding='latin1')\n",
    "validation_data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84ce8056-dc21-4231-bac1-10e0c9ec252f",
   "metadata": {},
   "source": [
    "# Preprocessing the validation data in the desired format. This includes the following steps:\n",
    "1. Only considering those urls which have doi in it\n",
    "2. Selecting the features ['Geographic Coverage\",\"Indicators\", \"Format\", \"Spatial Resolution\", \"Temporal Resolution\", \"Temporal Extent\"] for analysis\n",
    "3. We extract text based response from request for url and preprocess it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "99619be3-2817-48e2-98ea-8ae70c219fc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_data={}\n",
    "import requests\n",
    "for each_data in validation_data.iterrows():\n",
    "    if \"doi\" in each_data[1].iloc[11]:\n",
    "        final_data[each_data[1].iloc[11]]={\"Indicators\":each_data[1].iloc[1],\n",
    "                  \"Geographic_Coverage\":each_data[1].iloc[4],\n",
    "                  \"Format\":each_data[1].iloc[5],\n",
    "                  \"Spatial_Resolution\":each_data[1].iloc[7],\n",
    "                  \"Temporal_Resolution\":each_data[1].iloc[8],\n",
    "                  \"Temporal_Extent\":each_data[1].iloc[9]}\n",
    "url_lists=list(final_data.keys())\n",
    "url_content=[]\n",
    "urls=[]\n",
    "for url in url_lists:  \n",
    "    response=requests.get(url)\n",
    "    html_page = response.text\n",
    "    soup = BeautifulSoup(html_page, \"html.parser\")\n",
    "    text=soup.get_text()\n",
    "    text=text.lower()\n",
    "    # Tokenization\n",
    "    tokens = word_tokenize(text) \n",
    "    # Remove punctuation and make lowercase\n",
    "    tokens = [word.lower() for word in tokens if word.isalnum()]\n",
    "    # Remove stopwords\n",
    "    stop_words = set(stopwords.words('english'))\n",
    "    tokens = [word for word in tokens if word not in stop_words]\n",
    "    text=\" \".join(tokens)\n",
    "    text = re.sub(r'[\\t\\n\\r\\f\\v]+', '', text)\n",
    "    text=re.sub(r'[^\\w\\s]', '', text)\n",
    "    url_content.append(text)\n",
    "    urls.append(url.strip())\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b8e47b9a-23d9-4c99-83e1-1a31d791b4ae",
   "metadata": {},
   "source": [
    "# Initial Extractions\n",
    "## This is carried out using gpt-3.5-turbo-0125 model: \n",
    "**gpt-3.5-turbo-0125 model details**\n",
    "*It is the latest GPT-3.5 Turbo model with higher accuracy at responding in requested formats and a fix for a bug which caused a text encoding issue for non-English language function calls. It returns a maximum of 4,096 output tokens. The context window is upto 16,385 tokens.*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "4b32d142-ea4a-44ea-ace9-55c7296bf5d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "SYSTEM_PROMPT =\"Extract metadata and entities details accurately from my requests.\" + \"Metadata extraction for Indicators should clearly be one of the element in the list ['Disasters','Human Dimensions','Food Availability','Health & Air Quality','Water Availability', 'Extreme Heat','Urban Flooding','Climate Change']. Do not create new element for indicators \" + \"Geographic_Coverage: If multiple countries, write global, else write name of location\"+ \"Make sure the following fields follow the following regex pattern\"+\"Spatial_Resolution: ^(\\d+(\\.\\d+)? [a-zA-Z]+|varies|N/A)$\"+\"Temporal_Resolution: ^(\\d+(\\.\\d+)? [a-zA-Z]+|N/A|varies|weekly|monthly|daily|yearly|varies-multiple datasets included|Daily < Weekly|Hourly < Daily|Weekly < Monthly|Monthly < Yearly|1 minute)$\"+\"Temporal_Extent: ^(\\d{4}-\\d{2}-\\d{2} to present|present|\\d{4}-\\d{2}-\\d{2} (?:to|until) present|\\d{4}-\\d{2}-\\d{2} to \\d{4}-\\d{2}-\\d{2}|varies(?:- multiple datasets (?:included|available))?)$.\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "648f3a58-fa2f-40e5-b3e8-2a91cab313a7",
   "metadata": {},
   "outputs": [],
   "source": [
    "metadata_extractor = InstructorBasedOpenAIMetadataExtractor(\n",
    "    model=\"gpt-3.5-turbo-0125\",\n",
    "    schema=schema,\n",
    "    preprocessor=text_processor,system_prompt=SYSTEM_PROMPT,\n",
    "    debug=False,\n",
    ")\n",
    "before_finetuning={}  # It will hold results for all the data for before finetuning extractions\n",
    "for enum,text in enumerate(url_content):\n",
    "    split=text.split()\n",
    "    if len(split)>1900:\n",
    "        text=\" \".join(split[0:1900])  #handles cases when the text tokens exceeds token limit of used model\n",
    "    result = metadata_extractor(text)\n",
    "    message=result.choices[0].message.function_call.arguments\n",
    "    metadata=json.loads(message)\n",
    "    before_finetuning[urls[enum]]=metadata"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7f7e943-0ff0-43b5-b1b0-ed4ee4f25469",
   "metadata": {},
   "source": [
    "## Extracting EJ metadata on Finetuned model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "a96cf723-ae19-4d8f-afb3-6747c3447c34",
   "metadata": {},
   "outputs": [],
   "source": [
    "metadata_extractor = InstructorBasedOpenAIMetadataExtractor(\n",
    "    model=\"ft:gpt-3.5-turbo-0125:nasa::95blPXsF\",system_prompt=SYSTEM_PROMPT,\n",
    "    schema=schema,\n",
    "    preprocessor=text_processor,\n",
    "    debug=False,\n",
    ")\n",
    "after_finetuning={} # It will hold results for all the data for after finetuning extractions\n",
    "for enum,text in enumerate(url_content):\n",
    "    split=text.split()\n",
    "    if len(split)>1900:\n",
    "        text=\" \".join(split[0:1900]) #handles cases when the text tokens exceeds token limit\n",
    "    result = metadata_extractor(text)\n",
    "    message=result.choices[0].message.function_call.arguments\n",
    "    metadata=json.loads(message)\n",
    "    after_finetuning[urls[enum]]=metadata\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a772468e",
   "metadata": {},
   "source": [
    "## Collecting true data in a desirable format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "3d7f3d2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "true_data = {}\n",
    "for i, row in validation_data.iterrows():\n",
    "    source_link_col = [\"Source/Link\",\"Indicators\",\"Geographic Coverage\",\"Format\",\"Spatial Resolution\",\"Spatial Resolution (Standard)\",\"Temporal Resolution\",\"Temporal Extent\"]\n",
    "    alternate_vals=row[source_link_col]\n",
    "    alternate_vals=list(alternate_vals)\n",
    "    alternate_values=alternate_vals[1:]\n",
    "    value={}\n",
    "    for enum,i in enumerate(source_link_col[1:]):\n",
    "            value[i]=alternate_values[enum]\n",
    "    true_data[alternate_vals[0]]=value\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "290b3a35-4096-46b8-90a6-6dd160c3f39d",
   "metadata": {},
   "source": [
    "## Script preparation for evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "43ffc70c",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# Evaluation Metric: will be used for indicators field\n",
    "def evaluate_exact_match(predicted, ground_truth):\n",
    "    predicted=predicted.lower()\n",
    "    ground_truth=ground_truth.lower()\n",
    "    if int(predicted == ground_truth)==1:\n",
    "        return 1\n",
    "    else:\n",
    "        #in case of multiple indicators, if one is correct, then it is true prediction\n",
    "        predicted_indicators=set(predicted.split(\",\"))\n",
    "        ground_truth_indicators=set(ground_truth.split(\",\"))\n",
    "        if predicted_indicators.intersection(ground_truth_indicators):\n",
    "            return 1\n",
    "    return 0\n",
    "                                 \n",
    "\n",
    "# Evaluation Metric: Exact Match or at least one match if multiple\n",
    "def evaluate_format(predicted, ground_truth):\n",
    "    predicted = np.nan if predicted==\"N/A\" else predicted\n",
    "    if type(predicted)==float and type(ground_truth)==float:\n",
    "        return 1  #returning true if both true data and predicted data do not have information about it\n",
    "    elif type(ground_truth)==float:\n",
    "        return 0\n",
    "    elif type(predicted)==float:\n",
    "        return 0\n",
    "    predicted=predicted.lower()\n",
    "    ground_truth=ground_truth.lower()\n",
    "    if predicted==ground_truth:\n",
    "        return 1\n",
    "    #in case of multiple Formats, if one is correct, then it is true prediction\n",
    "    predicted_formats = set(predicted.split(\",\"))\n",
    "    ground_truth_formats = set(ground_truth.split(\",\"))\n",
    "    if predicted_formats.intersection(ground_truth_formats):\n",
    "        return 1\n",
    "    return 0\n",
    "\n",
    "def evaluate_spatial_resolution(predicted, ground_truth1,ground_truth2):\n",
    "    \"\"\"\n",
    "        We have two features for spatial resolution: Spatial Resolution and Spatial Resolution (Standard).\n",
    "        Here we are evaluating considering commonality of predicted data with both the ground truths\n",
    "    \"\"\"\n",
    "    predicted = np.nan if predicted==\"N/A\" else predicted\n",
    "    if (type(predicted)==float and type(ground_truth1)==float) or (type(predicted)==float and type(ground_truth2)==float) or (type(ground_truth1)==float and type(ground_truth2)==float):\n",
    "        return 1 #returning true if both true data and predicted data do not have information about it\n",
    "    elif (type(ground_truth1)==float or type(ground_truth2)==float):\n",
    "        return 0\n",
    "    elif type(predicted)==float:\n",
    "        return 0\n",
    "    predicted=predicted.lower()\n",
    "    ground_truth1=ground_truth1.lower()\n",
    "    ground_truth2=ground_truth2.lower()\n",
    "    if (predicted==ground_truth1) or (predicted==ground_truth2):\n",
    "        return 1\n",
    "    # if the above condition fails, we aim to check on only the numerical representation of the true and predicted data and check whether these values are correct or not\n",
    "    predicted = re.sub(r'[a-zA-Z\\(\\)\\[\\]]', '', predicted).strip()\n",
    "    ground_truth1=re.sub(r'[a-zA-Z\\(\\)\\[\\]]', '', ground_truth1).strip()\n",
    "    ground_truth2=re.sub(r'[a-zA-Z\\(\\)\\[\\]]', '', ground_truth2).strip()\n",
    "    if not (ground_truth1 or ground_truth2) and predicted.find(',') != -1:\n",
    "        return 1\n",
    "    ground_truth1=re.sub(r'[^\\d.]', '', ground_truth1)\n",
    "    ground_truth2=re.sub(r'[^\\d.]', '', ground_truth2)\n",
    "    predicted=re.sub(r'[^\\d.]', '', predicted)\n",
    "    if predicted==ground_truth1 or predicted==ground_truth2:\n",
    "        return 1\n",
    "    if predicted and ((ground_truth1 and (predicted.replace(\" \",\"\")==ground_truth1.replace(\" \",\"\"))) or (ground_truth2 and (predicted.replace(\" \",\"\")==ground_truth2.replace(\" \",\"\")))):\n",
    "        return 1\n",
    "    return 0\n",
    "\n",
    "\n",
    "def evaluate_temporal_resolution_extent(predicted,ground_truth):\n",
    "    predicted = np.nan if predicted==\"N/A\" else predicted\n",
    "    ground_truth=np.nan if (type(ground_truth)==str and ground_truth.lower()==\"nan\") else ground_truth\n",
    "    if type(predicted)==float and type(ground_truth)==float:\n",
    "        return 1  #returning true if both true data and predicted data do not have information about it\n",
    "    if type(ground_truth)==float:\n",
    "        return 0\n",
    "    elif type(predicted)==float:\n",
    "        return 0\n",
    "    predicted=predicted.lower()\n",
    "    ground_truth=str(ground_truth).lower()\n",
    "    if predicted==ground_truth:\n",
    "        return 1\n",
    "    predicted = ''.join(re.findall(r'\\w', predicted))\n",
    "    ground_truth=''.join(re.findall(r'\\w', ground_truth))\n",
    "    if predicted==ground_truth:\n",
    "        return 1 \n",
    "    return 0    "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "67630598-4a03-4874-aa85-2c27431c9fa4",
   "metadata": {},
   "source": [
    "# Making sure our true data, before_finetuned extractions, after finetuned extractions have the same source of link"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "1f150f27",
   "metadata": {},
   "outputs": [],
   "source": [
    "truth_data={}\n",
    "for i, value in true_data.items():\n",
    "    if not isinstance(i,float):\n",
    "        truth_data[i.strip()]=value \n",
    "true_data=truth_data\n",
    "intersection_keys = set(before_finetuning.keys()) & set(after_finetuning.keys()) & set(true_data.keys())\n",
    "before_finetuned={k:before_finetuning[k] for k in intersection_keys}\n",
    "after_finetuned={k:after_finetuning[k] for k in intersection_keys}\n",
    "true_data={k:true_data[k] for k in intersection_keys}\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5557c165-4233-4af6-bce2-6dfe3fa9a131",
   "metadata": {},
   "source": [
    "## Defining the extraction performance in boolean field: 0 ==> wrong prediction, 1 ==> correct prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "22aad265",
   "metadata": {},
   "outputs": [],
   "source": [
    "def boolean_metrics(true_data,pred_data):\n",
    "    metrics={}\n",
    "    for key in true_data.keys():\n",
    "        reject_key=[np.nan]\n",
    "        if key not in reject_key:\n",
    "            true_source=true_data[key]\n",
    "            try:\n",
    "                pred_source=pred_data[key]\n",
    "            except:\n",
    "                pred_source=pred_data[key.strip()]  #removing whitespace characters\n",
    "            for each_key in pred_source.keys():\n",
    "                if each_key==\"Indicators\":\n",
    "                    result=evaluate_exact_match(pred_source[each_key],true_source[each_key])\n",
    "                    if each_key not in metrics.keys():\n",
    "                            metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result)\n",
    "                elif each_key==\"Geographic_Coverage\":\n",
    "                    result=evaluate_format(pred_source[each_key],true_source[\"Geographic Coverage\"])\n",
    "                    if each_key not in metrics.keys():\n",
    "                        metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result) \n",
    "                        \n",
    "                elif each_key==\"Format\":\n",
    "                    result=evaluate_format(pred_source[each_key],true_source[each_key])\n",
    "                    if each_key not in metrics.keys():\n",
    "                        metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result)\n",
    "                elif each_key==\"Spatial_Resolution\":\n",
    "                    result=evaluate_spatial_resolution(pred_source[each_key],true_source[\"Spatial Resolution\"],true_source[\"Spatial Resolution (Standard)\"])\n",
    "                    if each_key not in metrics.keys():\n",
    "                        metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result)\n",
    "                elif each_key==\"Temporal_Resolution\":\n",
    "                    result=evaluate_temporal_resolution_extent(pred_source[each_key],true_source[\"Temporal Resolution\"])\n",
    "                    if each_key not in metrics.keys():\n",
    "                        metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result)\n",
    "                elif each_key==\"Temporal_Extent\":\n",
    "                    result=evaluate_temporal_resolution_extent(pred_source[each_key],true_source[\"Temporal Extent\"])\n",
    "                    if each_key not in metrics.keys():\n",
    "                        metrics[each_key]=[result]\n",
    "                    else:\n",
    "                        metrics[each_key].append(result)\n",
    "    return metrics\n",
    "\n",
    "# We will measure accuracy based on obtained boolean result\n",
    "def get_metrics(predictions):\n",
    "    aspect_metrics = {}\n",
    "    for aspect, preds in predictions.items():\n",
    "        accuracy = accuracy_score(preds, [1]*len(preds))  # Assuming ground truth is always 1\n",
    "        aspect_metrics[aspect] = accuracy\n",
    "    return aspect_metrics"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "726525d4-6b2a-4ad7-bcd7-2c0b636ba52f",
   "metadata": {},
   "source": [
    "## Extracting accuracy metrics on boolean evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "9b958cd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "before_metrics=boolean_metrics(true_data,before_finetuned)\n",
    "metrics_before_finetuning=get_metrics(before_metrics)\n",
    "\n",
    "after_metrics=boolean_metrics(true_data,after_finetuned)\n",
    "metrics_after_finetuning=get_metrics(after_metrics)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4eea83f0-573d-4270-aba6-7afaab655222",
   "metadata": {},
   "source": [
    "# At this rate, we have following set of informations:\n",
    "1. True data\n",
    "2. Metadata Extraction before finetuning\n",
    "3. Metadata Extraction after finetuning\n",
    "   \n",
    "**But What if our defined evaluation system for generated prediction is not enough? What if we are gettting correct result and our system is failing to identify it as correct?**\n",
    "\n",
    "*Solution1: Lets ask gpt based model to evaluate the performance of features of extracted metadata before and after finetuning by comparing with the ground truth.*\n",
    "\n",
    "**How are we going to do this?**\n",
    "\n",
    "*By implementing prompt based gpt evaluation for true and extracted metadata. The prompt implementation is given below:*\n",
    "\n",
    "\n",
    "*Solution2: Lets perform human evaluation on extracted data before and after finetuning and true data*\n",
    "\n",
    "**How are we going to do this?**\n",
    "\n",
    "*We will dump the ground truth and extracted data in a desirable spreasheet format and let human manually define whether its correct or not and then do the analysis*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d606fb2a",
   "metadata": {},
   "source": [
    "## Implementing prompt based gpt evaluation for true and extracted metadata"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "88c87ad8",
   "metadata": {},
   "outputs": [],
   "source": [
    "prompt=\"\"\"You are given a dictionary of true data and its prediction. Based on the information, give boolean result for each \n",
    "feature in the data if its true data and prediction are same or nearly same.\\\n",
    "Evaluation criteria are as follows: \n",
    "Indicators: if true data and predicted indicator are same, set 1, if one of the predicted indicator is same, set 1, else set 0\n",
    "Geographic_Coverage: if same, set 1, elif multiple countries in one and global in another, set 1, if both nan set 1 else 0\n",
    "Format: if true format and predicted format are same, set 1, elif few samples are similar, set 1, if both nan the set 1 else set 0\n",
    "Spatial Resolution: if true spatial resolution and predicted spatial resolution are similar or same, set 1, if both are nan set 1 else set 0\n",
    "Temporal Resolution: if true temporal resolution and predicted temporal resolution are similar or same, set 1, if both are nan set 1 else set 0\n",
    "Output format:\n",
    "{given url: {\"Indicators\": 0 or 1,\n",
    "\"Geographic_Coverage\": 0 or 1,\n",
    "\"Format\": 0 or 1,\n",
    "\"Spatial_Resolution\": 0 or 1,\n",
    "\"Temporal_Resolution\": 0 or 1,\n",
    "\"Temporal_Extent\":0 or 1',\n",
    "}}\n",
    "\"\"\"\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "257b0077-c1b9-41dc-bcb5-cef9f28e5554",
   "metadata": {},
   "source": [
    "## Evaluating the prompt based model performance by feeding true data, predictions of before and after finetuning on gpt-3.5-turbo model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "7b1de18a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import openai\n",
    "openai.api_key=openai_api_key\n",
    "# Get the iterators for each dictionary\n",
    "iter1 = iter(true_data.items())\n",
    "iter2 = iter(before_finetuned.items())\n",
    "iter3 = iter(after_finetuned.items())\n",
    "before_data=[]\n",
    "after_data=[]\n",
    "openai_client = OpenAI(api_key=os.environ.get(\"OPENAI_API_KEY\"))\n",
    "while True:\n",
    "    # Get the next item from each dictionary\n",
    "    try:\n",
    "        true = next(iter1)\n",
    "        before = next(iter2)\n",
    "        after = next(iter3)\n",
    "        response_before = openai_client.chat.completions.create(\n",
    "        model='gpt-3.5-turbo',\n",
    "        temperature=0,\n",
    "        messages=[{\"role\": \"system\", \"content\": prompt,},{\"role\":\"user\",\"content\":\"Prediction: \"+str(before)+\"True data: \"+str(true)}])\n",
    "        before_data.append(response_before)\n",
    "        \n",
    "        response_after = openai_client.chat.completions.create(\n",
    "        model='gpt-3.5-turbo',\n",
    "        temperature=0,\n",
    "        messages=[{\"role\": \"system\", \"content\": prompt,},{\"role\":\"user\",\"content\":\"Prediction: \"+str(after)+\"True data: \"+str(true)}])\n",
    "        after_data.append(response_after)\n",
    "    except StopIteration:\n",
    "        break"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce96fac0-3747-4f50-a321-7afb3bcdc5d0",
   "metadata": {},
   "source": [
    "## preprocessing the gpt evaluation in desirable format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "f4ddbcc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "gpt_before_metrics={}\n",
    "for enum, response_before in enumerate(before_data):\n",
    "    data=response_before.choices[0].message.content\n",
    "    data=data.replace(\"'\",'\"').replace(\"True\",\"1\").replace(\"False\",\"0\").replace(\",\\n}\",\"}\")\n",
    "    data=json.loads(data)\n",
    "    url=list(data.keys())\n",
    "    values=list(data.values())\n",
    "    \n",
    "    gpt_before_metrics[url[0]]=values[0]\n",
    "\n",
    "\n",
    "gpt_after_metrics={}\n",
    "for enum, response_after in enumerate(after_data):\n",
    "    data=response_after.choices[0].message.content\n",
    "    data=data.replace(\"'\",'\"').replace(\"True\",\"1\").replace(\"False\",\"0\").replace(\",\\n}\",\"}\")\n",
    "    data=json.loads(data)\n",
    "    url=list(data.keys())\n",
    "    values=list(data.values())\n",
    "    gpt_after_metrics[url[0]]=values[0]\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f54ccb50-a0e9-43e9-9f25-fb5ce4eb00d0",
   "metadata": {},
   "source": [
    "## converting gpt evaluation in desirable format for measuring accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "0e5f85d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_gpt_before_metrics = {}\n",
    "# Iterate through the dictionary\n",
    "for url, attributes in gpt_before_metrics.items():\n",
    "    for key, value in attributes.items():\n",
    "        if key not in final_gpt_before_metrics:\n",
    "            final_gpt_before_metrics[key] = []\n",
    "        final_gpt_before_metrics[key].append(value)\n",
    "\n",
    "# Convert to the desired format\n",
    "final_gpt_before_metrics = {key: value for key, value in final_gpt_before_metrics.items()}\n",
    "\n",
    "final_gpt_after_metrics = {}\n",
    "# Iterate through the dictionary\n",
    "for url, attributes in gpt_after_metrics.items():\n",
    "    for key, value in attributes.items():\n",
    "        if key not in final_gpt_after_metrics:\n",
    "            final_gpt_after_metrics[key] = []\n",
    "        final_gpt_after_metrics[key].append(value)\n",
    "\n",
    "# Convert to the desired format\n",
    "finals_gpt_after_metrics = {key: value for key, value in final_gpt_after_metrics.items()}\n",
    "\n",
    "finals_gpt_before_metrics=get_metrics(final_gpt_before_metrics)\n",
    "finals_gpt_after_metrics=get_metrics(final_gpt_after_metrics)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6a05ae4-73f5-40c2-83f7-8115bc2f5214",
   "metadata": {},
   "source": [
    "## Comparing the performance of model before and after finetuning using defined evaluation metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "97d68129",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA90AAAJOCAYAAACqS2TfAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/H5lhTAAAACXBIWXMAAA9hAAAPYQGoP6dpAACuNklEQVR4nOzdeVhU5f//8deAAoKCCwouKIorLljgWqYW5kKWpmmaqVhmJZWSlaa5lvTJMq00stTMJa00My2XLDWX3LXFJfctRdzAFQTu3x/+OF9HFlEZJu35uK65au45y/sMZ8Z5nXOf+9iMMUYAAAAAACDXuTi7AAAAAAAA7lSEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgC4w9lsNg0dOtTZZdyyqVOnqmrVqsqfP78KFy7s7HJu2NChQ2Wz2Zxdxi2Li4tT+/btVaxYMdlsNo0ZM8bZJeWp7t27KzAwMFeXee7cOT399NPy9/eXzWZTnz59cnX5uWXZsmWy2WxatmyZs0vJNZ9//rlsNpv279/v7FIA3MEI3QDueHv27FGvXr1UoUIFeXh4yNvbW/fcc4/Gjh2rixcvOrs85MCOHTvUvXt3BQUF6dNPP9WECROynDY93Lq4uOjQoUMZXk9MTFSBAgVks9kUFRV1U/WMHDlSc+fOval580qTJk1ks9msh5ubm8qXL69nnnkm0/clp/r27atFixZpwIABmjp1qlq0aJGLVd+Z6tatK5vNpo8//jjT10eOHKnPP/9czz33nKZOnaonn3xSq1ev1tChQ3XmzJk8qzP9s5PZIzY2Ns/quB0+XwBwI/I5uwAAcKQFCxbosccek7u7u7p27aoaNWooOTlZK1eu1CuvvKK//vor2wB3J7h48aLy5bu9v+6XLVumtLQ0jR07VhUrVszRPO7u7vryyy/16quv2rXPmTPnlusZOXKk2rdvrzZt2uR4nkGDBql///63vO4bUaZMGcXExEiSkpOTtW3bNsXGxmrRokXavn27PD09b3iZP//8sx555BH169cvt8u9I+3atUvr169XYGCgpk+frueeey7DND///LPq16+vIUOGWG3vvvuuhg0bpu7du+d5z46PP/5YBQsWtGurV6+egoKCdPHiRbm5uTl0/Tfz+bpZTz75pB5//HG5u7s7fF0A/rtu719hAJCNffv26fHHH1e5cuX0888/q2TJktZrvXv31u7du7VgwQInVug4aWlpSk5OloeHhzw8PJxdzi07fvy4JN1Q+GjVqlWmoXvGjBmKiIjQ7Nmzc7PELJ0/f15eXl7Kly9fnh/88PHxUZcuXezaypcvr6ioKK1atUrNmjW74WUeP348V0PgpUuX5ObmJheXO7Pz3bRp01SiRAm99957at++vfbv35+he/rx48cVHBycJ/VcuHDhugdb2rdvL19f30xfuxO+T67m6uoqV1dXZ5cB4A53Z/4LBwCS3nnnHZ07d04TJ060C9zpKlasqJdeesl6npKSohEjRigoKEju7u4KDAzU66+/rqSkJLv5AgMD9dBDD2nZsmUKCwtTgQIFVLNmTes6xzlz5qhmzZry8PBQaGioNm/ebDd/9+7dVbBgQe3du1fNmzeXl5eXSpUqpeHDh8sYYzftu+++q4YNG6pYsWIqUKCAQkND9c0332TYlvSu0tOnT1f16tXl7u6uhQsXWq9dfU332bNn1adPHwUGBsrd3V0lSpRQs2bNtGnTJrtlfv311woNDVWBAgXk6+urLl266MiRI5luy5EjR9SmTRsVLFhQxYsXV79+/ZSamprFX8be+PHjrZpLlSql3r1723WpDQwMtM4AFi9ePMfXqHfu3FlbtmzRjh07rLZjx47p559/VufOnTOdJykpSUOGDFHFihXl7u6ugIAAvfrqq3b7gM1m0/nz5zVlyhSr62337t0l/V/33G3btqlz584qUqSI7r33XrvXrjVt2jTVrVtXnp6eKlKkiO677z4tXrzYen3Dhg1q3ry5fH19VaBAAZUvX149evS47vZnxd/fX5IyHAA4cuSIevToIT8/P7m7u6t69eqaNGmS9Xr6ta/GGI0bN87a9nR79+7VY489pqJFi8rT01P169fPcFAr/ZrgmTNnatCgQSpdurQ8PT2VmJgoSVq7dq1atGghHx8feXp6qnHjxlq1atV1tyk5OVmDBw9WaGiofHx85OXlpUaNGumXX36xm27//v2y2Wx69913NWHCBOuzXqdOHa1fvz7DcufOnasaNWrIw8NDNWrU0LfffnvdWq41Y8YMtW/fXg899JB8fHw0Y8aMDO/Hvn37tGDBArv96ZVXXpF05SBJevvV1x1PmzbN+nwWLVpUjz/+eIbLBpo0aaIaNWpo48aNuu++++Tp6anXX3/9hrfh2nqvvqY7fR3btm1T06ZN5enpqdKlS+udd97JMP+tfr6yup4+s89W+ndi+t8wfZ9O/15Ml9k13enf8StXrlTdunXl4eGhChUq6Isvvsiw7t9//12NGzdWgQIFVKZMGb355puaPHky14kDsMOZbgB3rO+//14VKlRQw4YNczT9008/rSlTpqh9+/Z6+eWXtXbtWsXExGj79u0Zfmzv3r1bnTt3Vq9evdSlSxe9++67at26tWJjY/X666/r+eeflyTFxMSoQ4cO2rlzp92ZvNTUVLVo0UL169fXO++8o4ULF2rIkCFKSUnR8OHDrenGjh2rhx9+WE888YSSk5M1c+ZMPfbYY5o/f74iIiLsavr555/11VdfKSoqSr6+vlkO9vTss8/qm2++UVRUlIKDg3Xy5EmtXLlS27dv19133y3pyg/RyMhI1alTRzExMYqLi9PYsWO1atUqbd682e5MZ2pqqpo3b6569erp3Xff1U8//aT33ntPQUFBmXalvdrQoUM1bNgwhYeH67nnntPOnTv18ccfa/369Vq1apXy58+vMWPG6IsvvtC3335rdXutVavWdf+e9913n8qUKaMZM2ZY7+msWbNUsGDBDO+ddKV3wMMPP6yVK1fqmWeeUbVq1fTHH3/o/fff199//21dYzp16lQ9/fTTqlu3rp555hlJUlBQkN2yHnvsMVWqVEkjR47McCDlasOGDdPQoUPVsGFDDR8+XG5ublq7dq1+/vlnPfjggzp+/LgefPBBFS9eXP3791fhwoW1f//+HHeRT01N1YkTJyRJly9f1vbt263Qc88991jTxcXFqX79+lZQKV68uH788Uc99dRTSkxMVJ8+fXTfffdZ1xs3a9ZMXbt2tZu/YcOGunDhgl588UUVK1ZMU6ZM0cMPP6xvvvlGbdu2tatrxIgRcnNzU79+/ZSUlCQ3Nzf9/PPPatmypUJDQzVkyBC5uLho8uTJuv/++/Xrr7+qbt26WW5nYmKiPvvsM3Xq1Ek9e/bU2bNnNXHiRDVv3lzr1q1T7dq17aafMWOGzp49q169eslms+mdd97Ro48+qr179yp//vySpMWLF6tdu3YKDg5WTEyMTp48qcjISJUpUyZH77105SDC7t27NXnyZLm5uenRRx/V9OnTreBbrVo1TZ06VX379lWZMmX08ssvS5Jq1qyp5ORkffnll3r//fets87FixeXJL311lt644031KFDBz399NOKj4/Xhx9+qPvuuy/D5/PkyZNq2bKlHn/8cXXp0kV+fn7XrfvUqVN2z11dXVWkSJEspz99+rRatGihRx99VB06dNA333yj1157TTVr1lTLli0l5e7nK6dWrlypOXPm6Pnnn1ehQoX0wQcfqF27djp48KCKFSuW7by7d+9W+/bt9dRTT6lbt26aNGmSunfvrtDQUFWvXl3SlQNVTZs2lc1m04ABA+Tl5aXPPvuMruoAMjIAcAdKSEgwkswjjzySo+m3bNliJJmnn37arr1fv35Gkvn555+ttnLlyhlJZvXq1VbbokWLjCRToEABc+DAAav9k08+MZLML7/8YrV169bNSDIvvPCC1ZaWlmYiIiKMm5ubiY+Pt9ovXLhgV09ycrKpUaOGuf/+++3aJRkXFxfz119/Zdg2SWbIkCHWcx8fH9O7d+8s34vk5GRTokQJU6NGDXPx4kWrff78+UaSGTx4cIZtGT58uN0y7rrrLhMaGprlOowx5vjx48bNzc08+OCDJjU11Wr/6KOPjCQzadIkq23IkCFGkt17k5Wrp+3Xr5+pWLGi9VqdOnVMZGSkMebK+3L1+zB16lTj4uJifv31V7vlxcbGGklm1apVVpuXl5fp1q1bluvu1KlTlq+l27Vrl3FxcTFt27a1235jruwPxhjz7bffGklm/fr1193uazVu3NhIyvCoVq2a2bt3r920Tz31lClZsqQ5ceKEXfvjjz9ufHx87PbDa983Y4zp06ePkWT33p09e9aUL1/eBAYGWtv3yy+/GEmmQoUKdstMS0szlSpVMs2bN7e23Zgr+3/58uVNs2bNst3WlJQUk5SUZNd2+vRp4+fnZ3r06GG17du3z0gyxYoVM6dOnbLav/vuOyPJfP/991Zb7dq1TcmSJc2ZM2estsWLFxtJply5ctnWky4qKsoEBARY25Q+/+bNm+2mK1eunImIiLBrGzVqlJFk9u3bZ9e+f/9+4+rqat566y279j/++MPky5fPrj19H4iNjc1Rven76LWP9O1N//td/X2Wvo4vvvjCaktKSjL+/v6mXbt2VltufL66deuW6Xt/7WfLmCv7qZubm9m9e7fVtnXrViPJfPjhh1bb5MmTM7zP6d/xK1assNqOHz9u3N3dzcsvv2y1vfDCC8Zms9n9PU+ePGmKFi2a6d8OwH8X3csB3JHSu6sWKlQoR9P/8MMPkqTo6Gi79vQzT9d2kw0ODlaDBg2s5/Xq1ZMk3X///SpbtmyG9r1792ZY59UjZ6efYUxOTtZPP/1ktRcoUMD6/9OnTyshIUGNGjXK0BVckho3bpyj60ILFy6stWvX6p9//sn09Q0bNuj48eN6/vnn7a7fjIiIUNWqVTO9Dv7ZZ5+1e96oUaNMt/lqP/30k5KTk9WnTx+7XgA9e/aUt7d3rlxv37lzZ+3evVvr16+3/ptV1/Kvv/5a1apVU9WqVXXixAnrcf/990tShq7K2bn2/cjM3LlzlZaWpsGDB2e4njm9q2z6Gcv58+fr8uXLOV5/usDAQC1ZskRLlizRjz/+qDFjxighIUEtW7ZUfHy8JMkYo9mzZ6t169Yyxthte/PmzZWQkJDp/na1H374QXXr1rW60ktSwYIF9cwzz2j//v3atm2b3fTdunWz27e3bNmiXbt2qXPnzjp58qS1/vPnz+uBBx7QihUrlJaWluX6XV1drcG90tLSdOrUKaWkpCgsLCzT2jt27Gh35rZRo0aS/u9zevToUW3ZskXdunWTj4+PNV2zZs1yfO11SkqKZs2apY4dO1p/z/vvv18lSpTQ9OnTc7SMzMyZM0dpaWnq0KGD3d/K399flSpVyrCfuru7KzIy8obWMXv2bGu/WbJkyXXrLViwoN3YAW5ubqpbt67dd0Bufr5yKjw83O4sea1ateTt7X3d7ybpynd8+n4hXellUKVKFbt5Fy5cqAYNGtj1pChatKieeOKJ3NkAAHcMupcDuCN5e3tLunL9ck4cOHBALi4uGUbG9vf3V+HChXXgwAG79quDtSTrh3lAQECm7adPn7Zrd3FxUYUKFezaKleuLEl21wHOnz9fb775prZs2ZLhusdrlS9fPsvtu9o777yjbt26KSAgQKGhoWrVqpW6du1q1ZO+rVWqVMkwb9WqVbVy5Uq7Ng8PD6vba7oiRYpk2OZrZbUeNzc3VahQIcN7fjPuuusuVa1aVTNmzFDhwoXl7+9v/ci/1q5du7R9+/YM25IufTC3nMjJ32LPnj1ycXHJNsQ1btxY7dq107Bhw/T++++rSZMmatOmjTp37pyjLqxeXl4KDw+3nrdo0UL33nuvwsLC9Pbbb+u9995TfHy8zpw5owkTJmQ5kv/1tv3AgQPWAaarVatWzXq9Ro0aVvu178+uXbskXQnjWUlISMi2i/OUKVP03nvvaceOHXYHKDL7W1z7+U1fbvo+m77vVapUKcO8VapUue5BCOlK9/T4+HjVrVtXu3fvttqbNm2qL7/8Uv/73/9uavC4Xbt2yRiTaW2SrO7x6UqXLn3Do43fd999WQ6klpkyZcpk+E4qUqSIfv/9d+t5bn6+curav3N6Xdf7bsrpvAcOHLA7+Joup3dYAPDfQegGcEfy9vZWqVKl9Oeff97QfJmF2cxkNdptVu0mm+t6s/Lrr7/q4Ycf1n333afx48erZMmSyp8/vyZPnmw3GFO6q88cZqdDhw5q1KiRvv32Wy1evFijRo3S//73P82ZM8e6/vJG/NtH/u3cubM+/vhjFSpUSB07dswy6KSlpalmzZoaPXp0pq9fe0AlOzn9W1yPzWbTN998o99++03ff/+9Fi1apB49eui9997Tb7/9luG2TjmRPtjYihUrJMk6g9ylS5csQ29OrqG/Ede+P+k1jBo1KsP11+my29Zp06ape/fuatOmjV555RWVKFFCrq6uiomJ0Z49ezJMn5uf06yknx3u0KFDpq8vX75cTZs2veHlpqWlyWaz6ccff8x0O659n3JrX8xOTt7P3Ph8ZfX9nNWgjbfyd86LfQTAfwehG8Ad66GHHtKECRO0Zs2aTM9GXK1cuXJKS0vTrl27rLNz0pUBos6cOaNy5crlam1paWnau3evdXZbkv7++29JsgZAmz17tjw8PLRo0SK7s5qTJ0++5fWXLFlSzz//vJ5//nkdP35cd999t9566y21bNnS2tadO3dmOCu8c+fOXHsvrl7P1Wf9k5OTtW/fPrsztLeic+fOGjx4sI4ePaqpU6dmOV1QUJC2bt2qBx544LoHX3J6cCY7QUFBSktL07Zt27IMmunq16+v+vXr66233tKMGTP0xBNPaObMmXr66advat2pqak6d+6cpCvdZgsVKqTU1NSbfs/LlSunnTt3ZmhPHzn+evtMehdgb2/vm6rhm2++UYUKFTRnzhy7v83V972+Een1pp+Bv1pm23mt8+fP67vvvlPHjh3Vvn37DK+/+OKLmj59erahO6t9LCgoSMYYlS9f3u77498uNz5fRYoUsbuzQbrc6BVzM8qVK2fXiyFdZm0A/tu4phvAHevVV1+Vl5eXnn76acXFxWV4fc+ePRo7dqykK/d0lqQxY8bYTZN+Viaz0a5v1UcffWT9vzFGH330kfLnz68HHnhA0pUzLTabze4szv79+61Rfm9GamqqEhIS7NpKlCihUqVKWd3Xw8LCVKJECcXGxtp1af/xxx+1ffv2XHsvwsPD5ebmpg8++MDu7NHEiROVkJCQa+sJCgrSmDFjFBMTk+0I2B06dNCRI0f06aefZnjt4sWLOn/+vPXcy8sr0x//N6JNmzZycXHR8OHDM1yvnP5+nD59OsOZtfSAfu2t7HLql19+0blz5xQSEiLpyn7Wrl07zZ49O9OeIenXfmenVatWWrdundasWWO1nT9/XhMmTFBgYOB1r4MODQ1VUFCQ3n33XetgwI3UkH5W8ur3au3atXb13IiSJUuqdu3amjJlit3nZcmSJRmuT8/Mt99+q/Pnz6t3795q3759hsdDDz2k2bNnZ/s39PLykqQM+9mjjz4qV1dXDRs2LMO+YYzRyZMnb2BL805ufL6CgoKUkJBg12396NGjN3Urt9zQvHlzrVmzRlu2bLHaTp06dUvX7AO4M3GmG8AdKygoSDNmzFDHjh1VrVo1de3aVTVq1FBycrJWr16tr7/+2rr/a0hIiLp166YJEybozJkzaty4sdatW6cpU6aoTZs2N9UNNDseHh5auHChunXrpnr16unHH3/UggUL9Prrr1vXPEZERGj06NFq0aKFOnfurOPHj2vcuHGqWLGi3Y/OG3H27FmVKVNG7du3V0hIiAoWLKiffvpJ69ev13vvvSfpyjWh//vf/xQZGanGjRurU6dO1i3DAgMD1bdv31x5D4oXL64BAwZo2LBhatGihR5++GHt3LlT48ePV506dewGZrpVV9+PPStPPvmkvvrqKz377LP65ZdfdM899yg1NVU7duzQV199pUWLFiksLEzSlZD4008/afTo0SpVqpTKly+f6TXN2alYsaIGDhyoESNGqFGjRnr00Ufl7u6u9evXq1SpUoqJidGUKVM0fvx4tW3bVkFBQTp79qw+/fRTeXt7WweKspOQkKBp06ZJujKwV/ot2QoUKKD+/ftb07399tv65ZdfVK9ePfXs2VPBwcE6deqUNm3apJ9++inDLaSu1b9/f3355Zdq2bKlXnzxRRUtWlRTpkzRvn37NHv27Oteu+zi4qLPPvtMLVu2VPXq1RUZGanSpUvryJEj+uWXX+Tt7a3vv/8+y/kfeughzZkzR23btlVERIT27dun2NhYBQcHZxricyImJkYRERG699571aNHD506dUoffvihqlevft1lTp8+XcWKFcvydoUPP/ywPv30Uy1YsECPPvpoptOEhoZKkgYOHKjHH39c+fPnV+vWrRUUFKQ333xTAwYM0P79+9WmTRsVKlRI+/bt07fffqtnnnlG/fr1u6ltdqTc+Hw9/vjjeu2119S2bVu9+OKLunDhgj7++GNVrlw5R9fZ57ZXX31V06ZNU7NmzfTCCy9YtwwrW7asTp06lSs9YgDcIfJ+wHQAyFt///236dmzpwkMDDRubm6mUKFC5p577jEffvihuXTpkjXd5cuXzbBhw0z58uVN/vz5TUBAgBkwYIDdNMZkfnsfYzK/lVL6LYpGjRpltXXr1s14eXmZPXv2mAcffNB4enoaPz8/M2TIkAy3jpo4caKpVKmScXd3N1WrVjWTJ0/O8vY4Wd0GTFfdMiwpKcm88sorJiQkxBQqVMh4eXmZkJAQM378+AzzzZo1y9x1113G3d3dFC1a1DzxxBPm8OHDdtOkb8u1MqsxKx999JGpWrWqyZ8/v/Hz8zPPPfecOX36dKbLu9FbhmUns/csOTnZ/O9//zPVq1c37u7upkiRIiY0NNQMGzbMJCQkWNPt2LHD3HfffaZAgQJGknV7o+zWndV7MmnSJOt9LlKkiGncuLFZsmSJMcaYTZs2mU6dOpmyZcsad3d3U6JECfPQQw+ZDRs2XPd9uPaWYTabzRQtWtQ8/PDDZuPGjRmmj4uLM7179zYBAQEmf/78xt/f3zzwwANmwoQJ133fjDFmz549pn379qZw4cLGw8PD1K1b18yfP99umvRbTn399deZ1rx582bz6KOPmmLFihl3d3dTrlw506FDB7N06dJstzUtLc2MHDnSlCtXzri7u5u77rrLzJ8/P8MtpjL7PF69XVffWs8YY2bPnm2qVatm3N3dTXBwsJkzZ06Wt61KFxcXZ/Lly2eefPLJLKe5cOGC8fT0NG3btjXGZP2dMmLECFO6dGnj4uKS4RZUs2fPNvfee6/x8vIyXl5epmrVqqZ3795m586d1jSNGzc21atXz7KOa13vs5PVLcMyW0dm79Otfr6MuXLbtRo1ahg3NzdTpUoVM23atBv6TixXrpzd8rK6ZVhmf4/GjRubxo0b27Vt3rzZNGrUyLi7u5syZcqYmJgY88EHHxhJ5tixYxmWAeC/yWYMI0IAQF7q3r27vvnmm5s+AwcA+Pfq06ePPvnkE507d+5fP9AkgLzBNd0AAADATbh48aLd85MnT2rq1Km69957CdwALFzTDQAAANyEBg0aqEmTJqpWrZri4uI0ceJEJSYm6o033nB2aQD+RQjdAAAAwE1o1aqVvvnmG02YMEE2m0133323Jk6cqPvuu8/ZpQH4F3H6Nd3jxo3TqFGjdOzYMYWEhOjDDz/M8pYuly9ftkZzPXLkiKpUqaL//e9/atGiRR5XDQAAAADA9Tn1mu5Zs2YpOjpaQ4YM0aZNmxQSEqLmzZvr+PHjmU4/aNAgffLJJ/rwww+1bds2Pfvss2rbtq02b96cx5UDAAAAAHB9Tj3TXa9ePdWpU0cfffSRJCktLU0BAQF64YUX7O4fmq5UqVIaOHCgevfubbW1a9dOBQoUsO5DCgAAAADAv4XTrulOTk7Wxo0bNWDAAKvNxcVF4eHhWrNmTabzJCUlycPDw66tQIECWrlyZZbrSUpKUlJSkvU8LS1Np06dUrFixWSz2W5xKwAAAAAA/0XGGJ09e1alSpWSi0vWncidFrpPnDih1NRU+fn52bX7+flpx44dmc7TvHlzjR49Wvfdd5+CgoK0dOlSzZkzR6mpqVmuJyYmRsOGDcvV2gEAAAAAkKRDhw6pTJkyWb5+W41ePnbsWPXs2VNVq1aVzWZTUFCQIiMjNWnSpCznGTBggKKjo63nCQkJKlu2rA4dOiRvb++8KBsAAAAAcIdJTExUQECAChUqlO10Tgvdvr6+cnV1VVxcnF17XFyc/P39M52nePHimjt3ri5duqSTJ0+qVKlS6t+/vypUqJDletzd3eXu7p6h3dvbm9ANAAAAALgl17ts2Wmjl7u5uSk0NFRLly612tLS0rR06VI1aNAg23k9PDxUunRppaSkaPbs2XrkkUccXS4AAAAAADfMqd3Lo6Oj1a1bN4WFhalu3boaM2aMzp8/r8jISElS165dVbp0acXExEiS1q5dqyNHjqh27do6cuSIhg4dqrS0NL366qvO3AwAAAAAADLl1NDdsWNHxcfHa/DgwTp27Jhq166thQsXWoOrHTx40G4UuEuXLmnQoEHau3evChYsqFatWmnq1KkqXLiwk7YAAAAAAICsOfU+3c6QmJgoHx8fJSQkcE03AAAA/tXS0tKUnJzs7DKA/6T8+fPL1dU1y9dzmi1vq9HLAQAAgP+K5ORk7du3T2lpac4uBfjPKly4sPz9/a87WFp2CN0AAADAv4wxRkePHpWrq6sCAgLsLrkE4HjGGF24cEHHjx+XJJUsWfKml0XoBgAAAP5lUlJSdOHCBZUqVUqenp7OLgf4TypQoIAk6fjx4ypRokS2Xc2zwyEzAAAA4F8mNTVV0pXb7AJwnvSDXpcvX77pZRC6AQAAgH+pW7mOFMCty43PIKEbAAAAAAAHIXQDAAAAuCPYbDbNnTvX2WVk68KFC2rXrp28vb1ls9l05swZZ5dkuR3ev6sFBgZqzJgxzi7juhhIDQAAALhNBPZfkKfr2/92xA1N3717d02ZMkW9evVSbGys3Wu9e/fW+PHj1a1bN33++ec5Wt6yZcvUtGlTnT59WoULF77u9EePHlWRIkVuqOacCgwM1IEDByRJLi4u8vPzU8uWLfXuu+/e0DqnTJmiX3/9VatXr5avr698fHwcUm9Wrt6OdKVLl9bhw4cd8v4FBgaqT58+6tOnT64uV5LWr18vLy+vXF9ubuNMNwAAAIBcExAQoJkzZ+rixYtW26VLlzRjxgyVLVvWIetMTk6WJPn7+8vd3d0h65Ck4cOH6+jRozp48KCmT5+uFStW6MUXX7yhZezZs0fVqlVTjRo1bvr+z6mpqbd0//b07Uh/bN68WZLj37/cVrx48dtidH9CNwAAAIBcc/fddysgIEBz5syx2ubMmaOyZcvqrrvusps2LS1NMTExKl++vAoUKKCQkBB98803kqT9+/eradOmkqQiRYrIZrOpe/fukqQmTZooKipKffr0ka+vr5o3by4pY/fow4cPq1OnTipatKi8vLwUFhamtWvXSpK2bt2qpk2bqlChQvL29lZoaKg2bNiQ7bYVKlRI/v7+Kl26tJo2bapu3bpp06ZNdtOsXLlSjRo1UoECBRQQEKAXX3xR58+ft+p+7733tGLFCtlsNjVp0kSSdPr0aXXt2lVFihSRp6enWrZsqV27dlnL/Pzzz1W4cGHNmzdPwcHBcnd318GDB5WUlKR+/fqpdOnS8vLyUr169bRs2bLr/o3StyP9Ubx48Qzv3/79+2Wz2TRnzhw1bdpUnp6eCgkJ0Zo1a25oew8cOKC+ffvKZrNZBxiGDh2q2rVr2y1nzJgxCgwMtJ53795dbdq00bvvvquSJUuqWLFi6t27t90o4td2L7fZbPrss8/Utm1beXp6qlKlSpo3b57deubNm6dKlSrJw8NDTZs21ZQpUxzezZ/QDQAAACBX9ejRQ5MnT7aeT5o0SZGRkRmmi4mJ0RdffKHY2Fj99ddf6tu3r7p06aLly5crICBAs2fPliTt3LlTR48e1dixY615p0yZIjc3N61atSpDV3ZJOnfunBo3bqwjR45o3rx52rp1q1599VXrDPETTzyhMmXKaP369dq4caP69++v/Pnz53gbjxw5ou+//1716tWz2vbs2aMWLVqoXbt2+v333zVr1iytXLlSUVFRkq4cfOjZs6caNGigo0ePWgcmunfvrg0bNmjevHlas2aNjDFq1aqVXcC8cOGC/ve//+mzzz7TX3/9pRIlSigqKkpr1qzRzJkz9fvvv+uxxx5TixYt7AL7rRo4cKD69eunLVu2qHLlyurUqZNSUlJyvL1lypSxO7N+I3755Rft2bNHv/zyi6ZMmaLPP//8upcmDBs2TB06dNDvv/+uVq1a6YknntCpU6ckSfv27VP79u3Vpk0bbd26Vb169dLAgQNv/E25QYRuAAAAALmqS5cuWrlypQ4cOKADBw5o1apV6tKli900SUlJGjlypCZNmqTmzZurQoUK6t69u7p06aJPPvlErq6uKlq0qCSpRIkS8vf3t7v+uVKlSnrnnXdUpUoVValSJUMNM2bMUHx8vObOnat7771XFStWVIcOHdSgQQNJ0sGDBxUeHq6qVauqUqVKeuyxxxQSEpLtdr322msqWLCgChQooDJlyshms2n06NHW6zExMXriiSfUp08fVapUSQ0bNtQHH3ygL774QpcuXVLRokXl6ekpNzc3+fv7q2jRotq1a5fmzZunzz77TI0aNVJISIimT5+uI0eO2J21v3z5ssaPH6+GDRuqSpUqOnHihCZPnqyvv/5ajRo1UlBQkPr166d7773X7oBHdtuR/vjggw+ynLZfv36KiIhQ5cqVNWzYMB04cEC7d+/O8fa6urranVm/EUWKFNFHH32kqlWr6qGHHlJERISWLl2a7Tzdu3dXp06dVLFiRY0cOVLnzp3TunXrJEmffPKJqlSpolGjRqlKlSp6/PHHrd4TjsRAagAAAAByVfHixRUREaHPP/9cxhhFRETI19fXbprdu3frwoULatasmV17cnJyhm7omQkNDc329S1btuiuu+6ygvu1oqOj9fTTT2vq1KkKDw/XY489pqCgoGyX+corr6h79+4yxujQoUN6/fXXFRERoRUrVsjV1VVbt27V77//runTp1vzGGOUlpamffv2qVq1ahmWuX37duXLl8/ujHmxYsVUpUoVbd++3Wpzc3NTrVq1rOd//PGHUlNTVblyZbvlJSUlqVixYjnajnTX/m2udvU6S5YsKUk6fvy4qlatelPbeyOqV68uV1dXu/X/8ccf2c5zdb1eXl7y9vbW8ePHJV3pMVGnTh276evWrXtLNeYEoRsAAABAruvRo4fVzXjcuHEZXj937pwkacGCBSpdurTdazkZzOt6o1YXKFAg29eHDh2qzp07a8GCBfrxxx81ZMgQzZw5U23bts1yHl9fX1WsWFHSlTPtY8aMUYMGDfTLL78oPDxc586dU69evTIdXO1WB5ErUKCA3aBr586dk6urqzZu3GgXTCWpYMGC2S7r6u24nqu73KevP72L/s1ur4uLi4wxdm1Xd6XPbN3p67/eAHI3M4+jEboBAAAA5LoWLVooOTlZNpvNGujsalcPCNa4ceNMl+Hm5ibpymjdN6pWrVr67LPPdOrUqSzPdleuXFmVK1dW37591alTJ02ePDnb0H2t9LCbPlL73XffrW3btuU40EpStWrVlJKSorVr16phw4aSpJMnT2rnzp0KDg7Ocr677rpLqampOn78uBo1apTj9eWmnGyvm5tbhr9f8eLFdezYMRljrCC/ZcsWR5YqSapSpYp++OEHu7b169c7fL1c0w0AAAAg17m6umr79u3atm1bhjOx0pURtPv166e+fftqypQp2rNnjzZt2qQPP/xQU6ZMkSSVK1dONptN8+fPV3x8vHV2PCc6deokf39/tWnTRqtWrdLevXs1e/ZsrVmzRhcvXlRUVJSWLVtmXXO+fv3663aHPnv2rI4dO6ajR49q3bp1euWVV1S8eHErLL/22mtavXq1oqKitGXLFu3atUvfffeddcY/M5UqVdIjjzyinj17auXKldq6dau6dOmi0qVL65FHHslyvsqVK+uJJ55Q165dNWfOHO3bt0/r1q1TTEyMFizIm/u552R7AwMDtWLFCh05ckQnTpyQdGVU8/j4eL3zzjvas2ePxo0bpx9//NHh9fbq1Us7duzQa6+9pr///ltfffWVNTDbzdy6Lac40w38BwT2z5sv3ryw/+0IZ5cAAAByyNvbO9vXR4wYoeLFiysmJkZ79+5V4cKFdffdd+v111+XJJUuXVrDhg1T//79FRkZqa5du1539Op0bm5uWrx4sV5++WW1atVKKSkpCg4O1rhx4+Tq6qqTJ0+qa9euiouLk6+vrx599FENGzYs22UOHjxYgwcPlnTlbG2dOnW0ePFi6xrqWrVqafny5Ro4cKAaNWokY4yCgoLUsWPHbJc7efJkvfTSS3rooYeUnJys++67Tz/88MN1R1OfPHmy3nzzTb388ss6cuSIfH19Vb9+fT300EM5eo9uVU62d/jw4erVq5eCgoKUlJQkY4yqVaum8ePHa+TIkRoxYoTatWunfv36acKECQ6tt3z58vrmm2/08ssva+zYsWrQoIEGDhyo5557zqH3J7eZazvT3+ESExPl4+OjhISE634JAHcKQjcAALeXS5cuad++fSpfvrw8PDycXQ5wx3rrrbcUGxurQ4cOZfp6dp/FnGZLznQDAAAAAP4Txo8frzp16qhYsWJatWqVRo0alW33/9xA6AYAAAAA/Cfs2rVLb775pk6dOqWyZcvq5Zdf1oABAxy6TkI3AAAAAOA/4f3339f777+fp+tk9HIAAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAwB3BZrNp7ty5Tlv/sWPH1KxZM3l5ealw4cJOq0OS9u/fL5vNpi1btji1jhvh7L+fo3CfbgAAAOB2MdQnj9eXcEOTd+/eXVOmTFGvXr0UGxtr91rv3r01fvx4devWTZ9//nmOlrds2TI1bdpUp0+fzlGIPXr0qIoUKXJDNd+omJgYDRo0SG+//bZeeeUVu9fef/99HT16VFu2bJGPj88N138zbDZbhrZ77rlHy5cv19GjR+Xr65vr6/v222/Vpk2bXF2ulDd/P2fgTDcAAACAXBMQEKCZM2fq4sWLVtulS5c0Y8YMlS1b1iHrTE5OliT5+/vL3d3dIetIN2nSJL366quaNGlShtf27Nmj0NBQVapUSSVKlMi1dRpjlJKSkuXrkydP1tGjR63HvHnz5OrqKn9/f+XLd/ucZ82Lv58zELoBAAAA5Jq7775bAQEBmjNnjtU2Z84clS1bVnfddZfdtGlpaYqJiVH58uVVoEABhYSE6JtvvpF0pXt006ZNJUlFihSRzWZT9+7dJUlNmjRRVFSU+vTpI19fXzVv3lxSxu7Jhw8fVqdOnVS0aFF5eXkpLCxMa9eulSRt3bpVTZs2VaFCheTt7a3Q0FBt2LAh221bvny5Ll68qOHDhysxMVGrV6+2XgsMDNTs2bP1xRdfWLVmVX922y1dOcNvs9n0448/KjQ0VO7u7lq5cmWWdRUuXFj+/v7Wo2jRohm6l6cvc+nSpQoLC5Onp6caNmyonTt32i3ru+++09133y0PDw9VqFBBw4YNswJ/YGCgJKlt27ay2WzW8+7du2c4892nTx81adLEet6kSRO9+OKLevXVV1W0aFH5+/tr6NChdvNc/fdLr3/OnDlq2rSpPD09FRISojVr1tjN8+mnnyogIECenp5q27atRo8e7fSu/dcidAMAAADIVT169NDkyZOt55MmTVJkZGSG6WJiYvTFF18oNjZWf/31l/r27asuXbpo+fLlCggI0OzZsyVJO3fu1NGjRzV27Fhr3ilTpsjNzU2rVq3K0JVdks6dO6fGjRvryJEjmjdvnrZu3apXX31VaWlpkqQnnnhCZcqU0fr167Vx40b1799f+fPnz3a7Jk6cqE6dOil//vzq1KmTJk6caL22fv16tWjRQh06dLBqzar+7Lb7av3799fbb7+t7du3q1atWtnWllMDBw7Ue++9pw0bNihfvnzq0aOH9dqvv/6qrl276qWXXtK2bdv0ySef6PPPP9dbb71lbaP0f2fW05/n1JQpU+Tl5aW1a9fqnXfe0fDhw7VkyZLr1tuvXz9t2bJFlStXVqdOnayDAKtWrdKzzz6rl156SVu2bFGzZs2sWv9Nbp++BgAAAABuC126dNGAAQN04MABSVfC0cyZM7Vs2TJrmqSkJI0cOVI//fSTGjRoIEmqUKGCVq5cqU8++USNGzdW0aJFJUklSpTIcPayUqVKeuedd7KsYcaMGYqPj9f69eut5VSsWNF6/eDBg3rllVdUtWpVa3nZSUxM1DfffGOdae3SpYsaNWqksWPHqmDBgipevLjc3d1VoEAB+fv7S1Km9edku9MNHz5czZo1y7YuSerUqZNcXV2t59OmTVPt2rUznfatt96y1tG/f39FRETo0qVL8vDw0LBhw9S/f39169bNqmvEiBF69dVXNWTIEBUvXlzS/51Zv1G1atXSkCFDJF15vz/66CMtXbo0223s16+fIiIiJEnDhg1T9erVtXv3blWtWlUffvihWrZsqX79+kmSKleurNWrV2v+/Pk3XJsjEboBAAAA5KrixYsrIiJCn3/+uYwxioiIyDCg1+7du3XhwoUMgSs5OTlDN/TMhIaGZvv6li1bdNddd1nB91rR0dF6+umnNXXqVIWHh+uxxx5TUFBQlsv78ssvFRQUpJCQEElS7dq1Va5cOc2aNUtPPfXUdetNdyPbHRYWlqNlvv/++woPD7eelyxZUvHx8ZlOe/UZ85IlS0qSjh8/rrJly2rr1q1atWqV3dni1NRUXbp0SRcuXJCnp2eO6snKtWfrS5YsqePHj+d4nqvrrVq1qnbu3Km2bdvaTV+3bl1CNwAAAIA7X48ePRQVFSVJGjduXIbXz507J0lasGCBSpcubfdaTgbT8vLyyvb1AgUKZPv60KFD1blzZy1YsEA//vijhgwZopkzZ2YIcekmTpyov/76y25gsrS0NE2aNOmGQveNbPf1tjGdv7+/3Vl8SVmG7qu70KePfJ7e5f7cuXMaNmyYHn300QzzeXh4ZLl+FxcXGWPs2i5fvpztutPXn77urGRX7+2C0A0AAAAg17Vo0ULJycmy2WzWQGdXCw4Olru7uw4ePGjXpfpqbm5ukq6cbb1RtWrV0meffaZTp05leba7cuXKqly5svr27atOnTpp8uTJmYbuP/74Qxs2bNCyZcvslnXq1Ck1adJEO3bssLqpX6/+nGy3s9x9993auXNnhgB/tfz582f4exQvXlx//vmnXduWLVuue438rapSpUqG68pv9DrzvEDoBgAAAJDrXF1dtX37duv/r1WoUCH169dPffv2VVpamu69914lJCRo1apV8vb2Vrdu3VSuXDnZbDbNnz9frVq1UoECBVSwYMEcrb9Tp04aOXKk2rRpo5iYGJUsWVKbN29WqVKlVLt2bb3yyitq3769ypcvr8OHD2v9+vVq165dpsuaOHGi6tatq/vuuy/Da3Xq1NHEiRM1atSoDK9lVn9OtttZBg8erIceekhly5ZV+/bt5eLioq1bt+rPP//Um2++KenKCOZLly7VPffcI3d3dxUpUkT333+/Ro0apS+++EINGjTQtGnT9Oeff+boMoFb8cILL+i+++7T6NGj1bp1a/3888/68ccfM713uTMxejkAAAAAh/D29pa3t3eWr48YMUJvvPGGYmJiVK1aNbVo0UILFixQ+fLlJUmlS5e2Bvfy8/OzuqvnhJubmxYvXqwSJUqoVatWqlmzpt5++225urrK1dVVJ0+eVNeuXVW5cmV16NBBLVu21LBhwzIsJzk5WdOmTcsykLdr105ffPFFpt2ps6r/etvtLM2bN9f8+fO1ePFi1alTR/Xr19f777+vcuXKWdO89957WrJkiQICAqxQ3bx5c73xxht69dVXVadOHZ09e1Zdu3Z1eL333HOPYmNjNXr0aIWEhGjhwoXq27dvtl3hncFmru18f4dLTEyUj4+PEhISsv0CAO4kgf0XOLuEXLP/7QhnlwAAgMNdunRJ+/btU/ny5f91AQL4N+vZs6d27NihX3/9NVeWl91nMafZku7lAAAAAIDb0rvvvqtmzZrJy8tLP/74o6ZMmaLx48c7uyw7hG4AAAAAwG1p3bp1euedd3T27FlVqFBBH3zwgZ5++mlnl2WH0A0AAAAAuC199dVXzi7hupw+kNq4ceMUGBgoDw8P1atXT+vWrct2+jFjxqhKlSoqUKCAAgIC1LdvX126dCmPqgUAAAAAIOecGrpnzZql6OhoDRkyRJs2bVJISIiaN2+u48ePZzr9jBkz1L9/fw0ZMkTbt2/XxIkTNWvWLL3++ut5XDkAAAAAANfn1NA9evRo9ezZU5GRkQoODlZsbKw8PT01adKkTKdfvXq17rnnHnXu3FmBgYF68MEH1alTp+ueHQcAAABuR/+xGw0B/zppaWm3vAynXdOdnJysjRs3asCAAVabi4uLwsPDtWbNmkznadiwoaZNm6Z169apbt262rt3r3744Qc9+eSTWa4nKSlJSUlJ1vPExMTc2wgAAADAAfLnzy+bzab4+HgVL15cNpvN2SUB/ynGGCUnJys+Pl4uLi5yc3O76WU5LXSfOHFCqamp8vPzs2v38/PTjh07Mp2nc+fOOnHihO69914ZY5SSkqJnn3022+7lMTExmd7kHgAAAPi3cnV1VZkyZXT48GHt37/f2eUA/1menp4qW7asXFxuvpP4bTV6+bJlyzRy5EiNHz9e9erV0+7du/XSSy9pxIgReuONNzKdZ8CAAYqOjraeJyYmKiAgIK9KBgAAAG5KwYIFValSJV2+fNnZpQD/Sa6ursqXL98t9zRxWuj29fWVq6ur4uLi7Nrj4uLk7++f6TxvvPGGnnzySeu+azVr1tT58+f1zDPPaODAgZkefXB3d5e7u3vubwAAAADgYK6urnJ1dXV2GQBugdMGUnNzc1NoaKiWLl1qtaWlpWnp0qVq0KBBpvNcuHAhQ7BO/xJikAkAAAAAwL+NU7uXR0dHq1u3bgoLC1PdunU1ZswYnT9/XpGRkZKkrl27qnTp0oqJiZEktW7dWqNHj9Zdd91ldS9/44031Lp1a44AAgAAAAD+dZwaujt27Kj4+HgNHjxYx44dU+3atbVw4UJrcLWDBw/andkeNGiQbDabBg0apCNHjqh48eJq3bq13nrrLWdtAgAAAAAAWbKZ/1i/7MTERPn4+CghIUHe3t7OLgfIE4H9Fzi7hFyz/+0IZ5cAAAAA5DhbOu2abgAAAAAA7nS31S3DAEBDfZxdQe4ZmuDsCgDcKr6TAADXwZluAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEHyObsAAADw3xLYf4GzS8g1+z2cXQEA4N+OM90AAAAAADgIoRsAAAAAAAf5V4TucePGKTAwUB4eHqpXr57WrVuX5bRNmjSRzWbL8IiIiMjDigEAAAAAuD6nh+5Zs2YpOjpaQ4YM0aZNmxQSEqLmzZvr+PHjmU4/Z84cHT161Hr8+eefcnV11WOPPZbHlQMAAAAAkD2nh+7Ro0erZ8+eioyMVHBwsGJjY+Xp6alJkyZlOn3RokXl7+9vPZYsWSJPT09CNwAAAADgX8epoTs5OVkbN25UeHi41ebi4qLw8HCtWbMmR8uYOHGiHn/8cXl5eWX6elJSkhITE+0eAAAAAADkBaeG7hMnTig1NVV+fn527X5+fjp27Nh151+3bp3+/PNPPf3001lOExMTIx8fH+sREBBwy3UDAAAAAJATTu9efismTpyomjVrqm7dullOM2DAACUkJFiPQ4cO5WGFAAAAAID/snzOXLmvr69cXV0VFxdn1x4XFyd/f/9s5z1//rxmzpyp4cOHZzudu7u73N3db7lWAAAAAABulFPPdLu5uSk0NFRLly612tLS0rR06VI1aNAg23m//vprJSUlqUuXLo4uEwAAAACAm+LUM92SFB0drW7duiksLEx169bVmDFjdP78eUVGRkqSunbtqtKlSysmJsZuvokTJ6pNmzYqVqyYM8oGAAAAAOC6nB66O3bsqPj4eA0ePFjHjh1T7dq1tXDhQmtwtYMHD8rFxf6E/M6dO7Vy5UotXrzYGSUDAAAAAJAjTg/dkhQVFaWoqKhMX1u2bFmGtipVqsgY4+CqAAAAAAC4Nbf16OUAAAAAAPybEboBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAcJJ+zC8B/xFAfZ1eQu4YmOLsCAAAAALcBznQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAgTg/d48aNU2BgoDw8PFSvXj2tW7cu2+nPnDmj3r17q2TJknJ3d1flypX1ww8/5FG1AAAAAADkXD5nrnzWrFmKjo5WbGys6tWrpzFjxqh58+bauXOnSpQokWH65ORkNWvWTCVKlNA333yj0qVL68CBAypcuHDeFw8AAAAAwHU4NXSPHj1aPXv2VGRkpCQpNjZWCxYs0KRJk9S/f/8M00+aNEmnTp3S6tWrlT9/fklSYGBgXpYMAAAAAECOOa17eXJysjZu3Kjw8PD/K8bFReHh4VqzZk2m88ybN08NGjRQ79695efnpxo1amjkyJFKTU3Nq7IBAAAAAMgxp53pPnHihFJTU+Xn52fX7ufnpx07dmQ6z969e/Xzzz/riSee0A8//KDdu3fr+eef1+XLlzVkyJBM50lKSlJSUpL1PDExMfc2AgAAAACAbDh9ILUbkZaWphIlSmjChAkKDQ1Vx44dNXDgQMXGxmY5T0xMjHx8fKxHQEBAHlYMAAAAAPgvc1ro9vX1laurq+Li4uza4+Li5O/vn+k8JUuWVOXKleXq6mq1VatWTceOHVNycnKm8wwYMEAJCQnW49ChQ7m3EQAAAAAAZMNpodvNzU2hoaFaunSp1ZaWlqalS5eqQYMGmc5zzz33aPfu3UpLS7Pa/v77b5UsWVJubm6ZzuPu7i5vb2+7BwAAAAAAecGp3cujo6P16aefasqUKdq+fbuee+45nT9/3hrNvGvXrhowYIA1/XPPPadTp07ppZde0t9//60FCxZo5MiR6t27t7M2AQAAAACALDn1lmEdO3ZUfHy8Bg8erGPHjql27dpauHChNbjawYMH5eLyf8cFAgICtGjRIvXt21e1atVS6dKl9dJLL+m1115z1iYAAAAAAJAlp4ZuSYqKilJUVFSmry1btixDW4MGDfTbb785uCoAAAAAAG7dbTV6OQAAAAAAtxNCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADkLoBgAAAADAQQjdAAAAAAA4CKEbAAAAAAAHIXQDAAAAAOAghG4AAAAAAByE0A0AAAAAgIMQugEAAAAAcBBCNwAAAAAADvKvCN3jxo1TYGCgPDw8VK9ePa1bty7LaT///HPZbDa7h4eHRx5WCwAAAABAzjg9dM+aNUvR0dEaMmSINm3apJCQEDVv3lzHjx/Pch5vb28dPXrUehw4cCAPKwYAAAAAIGecHrpHjx6tnj17KjIyUsHBwYqNjZWnp6cmTZqU5Tw2m03+/v7Ww8/PLw8rBgAAAAAgZ5waupOTk7Vx40aFh4dbbS4uLgoPD9eaNWuynO/cuXMqV66cAgIC9Mgjj+ivv/7Ki3IBAAAAALghTg3dJ06cUGpqaoYz1X5+fjp27Fim81SpUkWTJk3Sd999p2nTpiktLU0NGzbU4cOHM50+KSlJiYmJdg8AAAAAAPJCPmcXcKMaNGigBg0aWM8bNmyoatWq6ZNPPtGIESMyTB8TE6Nhw4blZYkAgH+7oT7OriB3DU1wdgUAgH+DO+nftzvo3zannun29fWVq6ur4uLi7Nrj4uLk7++fo2Xkz59fd911l3bv3p3p6wMGDFBCQoL1OHTo0C3XDQAAAABATjg1dLu5uSk0NFRLly612tLS0rR06VK7s9nZSU1N1R9//KGSJUtm+rq7u7u8vb3tHgAAAAAA5AWndy+Pjo5Wt27dFBYWprp162rMmDE6f/68IiMjJUldu3ZV6dKlFRMTI0kaPny46tevr4oVK+rMmTMaNWqUDhw4oKefftqZmwEAAAAAQAZOD90dO3ZUfHy8Bg8erGPHjql27dpauHChNbjawYMH5eLyfyfkT58+rZ49e+rYsWMqUqSIQkNDtXr1agUHBztrEwAAAAAAyFSuhO7ExET9/PPPqlKliqpVq3bD80dFRSkqKirT15YtW2b3/P3339f7779/M2UCAAAAAJCnbuqa7g4dOuijjz6SJF28eFFhYWHq0KGDatWqpdmzZ+dqgQAAAAAA3K5uKnSvWLFCjRo1kiR9++23MsbozJkz+uCDD/Tmm2/maoEAAAAAANyubip0JyQkqGjRopKkhQsXql27dvL09FRERIR27dqVqwUCAAAAAHC7uqnQHRAQoDVr1uj8+fNauHChHnzwQUlXBjnz8PDI1QIBAAAAALhd3dRAan369NETTzyhggULqmzZsmrSpImkK93Oa9asmZv1AQAAAABw27qp0P3888+rbt26OnTokJo1a2bd0qtChQpc0w0AAAAAwP9307cMCwsLU61atbRv3z4FBQUpX758ioiIyM3aAAAAAAC4rd3UNd0XLlzQU089JU9PT1WvXl0HDx6UJL3wwgt6++23c7VAAAAAAABuVzcVugcMGKCtW7dq2bJldgOnhYeHa9asWblWHAAAAAAAt7Ob6l4+d+5czZo1S/Xr15fNZrPaq1evrj179uRacQAAAAAA3M5u6kx3fHy8SpQokaH9/PnzdiEcAAAAAID/spsK3WFhYVqwYIH1PD1of/bZZ2rQoEHuVAYAAAAAwG3uprqXjxw5Ui1bttS2bduUkpKisWPHatu2bVq9erWWL1+e2zUCAAAAAHBbuqkz3ffee6+2bt2qlJQU1axZU4sXL1aJEiW0Zs0ahYaG5naNAAAAAADclm74TPfly5fVq1cvvfHGG/r0008dURMAAAAAAHeEGz7TnT9/fs2ePdsRtQAAAAAAcEe5qe7lbdq00dy5c3O5FAAAAAAA7iw3NZBapUqVNHz4cK1atUqhoaHy8vKye/3FF1/MleIAAAAAALid3VTonjhxogoXLqyNGzdq48aNdq/ZbDZCNwAAAAAAusnQvW/fvtyuAwAAAACAO85NXdN9NWOMjDG5UQsAAAAAAHeUmw7dX3zxhWrWrKkCBQqoQIECqlWrlqZOnZqbtQEAAAAAcFu7qe7lo0eP1htvvKGoqCjdc889kqSVK1fq2Wef1YkTJ9S3b99cLRIAAAAAgNvRTYXuDz/8UB9//LG6du1qtT388MOqXr26hg4dSugGAAAAAEA32b386NGjatiwYYb2hg0b6ujRo7dcFAAAAAAAd4KbCt0VK1bUV199laF91qxZqlSp0i0XBQAAAADAneCmupcPGzZMHTt21IoVK6xruletWqWlS5dmGsYBAAAAAPgvuqkz3e3atdPatWvl6+uruXPnau7cufL19dW6devUtm3b3K4RAAAAAIDb0k2d6Zak0NBQTZs2LTdrAQAAAADgjnJTZ7p/+OEHLVq0KEP7okWL9OOPP95yUQAAAAAA3AluKnT3799fqampGdqNMerfv/8tFwUAAAAAwJ3gpkL3rl27FBwcnKG9atWq2r179y0XBQAAAADAneCmrun28fHR3r17FRgYaNe+e/dueXl55UZdkBTYf4GzS8g1+z2cXQEAAMC/1FAfZ1eQu4YmOLsC4F/lps50P/LII+rTp4/27Nljte3evVsvv/yyHn744VwrDgAAAACA29lNhe533nlHXl5eqlq1qsqXL6/y5curatWqKlasmN59993crhEAAAAAgNvSTXcvX716tZYsWaKtW7eqQIECCgkJUaNGjXK7PgAAAAAAbls3dKZ7zZo1mj9/viTJZrPpwQcfVIkSJfTuu++qXbt2euaZZ5SUlOSQQgEAAAAAuN3cUOgePny4/vrrL+v5H3/8oZ49e6pZs2bq37+/vv/+e8XExOR6kQAAAAAA3I5uKHRv2bJFDzzwgPV85syZqlu3rj799FNFR0frgw8+0FdffZXrRQIAAAAAcDu6odB9+vRp+fn5Wc+XL1+uli1bWs/r1KmjQ4cO5V51AAAAAADcxm4odPv5+Wnfvn2SpOTkZG3atEn169e3Xj979qzy58+fuxUCAAAAAHCbuqHQ3apVK/Xv31+//vqrBgwYIE9PT7sRy3///XcFBQXlepEAAAAAANyObih0jxgxQvny5VPjxo316aef6tNPP5Wbm5v1+qRJk/Tggw/ecBHjxo1TYGCgPDw8VK9ePa1bty5H882cOVM2m01t2rS54XUCAAAAAOBoN3Sfbl9fX61YsUIJCQkqWLCgXF1d7V7/+uuvVbBgwRsqYNasWYqOjlZsbKzq1aunMWPGqHnz5tq5c6dKlCiR5Xz79+9Xv379uDc4AAAAAOBf64bOdKfz8fHJELglqWjRonZnvnNi9OjR6tmzpyIjIxUcHKzY2Fh5enpq0qRJWc6TmpqqJ554QsOGDVOFChVuuH4AAAAAAPLCTYXu3JKcnKyNGzcqPDzcanNxcVF4eLjWrFmT5XzDhw9XiRIl9NRTT+VFmQAAAAAA3JQb6l6e206cOKHU1FS725BJV0ZJ37FjR6bzrFy5UhMnTtSWLVtytI6kpCQlJSVZzxMTE2+6XgAAAAAAboRTz3TfqLNnz+rJJ5/Up59+Kl9f3xzNExMTIx8fH+sREBDg4CoBAAAAALjCqWe6fX195erqqri4OLv2uLg4+fv7Z5h+z5492r9/v1q3bm21paWlSZLy5cunnTt3Zrhl2YABAxQdHW09T0xMJHgDAAAAAPKEU0O3m5ubQkNDtXTpUuu2X2lpaVq6dKmioqIyTF+1alX98ccfdm2DBg3S2bNnNXbs2EzDtLu7u9zd3R1SPwAAAAAA2XFq6Jak6OhodevWTWFhYapbt67GjBmj8+fPKzIyUpLUtWtXlS5dWjExMfLw8FCNGjXs5i9cuLAkZWgHAAAAAMDZnB66O3bsqPj4eA0ePFjHjh1T7dq1tXDhQmtwtYMHD8rF5ba69BwAAAAAAEn/gtAtSVFRUZl2J5ekZcuWZTvv559/nvsFAQAAAACQCziFDAAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHCSfswsAAAAAblRg/wXOLiHX7PdwdgUAHIkz3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA5C6AYAAAAAwEEI3QAAAAAAOMi/InSPGzdOgYGB8vDwUL169bRu3bosp50zZ47CwsJUuHBheXl5qXbt2po6dWoeVgsAAAAAQM44PXTPmjVL0dHRGjJkiDZt2qSQkBA1b95cx48fz3T6okWLauDAgVqzZo1+//13RUZGKjIyUosWLcrjygEAAAAAyJ7TQ/fo0aPVs2dPRUZGKjg4WLGxsfL09NSkSZMynb5JkyZq27atqlWrpqCgIL300kuqVauWVq5cmceVAwAAAACQPaeG7uTkZG3cuFHh4eFWm4uLi8LDw7VmzZrrzm+M0dKlS7Vz507dd999jiwVAAAAAIAbls+ZKz9x4oRSU1Pl5+dn1+7n56cdO3ZkOV9CQoJKly6tpKQkubq6avz48WrWrFmm0yYlJSkpKcl6npiYmDvFAwAAAABwHU4N3TerUKFC2rJli86dO6elS5cqOjpaFSpUUJMmTTJMGxMTo2HDhuV9kQAAAACA/zynhm5fX1+5uroqLi7Orj0uLk7+/v5Zzufi4qKKFStKkmrXrq3t27crJiYm09A9YMAARUdHW88TExMVEBCQOxsAAAAAAEA2nHpNt5ubm0JDQ7V06VKrLS0tTUuXLlWDBg1yvJy0tDS7LuRXc3d3l7e3t90DAAAAAIC84PTu5dHR0erWrZvCwsJUt25djRkzRufPn1dkZKQkqWvXripdurRiYmIkXekuHhYWpqCgICUlJemHH37Q1KlT9fHHHztzMwAAAAAAyMDpobtjx46Kj4/X4MGDdezYMdWuXVsLFy60Blc7ePCgXFz+74T8+fPn9fzzz+vw4cMqUKCAqlatqmnTpqljx47O2gQAAAAAADLl9NAtSVFRUYqKisr0tWXLltk9f/PNN/Xmm2/mQVUAAAAAANwap17TDQAAAADAnYzQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAfJ5+wCAAC3h8D+C5xdQq7Z7+HsCgAAwH8FZ7oBAAAAAHAQQjcAAAAAAA7yrwjd48aNU2BgoDw8PFSvXj2tW7cuy2k//fRTNWrUSEWKFFGRIkUUHh6e7fQAAAAAADiL00P3rFmzFB0drSFDhmjTpk0KCQlR8+bNdfz48UynX7ZsmTp16qRffvlFa9asUUBAgB588EEdOXIkjysHAAAAACB7Tg/do0ePVs+ePRUZGang4GDFxsbK09NTkyZNynT66dOn6/nnn1ft2rVVtWpVffbZZ0pLS9PSpUvzuHIAAAAAALLn1NCdnJysjRs3Kjw83GpzcXFReHi41qxZk6NlXLhwQZcvX1bRokUdVSYAAAAAADfFqbcMO3HihFJTU+Xn52fX7ufnpx07duRoGa+99ppKlSplF9yvlpSUpKSkJOt5YmLizRcMAAAAAMANcHr38lvx9ttva+bMmfr222/l4ZH5TVdjYmLk4+NjPQICAvK4SgAAAADAf5VTQ7evr69cXV0VFxdn1x4XFyd/f/9s53333Xf19ttva/HixapVq1aW0w0YMEAJCQnW49ChQ7lSOwAAAAAA1+PU0O3m5qbQ0FC7QdDSB0Vr0KBBlvO98847GjFihBYuXKiwsLBs1+Hu7i5vb2+7BwAAAAAAecGp13RLUnR0tLp166awsDDVrVtXY8aM0fnz5xUZGSlJ6tq1q0qXLq2YmBhJ0v/+9z8NHjxYM2bMUGBgoI4dOyZJKliwoAoWLOi07QAAAAAA4FpOD90dO3ZUfHy8Bg8erGPHjql27dpauHChNbjawYMH5eLyfyfkP/74YyUnJ6t9+/Z2yxkyZIiGDh2al6UDAAAAAJAtp4duSYqKilJUVFSmry1btszu+f79+x1fEAAAAAAAueC2Hr0cAAAAAIB/M0I3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAO4vTQPW7cOAUGBsrDw0P16tXTunXrspz2r7/+Urt27RQYGCibzaYxY8bkXaEAAAAAANwgp4buWbNmKTo6WkOGDNGmTZsUEhKi5s2b6/jx45lOf+HCBVWoUEFvv/22/P3987haAAAAAABujFND9+jRo9WzZ09FRkYqODhYsbGx8vT01KRJkzKdvk6dOho1apQef/xxubu753G1AAAAAADcGKeF7uTkZG3cuFHh4eH/V4yLi8LDw7VmzRpnlQUAAAAAQK7J56wVnzhxQqmpqfLz87Nr9/Pz044dO3JtPUlJSUpKSrKeJyYm5tqyAQAAAADIjtMHUnO0mJgY+fj4WI+AgABnlwQAAAAA+I9wWuj29fWVq6ur4uLi7Nrj4uJydZC0AQMGKCEhwXocOnQo15YNAAAAAEB2nBa63dzcFBoaqqVLl1ptaWlpWrp0qRo0aJBr63F3d5e3t7fdAwAAAACAvOC0a7olKTo6Wt26dVNYWJjq1q2rMWPG6Pz584qMjJQkde3aVaVLl1ZMTIykK4Ovbdu2zfr/I0eOaMuWLSpYsKAqVqzotO0AAAAAACAzTg3dHTt2VHx8vAYPHqxjx46pdu3aWrhwoTW42sGDB+Xi8n8n4//55x/ddddd1vN3331X7777rho3bqxly5bldfkAAAAAAGTLqaFbkqKiohQVFZXpa9cG6cDAQBlj8qAqAAAAAABu3R0/ejkAAAAAAM5C6AYAAAAAwEEI3QAAAAAAOAihGwAAAAAAByF0AwAAAADgIIRuAAAAAAAchNANAAAAAICDELoBAAAAAHAQQjcAAAAAAA6Sz9kFAAAAAICzBPZf4OwScs1+D2dXgMxwphsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAMAAAAA4CCEbgAAAAAAHITQDQAAAACAgxC6AQAAAABwEEI3AAAAAAAOQugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAf5V4TucePGKTAwUB4eHqpXr57WrVuX7fRff/21qlatKg8PD9WsWVM//PBDHlUKAAAAAEDOOT10z5o1S9HR0RoyZIg2bdqkkJAQNW/eXMePH890+tWrV6tTp0566qmntHnzZrVp00Zt2rTRn3/+mceVAwAAAACQPaeH7tGjR6tnz56KjIxUcHCwYmNj5enpqUmTJmU6/dixY9WiRQu98sorqlatmkaMGKG7775bH330UR5XDgAAAABA9vI5c+XJycnauHGjBgwYYLW5uLgoPDxca9asyXSeNWvWKDo62q6tefPmmjt3bqbTJyUlKSkpyXqekJAgSUpMTLzF6h0vLemCs0vINYk24+wSctdtsP9cjX3pX4r9yGnuqP1IYl9yojtqX2I/cpo7aj+S2Jec6I7al26D/Sg9UxqT/fvu1NB94sQJpaamys/Pz67dz89PO3bsyHSeY8eOZTr9sWPHMp0+JiZGw4YNy9AeEBBwk1XjZvg4u4Dc9vYdt0W3jTvqnWc/cpo77p1nX3KaO+qdZz9ymjvunWdfcpo76p2/jfajs2fPyscn63qdGrrzwoABA+zOjKelpenUqVMqVqyYbDabEyv770hMTFRAQIAOHTokb29vZ5eD2xj7EnID+xFyC/sScgP7EXIL+1LeM8bo7NmzKlWqVLbTOTV0+/r6ytXVVXFxcXbtcXFx8vf3z3Qef3//G5re3d1d7u7udm2FCxe++aJx07y9vfkCQK5gX0JuYD9CbmFfQm5gP0JuYV/KW9md4U7n1IHU3NzcFBoaqqVLl1ptaWlpWrp0qRo0aJDpPA0aNLCbXpKWLFmS5fQAAAAAADiL07uXR0dHq1u3bgoLC1PdunU1ZswYnT9/XpGRkZKkrl27qnTp0oqJiZEkvfTSS2rcuLHee+89RUREaObMmdqwYYMmTJjgzM0AAAAAACADp4fujh07Kj4+XoMHD9axY8dUu3ZtLVy40Bos7eDBg3Jx+b8T8g0bNtSMGTM0aNAgvf7666pUqZLmzp2rGjVqOGsTcB3u7u4aMmRIhm7+wI1iX0JuYD9CbmFfQm5gP0JuYV/697KZ641vDgAAAAAAbopTr+kGAAAAAOBORugGAAAAAMBBCN0AAAAAADgIoRsAAAAAAAchdAPAbcwYo9TUVKWmpopxMfPe2rVrNX/+fKWlpfH+447G/g0gryUkJGj+/Pkyxujy5cvOLueWMHo57jjGGKWlpclms9ndbg643aXv1zabzdml/OcYY2SMyfD+R0VF6eeff9a2bducWB2QM+n7cE6kpKQoXz6n31kWd7i0tDRJ4t82WAevXV1drbYVK1aoSZMm1n5yOyOR4LaVlpaW6dklm80mV1dXK3CfOHFCSUlJzigRuCXX7tsuLi4ZfpScPHlSI0eOVOPGjRUeHq6PP/5YFy9ezMsy/xPSD+Jl9qMwLi5O9913n0qVKqWePXtK4qwgnCezfW/Tpk16//33ZbPZ7H68pqWlKSUlRampqRnmuTpwnzp1Shs2bNCpU6ccUzTuCOfOndPPP/+sffv2SbLfF9PS0jLtkeXi4mL33ZqQkKCjR4/mXdHIFTf7b97V3z0uLi52gTt9ue7u7urQoYPq1Kmjhx56SP/8888t1eoshG786+X0izouLk4XLlzQiRMn9NJLL6l8+fLy8fFRz5499ccff0jihzD+XdL37cx+8Eqy9u1Tp05p3759+uuvv/Tyyy9br1+6dElvvfWWfv75Zz3wwAN66KGHNG7cOD311FN5Uv+dIr13TFYuXbqktWvXasKECZo1a5b19zpz5oymTZumhIQE1apVS59++qneeustSeKMDZzm2n0vLS1NW7Zs0caNGyXJrgeYi4uL8uXLl+GHriT1799fP/74o/r06aOKFSuqefPmeuqpp6xABUj/1wtIktavX6/33ntPhw4dkmS/L6YHqmv3z99//11vv/22tm/frooVKyogIEBt2rTRzz//nHcbgVuW3b956Qf3rv0NfvVZ7UuXLmnZsmUaM2aMVq5caXUlj4mJUVJSkvbu3asuXbpo0KBBKlGihOM2xIHoXg6nSu/qdurUKRUqVMj6Qs7JD9Y//vhDCxYs0N13363o6GglJyfrgw8+0N9//60pU6Zo0KBBqlChgs6fP6/ChQsrODg4D7YIyFz6tdeZ/ejIytq1a9WtWzcdPnxYjRs3lpubm7777jtduHBBHh4e2rJli1q1aqUtW7ZY/wht3bpVd911l7Zt26aqVas6cpNuO+nv/9Wu7W575swZFS5c2Hq+bNkyvf766zp06JACAgJ0+vRpBQcHa/r06fLw8NClS5dUtGhR/fDDD2rSpEkebQmQOWOMZs2apcKFC+uBBx5Q/vz5s5w2NTVVq1at0pw5c3Tw4EE9/PDDateunQoVKiRJqlKlivbv36+XXnpJ/fr106FDhxQREaGOHTtq7NixebVJ+Jf6559/tHLlSnXo0EHJyclyc3PLctrTp0/rl19+0dq1a1W8eHH17NlTPj4+kqTY2Fg9//zzioiI0DPPPKNatWqpd+/e2rdvn1auXKkiRYrk1SbhFqxcuVI2m0116tRR/vz5s/ydk5aWZnfg77XXXtOpU6dUtmxZzZkzR8nJyUpJSdELL7ygqKgonTt3Tp06dZKvr68mT56cV5vjEJzphtOk/9gdMWKE7r77buuDaLPZrKNhZ86c0axZsxQVFaU33nhDO3futOY/deqUXn/9db311lvq37+/Nm3apGrVqmnmzJl64IEH1LZtW4WEhKhhw4YEbuSJq4/4X8tmsylfvnzWP0QXLlzQkiVL9Oqrr6pt27YaOHCgTpw4YU1/+fJlPfvsswoNDdW2bdv09ttv68yZM5JknWmaMmWKdYa1Xbt2qly5svVD+798NiqrQeWuDtxHjx7VH3/8IZvNpiNHjqhZs2by9vZWy5YttX79emv+s2fP6uWXX9a+ffu0evVqLV++XKtWrdKXX34pSfLw8FCpUqW0evXqvNk4/KecPXtWa9asyXaa9Mus0sd8ePPNNzV//nwrcKekpOjSpUv65JNP9Ndff1nzzZs3Ty+88IL27NmjMmXK6I033lB0dLTi4+MlSa1atVL+/PnVpk0blShRQqGhoXrhhRe0fPlyHTx40HEbDac4fvz4dS8fSElJsf5/7dq1evzxxyXJLnD/+eefWrFihfU8KSlJzzzzjPr376+dO3dq+vTpatKkibUPVa5cWUWKFFFYWJhat26tcuXKadiwYYqPj9f69esl0UsxL6T3vMvptdPpJxLS94no6Gi99957cnNzyxC4f/rpJzVv3lwBAQFq1aqVvvrqK124cEGSVLhwYU2cOFHbt2/XsmXLtHXrVoWHh2v06NE6efKkChYsqCpVqmjDhg1WnbcrQjccLv2Dee2XZnoXzXvvvVdnzpzR5MmT1b9/f73++uuy2Ww6e/asXnvtNQ0ZMkTHjh3Thg0bdP/991td5EJDQyVJ/v7+6ty5swoWLKhy5crpnnvu0YwZM/TII49o4MCB+vrrr7V7926+tOEw6ftWZr00Ll68qD179ujPP/9U9+7dFRYWpsuXL2vevHkaOXKkdu/ererVq2vx4sVq3bq1jhw5IknauHGjDh8+rJdeeklly5ZVzZo1NWrUKBUsWFDr1q2TdCVErl69WosWLVKlSpU0bNgwLVu2TMePH1fLli3z9k3IY1eP2n6tq3sTXLhwQRs3blRcXJxefPFFrV+/Xu3atVPFihV1//33a8yYMRo6dKhat26tDRs2KH/+/Orfv7/27t0rSXrggQfUrl07HT58WJ9//rk++ugjnT59Wr/99psSEhIkScHBwdq6dWvebTz+M6ZMmaKIiAjt2rVLkjLd39MPVqf/aC5atKjmzZun4OBgubi46Msvv9TFixc1aNAg62BRfHy83n//fVWqVEnff/+9PvjgA40aNUobNmzQhAkTJElVq1aVl5eX/P39rXXVqVNHx44d07Fjx/Jg65FXTpw4oYoVK2rmzJlW2/Wu809KSlK+fPn08MMPq0qVKnrkkUckSQMHDtTw4cMVFxcn6Ur34PXr1+uLL77Q3LlzNXfuXCUnJ2vYsGG6dOmSypUrp/z58ysgIMBadkBAgPz8/LRq1SpHbTKukd7938XFRcYYbdmyRYsXL7YOxFy9P6SfNHN1dbX2ifLly2vjxo164IEHVKJECQ0ZMkSStGbNGg0ePFi1atXSZ599pkaNGmnYsGFWb5kGDRpIktq1aycfHx/ly5dPffr0UXx8vH7//XdJUkhIiPbt2ydjzG09QPLtWznyRGZB9YcfflDv3r0zHG1KP+J1bXv6B9Nms+nixYvW0c18+fLpzJkz6tixo86ePauhQ4fq999/V8mSJSVJv/zyi6ZMmaLPPvtM33zzjaZPn67atWvr1VdfVWJiogoWLCgfHx8FBwfbBZ0BAwZo4MCBql27tk6ePKkXXnhBbdu21ZYtW7LcJuBW2Gw2Xb58WVu2bNHSpUvtbmsxcOBAPfzwwxo2bJhKlCih5557Tvnz51exYsU0btw4zZkzR2+++aZ+/fVXxcXFac6cOZKuXOdWoUIFu/UEBQWpWrVqWrt2rSSpTJkyCgwM1KhRo/T222+rU6dOqlGjhjw8PKyBaG7n/f3MmTNasmSJhg4dqlatWqlEiRLWP8Lp3yuZXYv69ddfa+zYsRozZoxq166tHj16aPv27Zo9e7batGmj+++/XydPnlSvXr00YsQIJSQkqEePHqpcubIGDBig8+fPW2euPT09NXnyZOt6+WPHjqlRo0Zas2aNzp8/L+nKAcAdO3bYnQUCbtTVYwukf24rVKigsmXLas+ePZL+r7fG1d8xy5cvV+vWrbV+/XqtXr3aGt+kU6dO2rRpkzp37qy0tDQ1adLECu/bt29XXFycIiMjreVEREQoLCxMP/30kyQpLCxMp06dsnrYSFKtWrV04cIFBrq6zV3bK8vX11dly5bVqVOnrO+xzL5bu3fvrgkTJiglJUXjxo1TSkqKkpKS9Prrr+vNN9+UJNWuXVuurq46fvy4pCujT7do0UL169fX5cuXVa5cOfXo0UN//fWXdu7cqRIlSqhixYravHmztZ6CBQuqatWq1pnu2/ns5u1g//79Gjp0qO6//36VKFFCbm5uCg0N1eeff24dXHZ1dbUbaf7EiRN6/PHHNXbsWP31119atGiR4uLiVLlyZU2YMEHPP/+8UlJS9P3336tSpUoaNWqUmjdvrgEDBqhNmzYaOXKkJKl69eqSrvzN01WoUEH58+fX/v37JV251OXSpUu37QBq6QjdyJbNZtPKlStVtmxZ7dixQ5Lk7u6uEiVKZPgSTD/ilX4UKv31w4cP68UXX1T58uVVsmRJRURE6M0331RKSooKFy6sTZs2yd3dXaNGjdIPP/ygF154QZI0Y8YMtW3bVvXr15ckFS1aVL169dKePXusH94VKlTQP//8o+TkZElX/iEpWrSonnvuOQ0ZMkSxsbH67bffdPHiRS1cuNDxbxj+dXLaZep6g2ld7dChQ9Y/RJI0evRolS1bVi1bttQLL7ygbt26Wd04q1WrpmPHjsnNzU0jRoywBjlr1qyZypQpo08//VTt27dXnTp1tH//fqsnh5+fn/Lly2cNSCNd6crs5+dnLTs8PFwlS5bUyy+/rO3btys1NVWHDx/W2LFjb/trLsePH6+iRYuqU6dO+u2331S7dm19+OGHqlixoiRp586dGjZsmCIiIhQdHW2FEknasmWLBg8erG+//VYTJkzQhg0bFBISojp16sjf31+PPfaYPDw81KtXL1WsWFHe3t7WP/gVK1aUj4+PNm3aJEn67bff9Nprr6lz585atmyZJkyYoOeee04HDhzQyZMnJUmNGjXSH3/8YR0MSf+xCdyIq29zmf5dVLJkSXl5eVn/5i1atEgNGjTQvHnzrPlcXV11+PBhbdq0SY0aNdLkyZPl7e2tJk2aWAGoYMGCqlChgjWoaKlSpRQXF6cCBQpYy/Hw8FBQUJB1hjIkJESpqal2Xcn9/f3l4+OjgwcPEoRuQ5n1yko/g1mmTBnt2bNHZ8+elSQ98sgj1sCQ6X/r7du3a+nSpcqXL5+WLFmievXqKSwsTN26dVONGjUkSZUqVdKpU6es70EvLy/rwE16oK9du7bOnTun+Ph4eXl5qUqVKlZvofTRqmvVqmX97mRgyhu3detWffTRR1bvucwOwKf/XVevXq3hw4erfPny+vrrr7V3716lpKRoxowZKl++vE6dOqUePXqoUaNGOnfunKQrB/t+/PFHdezYUdWrV9fOnTuVnJysZ555Rm3atJGfn5/S0tI0b948Xbx40ephVqRIEU2YMEHVq1dXQkKCihcvLi8vL+tMtnTlO83Pz09///23Ll++rMqVK8vT09M6KZGWlnZb3rOb0I0spX85litXTocPH7au83rggQc0ZMiQDPfvnD9/vtq0aaMKFSqofv361pnlgwcPys3NTaNHj9aOHTs0fPhwvf322/ryyy+VkpJina3btGmTkpKSrC+BS5cuydXV1fqAp9dSuHBh64dD+oBR6bdIstlsio+P14oVK5SQkKC0tDQtWbJE3t7eqlmzpjUN/juu7jIlKcsv6szu656YmJjhyOr+/fsVGhpqffl//vnn+uKLLzR+/HgdPXpUy5cv1+nTp60fK5UrV5aHh4caNGggd3d36wdOYmKiBg0apAkTJigwMFCDBw9W7969re50wcHByp8/v3XWKX3dGzdutP4RrVGjhmJiYmSz2dStWzdVrFhRNWrU0Lx58xQWFmZt1+2ofPnyqlu3rhYtWqSFCxdq5MiR6tixozw9PbVv3z69/PLLWrhwoWrXrq0NGzboscce05IlSyRJ9evXl4uLixo2bKgmTZrI1dVVhQsXVunSpZUvXz5r0LnChQurVq1admNF+Pr6qmTJktYZQenKLWzat28vLy8vHT9+XJ988olSU1Ots31NmjTRww8/rCeeeEIFChRQu3btOOsNO9c78Hf58mVt2LBBX375pTZv3mydZSxevLhKlChhHWgrXry4Ll68aHdgJ/2HbHo4DgwMVHJyst2BKHd3d1WsWNEK1BUrVpS7u7vdNd758+fXP//8o3LlyikhIUH58uVTqVKl9Pfff9vdntPd3d06mI1/l/RRorPa167ulbVp0ya7wSWDg4N14MABqzvx5cuXrTPN6f+OPPDAA9qzZ48uXLggd3d3Va5c2eoVlP6dV6VKFaWkpFgHjKtWrWr9HnR3d5d05bfcgQMHVKxYMUlXDk5fPeK5q6urypcvb53pvJ27FOe19L/95s2bNWTIkGzHd0n/u4aEhCgwMFDt27dX48aNFRAQYPfboWjRonr99deVlJSkiIgI7du3T++8846eeeYZ6/ITPz8/5c+fX7///rv1XeHm5qaUlBRt3bpVgYGBGjlypFatWqW9e/fqt99+swbTq1ChgrZt22b3+6x69eras2ePEhMTVbhwYT399NMaP368/Pz8VKNGjeuOdfFvlO/6k+C/Kj1UBwQEyN3dXXv37lWjRo2UkpKivn37qkOHDmrUqJEkafHixRo6dKhq166tMWPG6MyZM3bd44YOHaqCBQsqOTlZzZs3V/Xq1fX999+radOmKlOmjEJCQvTXX3/p0qVL1pdySEiIfvzxR50+fdoaTfjaLqX33HOPFi9ebH0opStdUj/44APt2rVL+/fvl4+Pj6Kjo/XQQw/l4buHf4PTp09r9uzZmj17tvbu3SsfHx81btxYzzzzjCpVqmT94Ej/R+HAgQMKDQ1VuXLlJEndunXTyZMnNW3aNJUtW1bSlbOoFStWtEYGX7Vqle6//361bdtWf/zxh+Lj45WcnKxFixbpjz/+UEBAgCpUqGD9SE7/XGzevFkfffSRli5dqqZNm0qSDhw4oPj4eB09elQVK1bUs88+q86dOysgIEChoaFavHix6tWrp7lz5+rChQvy9PTU3XffrSVLlmjJkiXy8PBQSEjIbXs7jatVrFhRNpvNOshxtWnTpmnDhg3avHmzFZBfeeUVDRkyRM2aNVPZsmVVsmRJa9Tb9B9sNWrU0KxZs6zleHp6qkqVKvrhhx+sNh8fH5UtW1bbt2+XJNWtW1eFChVSVFSUmjdvrj///FMPPPCAfvrpJ+3Zs8caAHLChAn6+++/rYOIuDOkh83MDspdK33skvTbWV7t6q66R44cUf78+a3P6fbt29WlSxcdPHjQ+gHbvXt3vfzyyypcuLBKlSplheNKlSrJy8vLrgeMv7+/ChcurGPHjik1NVWlSpVSwYIFdeDAAaWkpFj/lpcrV04pKSnauXOnqlSpopYtW2rKlCkKDg7WAw88oN27d2v69Ol6/fXX5e3tLUkqUqSIVq5cqRdeeME6Kz569GgVL17c7iw5bl1O97Wre2Vd2wX82n3v7NmzKliwoGw2m5KSkhQTE6PRo0fL3d1dJUuWVMuWLfXaa6+paNGiqlmzphYuXKgTJ04oKChIjRo10pQpU5SUlGT9Lrv77rsVGxurhIQEeXp6qnbt2tbBzvT1VqhQQW5ubtZBoLZt2+r999/X119/rccee0zGGI0bN846GeLi4iJ/f38r8BctWlTSlfED3njjDZ0/f15eXl659C7fea7db9LDcr169eTt7W2dMMvsAHx6W6VKlZSWlmb13pKufE9t3LhRVatWVeXKlVWxYkUtXrxYTZs21bPPPqtjx45ZPffSv2fKlSunbdu2Wc9tNpvKlSunggUL6q233pKHh4e1/AMHDki68r0UGhqqP//8UxcvXrQG5atQoYJmz56thIQEFStWTIMGDdKDDz4o6colXcWLF8/tt9LxDP4Tzp8/b4wxJjU11aSmppq0tLRspz9x4oSZPn266dOnj9m3b58JCgoyr732mrl06ZIxxpiiRYuagQMHWtPXq1fPPPLIIyYpKSnT5f3999+mQ4cOxs/Pz/j5+ZlixYqZOnXqmPXr1xtjjBkzZowpX768+eeff6x5VqxYYWrUqGFefvllq+2NN94wfn5+5ujRo8YYYzZv3mxsNptZvXq1Nc3ly5fN8uXLzaJFi8y+fftu4F3CnSItLc1cunTJ9OnTx4SFhZno6Gjz3XffmeXLl5uHH37YTJs2zZr2l19+McHBwcbX19dUqlTJNGjQwEydOtUYY8yyZctM06ZN7fb1qVOnmsDAQJOammri4uJM+/btjbu7uylSpIgpUqSIqVGjhmnXrp359NNPzZkzZ8yZM2fMI488Ynr06GFX4/Lly43NZjM7d+40xhjz+++/m+DgYGOz2cyqVaus6WJjY03NmjVNsWLFTK9evczixYuNt7e32bFjh7WtWb0Ht7OEhATTpEkT06tXLzNt2jTTrVs306hRI9O3b1/TvXt307lzZ2PM/23nt99+azw9Pc25c+dMYmKiqV+/vt3fzZgrf2ubzWZOnDhhtc2ePdsULVrUnDx50mr74IMPjK+vr9m9e7cxxph169aZjh07mrCwMBMVFWUSEhLM/v37b/v3GDcmLS3N7Nu3zxw/ftx6nlOxsbGmRo0axtPT01SrVs28//77xpgr/141adLENG/e3Jw8edIkJCSYESNGGDc3N7N161ZjjDExMTGmevXq5ty5c8YYY1q3bm26d+9uzp49ay3/+eefNy1btjSHDx82xhjz4IMPmmeeecacOnXKmmbLli2mfPny5quvvjLGGHPgwAETERFhatWqZZo0aWLKlCljOnXqZI4dO2bNs2HDBrNr1y679wB54/Tp0zc8z99//22efvppExQUZHx9fc1zzz1n/T2nT59uihUrZubNm2cuXrxoYmNjTVBQkOnXr58xxpj169ebMmXKmLlz5xpjjPnhhx+Mp6en9Xsrffk2m81s3LjRGGPMwoULTf78+a3fmOlatWplnnvuOes343PPPWfKli1rnnzySfPggw+aoKAgs2DBAmv6vXv3mhkzZmRYDuylpKSYlJSUHE2bnJxsypUrZz744INsp0v/TNerV89UrlzZVK9e3RQpUsQULVrU+Pv7m/nz59tNt3nzZlO4cGFTsmRJ69/I9L9z+/btzSOPPGL33fTNN9+YGjVqmE6dOpm9e/eaEydOmO+//9707NnTzJo1yxhjzLvvvmv8/f3tMkBcXJw5ePBgjrb1dkHo/g/48MMPTf369a1/jK/2zz//mF27dpnU1FRjzJVQfvHiRdOlSxcTEBBgunTpYp599lljs9lM165drX/A27VrZ9q1a2eMMWbt2rUmODjYTJw40RhjMnwhHD9+3HTo0ME0a9bMCscrVqwwhQoVMosWLTLGGLNmzRpToEABs3z5cmu+y5cvm0mTJplixYqZdu3amcaNG5vKlSubyZMn2y1/xowZJjExMRfeKdwJ0vflN9980xQoUMB89dVXWf4jdeTIEdOwYUPToUMHc/nyZXP48GHz9NNPm1KlSpm9e/caY4wZO3asKV68uBXKRo4caUJCQowxxpw9e9Z07tzZ1K1b1+zcudOcOXMm0/X06tXLtG7d2vqHKb3OsLAwU7NmTdOkSRNz9913m5kzZ5qKFSua7777zm7+q+t//fXXTeXKlc1ff/1lN01aWpq17XeC1NRUExERYdzc3EzDhg1NZGSkiY2NNVu3bjWtW7c2L7zwgrW9aWlpZseOHcZms1k/Ah566CETGRlpLly4YC1z9+7dplChQubXX3+12lasWGFsNptZvHix1bZ8+XIzaNAgc+jQoTzaWvwbJSUlmTlz5pjw8HDj6+tr3N3dTVBQkGndurX55ZdfjDH/90M0MTHRTJ061Tz55JOmY8eOZs6cOdZyNm7caIKDg837779vtm/fbv7++2+zYsUKk5aWZg4dOmTc3NysA9DpypcvbwYNGmSMuRKWKleubDZt2mSMMSYqKsq0bNnS7qDyyy+/bKpVq2Ytp3///qZ27dpm2bJl5sKFCyYuLs6cOHHClC9f3jz77LPWfKdPnzZffPGFGTlypFm0aJEV7K/nTvqu+TfYt2+fGTJkiGnatKnx9/c3vr6+pnnz5mbGjBnWNOn72u7du8348eNNVFSUmThxovUdd/r0afPcc8+ZBx54wHz99dfm119/NYsXLzZxcXEmKSnJ9OjRw7Rs2dJuve+8847x9fU1xhhz7tw5ExAQYMaPH2/V5OLiYtatW2dNHx8fb9zc3KwDN4cPHzY2m+3/tXffYVFdWx+Af8NQpHcEBnDo0qSICiJWNKixYuxGsUQsMSoksddExUs0mmvXaEyssUUx18QSBBtKFBFRsVAFlSII0pn1/cE3J4wCYhITy3qf5z43TjtnDmf2OWvvtdcW9lMebA0bNoz8/f0pLS1NeO+PP/5IY8eOpblz59L58+cbdQ5VVVVxR08DioqK6LfffqNFixZRaGgoXb58WeF4tW7dmqZNm6ZwHXyW/P5i9OjRJJVKaeXKlXT16lV69OiRwmfJ/17bt28nFxcX8vb2ph49eih0yq1atYosLS2Fc0Z+T7R7927q2rUr2dvbk4aGBllaWtLUqVOF63VpaWm9g3a1ven3ORx0vwOOHj1Kvr6+9Pvvv1NycjJt2LCBdu7cSZ06dSJVVVVycXGhtWvXCj+ujRs3krGxMUVFRRERUWJiInXv3p1cXV2FwH3x4sXk4OBARDW9Xt7e3rRmzRoiev5ifOfOHdLV1VUIJNasWUNNmjShbdu2EVFNL1nr1q3J39+fBgwYQB06dBA+JyoqiqZPn05LliyhCxcuKAQg3Bi/Xfr160cTJ05UeOzPNLJpaWnk7OxMw4cPr/N5+Tl07tw5UldXp7t37wrPFRUVkVQqpVWrVhFRzbnp7OxMs2bNEvZx2rRpwmcsX778uR7a6upq2rVrFyUkJBBRTYZGixYthFFt+XvT0tJo+fLltGDBAjp58iRVVlY+t69Xrlyhw4cP09mzZyk8PJxcXV0pPDz8pY7Hm2rixIkUGBj4XKfJuHHjKDAwUOFGIjo6mvT09Cg+Pp6IagKT3r17U2pqqvCanJwc0tTUpCVLlgiPZWdn08aNGxVex95epaWltG3bNvriiy+IqOHgsaSkhEaOHEk6OjoUFRVFqampFBcXRy1btiQXFxeF186dO5fatGlDo0aNotmzZ5OJiYmwjfDwcHJ2dhbO19rbjI6OJgsLC6EjSP6agQMHUv/+/amyspKio6PJzc1NyND59ttvydvbm7Zv305EROnp6dSuXTuysrKiH3/8kYiIrl69SsOHDycjIyMSiURCps3evXvp/PnzL3XM+Dr758THx9M333wj3DfVdRzl58KOHTtIJBJRcHAwnT59mi5evEjBwcGkra0tdPAQEV27do3atGlDbdq0oSFDhlC7du2oTZs2RFQziKKkpKQwglx7Oz4+PsJ1TL4/Z86cIbFYLFy/PDw86LPPPhPOQxMTE1q8eDFVVFQQEdGXX35JIpGIli9fLjz20UcfUfPmzUlLS4tcXFwoIyODjh49Shs3bqy3I7oub3Iw9VfJs1DrU15eTmVlZXTs2DEKCgqigwcPUn5+Po0fP54cHR2pW7du9P7775OpqSnt3r1beN/QoUOpX79+CtkKz5JfX5csWUI+Pj6UnJxMRH+cr7XP26dPn1KvXr1o3LhxlJqaSp07d6Y2bdoI2WO3b9+mXr16kbm5Oamrq1O3bt2E9+bk5FBsbKyQKfQu4qD7HZCcnEze3t70yy+/0JEjR0hbW5vatGlDu3btory8PJo6dSo1a9aMzpw5Q0REn3zyCfn4+BDRHz/GI0eOkLq6uhBEREZGkoaGBuXn59Pjx4+pR48ezwU41dXVVFFRQTKZjAwNDWnUqFF07949Onr0KPXs2ZNEIhGFhYUJvVtJSUkUFhZGYWFhtG/fvjoDEPbmevjwIZ04cUJonOu6AYmLi2swnejevXt09erVF16c09LSSEVFhb7++ut6t0VEdPLkSRKLxcI5KP/cLl260NSpU6m0tJSIiLZu3UpmZmaUlpZGdnZ2QlYHUc3ogI+PDzk6OtLKlStp06ZNFBISQr6+vnTs2DEiqrnRnTZtmkJw35DaU0B+//136tSpExkZGZGXlxetXLnyT6Udvok2bNhA7dq1E46bvE04ePAgWVpa0rJly6iyslIYFX/vvfeEm8fFixeTk5OTQqp+VVUVRUdHK3SQsHdLaWkpBQcHC9kqL/LVV1+Ru7u7QtrrsmXLSCQSCdkwv/32G7m6ugopt0Q10xacnZ0pLi6OLl++TG5ubuTg4EAjRoygiIgIOnr0KJWXl1NiYiJ5eHgIndbybJipU6eSn58fERFdv36dOnfuLIx8p6am0qBBg8jW1pamTZtGQ4YMoTFjxpCPj4/Q5hHVtLmJiYlCO1YfmUxGVVVV73TQ83eSH8etW7eSgYGB0KFS13VI/tj169fJ2tqafv75Z+G57Oxs0tDQEALlyspK8vf3p5kzZypsy87OTugk9vHxIU9PTxo9ejRFRERQTEyMkF7erVs3mjBhAhUWFgrvv3DhApmYmFB0dDQR1WQxDh48mO7fv09ERNOmTSMLCwv69NNPae7cuTRz5kyytbWlHj16COdVbm4unTx5ks6fP//Cc02eHv0unmtVVVXC9aq2Z8+LjIwMhYyThw8fkomJCY0ZM4ZCQkJo0KBBdP78ecrIyKB9+/YpZGR99tlnFBgYSImJiURU00nSqlUr4d8Ndfz8/PPPZGNjQydPnqz3O5w7d450dXXpxo0bRFSTDeHn50eenp4K+3/s2DEheK9PY6a6vm046H4HFBYWkr+/P0VERNCDBw/IwsKC+vfvLzyflZVFbdq0EXrlP/vsM2rdujUR/RF0Z2dnk0gkEoIIedqRPJVt69atpK6uTt988w3l5uZSYWEhbdy4Ubjh3bNnD7Vs2ZLU1dVJIpHQmTNnaNOmTRQbG/tONr7vEvnfd/PmzWRgYCBMIWhsp0pubi4NHz6ctLW1ydDQkHr16vXCufqPHz8mZWVlWr58eZ3nl7yhl09rkI+Oyvepc+fOFBISonAD4efnR926dSNzc3NhjpM8WM/KyqLVq1dTmzZtyNXVlUaPHk0//fRTo6Y9vOiGt6qqqtEpn2+bmJgY8vHxEW4Cao94f/HFF+Tg4EC+vr5kb29P7u7uCtNTMjIyKDExsdHz39ib4dmbtLKyMpo/fz5du3btudfJb/Cffc+KFSvIzc2twVE4+Xt27NhBTk5OQprs3bt3ydfXl8zMzIR2Y+zYsdS3b1/atWsXffLJJ0KKsEgkUqgP8d///pfmzp1LHTt2JAMDAyF4CgoKos6dOwvnamZmJrVq1YpGjBhBRDXtS69evahLly7C/uXk5NCMGTOob9++tHr1anrw4EGD7URdaaLs71FdXa3QhsuPdVJSEkmlUoWpBvUpLy+nZs2aCecLEdGmTZtIJBLRhg0biKims0VPT48uXLhABw4coIkTJ9L7779PIpGIhgwZQk+fPqUbN27QvHnzaPr06RQUFESqqqo0aNAgKisroyVLlpCXl5dCpsPnn39Onp6elJSUREREs2fPJhcXF6GeQGFhIa1fv55at25NPXr0oAsXLlBWVlajRrC57X0x+bmSmJhI77//PqmpqZGJiQmNGDFCSNHOz8+nvn37kkgkErJDif74HWdkZNC6deto4sSJJJVKydramvbt20dENZmuNjY2dOrUKYXt1SUtLY2kUikdPHiQiGraukOHDtHs2bNpw4YNJJPJqFOnTkI9Fbn8/PxGpYa/aPvvAg663wFVVVXUu3dvGj9+PFVWVpK3tzdNnTpVeL6yspL69OlDwcHBRFST+m1paanwI0pPTyc1NTVau3YtVVVVUUVFBRkbGytcIGbMmEGenp4klUpJV1eXHB0dFdKiMjMzXyrViL255DchRH80sgkJCSSVSunw4cPC6yoqKhTmOd+6dYsMDQ0V0rAHDRpEbdu2pejoaHr8+HGj05MsLS1p5MiRQq9+XY39kydPyMnJSWHO0927d8nMzExI4Zbv34EDB0hNTY0MDQ0V5rg19iLyLvbq/lUZGRnk7e0tFIJ5dmrJhQsXaPny5bR//35hNIeP8bunS5cuz82JbsiuXbvI0dFReE9DI5BRUVHk6upKRkZGZGhoSCoqKiQSiRRGG5cuXUoikYg8PT1p4MCBtGzZMqGQ57OBh/zfc+bMIYlEQkQ1o0cmJiYUGBhIa9eupX79+pGPj48wV7KyspK2bNlCmzZteomjwl6Ff6KQlbOzM2lpaZFIJCI7OzshoyImJoakUimpqamRu7s7DRw4kJYuXUrHjh2j7Oxshc4UeQfMwYMHycnJiSIjI6m4uJi6detGdnZ2tGvXLvrmm2+oRYsWtHTpUuF9v/76K40ZM+alp9vIZLJ3ru2t/X0b+u6VlZWUl5dHkZGRNHDgQBozZgzFxcUJzz958oSGDRtGHTp0oJs3b9LZs2fJx8eHWrVqJWRlTZgwgZydnZ+7r0pMTKT27duTv78/hYSE0MKFC8nBwYEiIiKIqCaQNjc3Vyge2xAdHR1SV1cndXV1EovFJJFIqHPnzrRr1y5hX1/22LA/8JJh7wCxWAwLCwukpqZCWVkZZmZmKCoqQmlpKdTV1aGsrAxTU1NkZGSgoKAAXbt2xaxZs7BmzRpMmjQJqqqqiIiIQEVFBW7fvo2ysjJoampCTU0NZ8+exfDhwwEAS5cuxZAhQ5CdnQ1HR8fnls2RSCT/wrdn/xT50klA3WtqOjg4QCaT4eHDh1i5ciVWrFgBIsLgwYMxf/58aGtrw8jICI8fP0ZKSgocHBxw48YNREZGYt++ffD39wcRoXXr1g3uh3wZsC5duuD48eNISEhAu3btFJbLSExMxKNHj9C5c2dMnz4dCxYswKNHj9ClSxfs3LkTHh4eGDp0KAAIy1d07doVMTExMDU1haWlpfBZzy7DUd+yQbzO6MszNDSEVCoVlqupvTyOSCRCmzZt0KZNG4X3vKnrkrMXu3PnDg4cOIDhw4fD3NxceLz2WvZAzfrCV69exbFjx5CQkABra2uMGTMGDg4OAGqW2NLQ0EBSUhK8vb2F5XbqYmpqChMTE5iZmeHbb79FVlYWDh48iNzcXKSnp8PKygoSiQR6eno4ePCgsNyg3JMnT6Cjo4Po6GgYGRkJy9FduHBBWMbS19cXp06dwrJly7Blyxa0aNECoaGhsLOzA1CzfOfo0aPr3L/q6moAzy8dxV6NZ49zcXEx4uLiEBMTg8LCQgwbNgweHh4QiUTCsnApKSnC/VZdZDIZxGIxXFxccOrUKUyYMAFdunTBt99+i5KSEqSmpsLa2hoqKirQ1dXFuHHjMGvWrDo/69GjR7h16xb8/PwAAFevXoWWlhakUik0NTWxbt06bNq0CbNmzYKGhgaGDh2K8ePHC+/v2rUrunbtWuc+yq/xdV3L3rR2l2oGHRu9HCBQ81urvSSX/P9rtx/p6enQ1NQU1h+Pjo7G3LlzYWxsDLFYDF1dXaSkpGDo0KFYsmQJgoKCcPnyZZw4cQKbNm2Co6MjAOC7775D7969sWfPHkydOhVOTk7Yu3cvxGKx8HcoKyvDokWLUF1djR9//BFmZmYAgFWrViE7OxsAYGVlBR0dHWRlZSnco9X13cRiMdatWwdVVVW4u7vD3t7+uddpa2s36vi+aefDP+bfi/fZP2nDhg3k6+tLBQUFFBoaSj169FCoKhkeHk6dOnUS5qQtWLCApFIpBQQEUIcOHWj+/PkUGBhIffr0EXq6YmNj37py/qx+DaVr1lZSUkInTpyg5cuX05EjRxSe8/b2prZt29L48ePp2LFjtGfPHlJWVqb//Oc/Qmq3ra2tUOhqzZo15OrqKqS6yXvTG9q+vLf/4sWL5O7uTs7OzvTjjz/So0eP6MmTJ3Tx4kXq06cPffLJJ0REQqGiXr16kZubG3300Ud05cqVP3uYGGN/UX3puj///DNZWFgImSby6R+nT5+mNWvWCKN7x44dI39/f+revTuFhoZS586dydvbW8i8un79OgUEBNCnn35KRA2nwebm5tKwYcOE1Trk2w0ODhYKWD18+JCsrKxo5MiRwjUxNzeXtm3bJowifvHFF+Tr60v6+vqkp6dHQ4cOVVi6rjE4XffVeR0LWRHV1A9wc3MTRkaHDx9OHh4eCq85d+4cffnll/Tw4UNKSEigVq1akaenJ+no6JCjo+Nzo5wvml7wtlYMb+g7VVVV0e3bt4UVeuRF4urz6NEjOnnypLCiwI8//kgmJiakq6tLAwYMEFYXuXnzJvn6+pKRkZFQTDgzM5NGjx5NdnZ2RFSTTaOsrKyQCVpcXEwDBgygcePGERHRiRMnSElJSWEJQKKa6uTTp08X/h0eHk7a2to0cOBAysnJISIiZ2dnGjx4sJD1J29fKysrX9imVFdXN+p1rHF4pPsd4ezsjMrKSqSnp8PV1RWxsbF48OABrKysANSMQqenp+PWrVvw8vLC3Llz4e/vj/3790MikWD48OGYM2cOlJX/OGVeNOLI3kzy3tDq6mqFHmCRSNTgaMr8+fORl5cHsViM6OhoqKqqIjMzE5MmTcLEiROhp6eHFi1aYPv27QgNDcV7770HoKYn+NixY+jbty/s7Ozg6OiIK1euAAB0dXWhra2NjIwMtGjRQhhBboj8+VatWmHHjh349NNPMW/ePIjFYqSnp0NfXx/9+vXD7NmzAdSMIvn7+8Pf3/+Fx4aIhGPBGHse/f/oEVDzO/kzv5X6fuOZmZkoLS3F0KFDkZKSgjZt2uDUqVM4c+YMwsPD0atXL2hqakIikSAsLAy9e/cGABQVFeHjjz/Ghg0b0LFjRxgaGsLU1BS3b98W9rM+enp6sLCwwKlTp4TH1NTUsGLFCjg6OiI8PByff/45Nm/ejClTpiA4OBgFBQXIzs6GqakpPv/8cwDAsGHD0L59e0gkEtjY2NS7vfoyZQAezf6r6ju2z15XMjMzoa+vD01NTQA1o8dubm7o1asXVFRUhOzAp0+fomvXrpgzZw4sLCwAAJ9//jm2bdsGV1dXuLi4wMXFBYcOHUJeXh5MTU3rzKqQ/9vDwwObN29GZmamMMo4duxYZGVlYcyYMYiPj8eiRYswePBgBAUFwcPDA8nJyXj8+DE6duwIdXV12NvbIywsDGpqanBxcRGyJWqTf1d5lkTt0Vvg7T3P5N8xLS0Nly5dQlJSEs6dO4e4uDjk5+ejVatW2L59O/T19aGiovLc+3/77Tds2LABw4cPx9SpU1FQUAALCwv07NkTKSkpOH78OMrKyjB27FgsXLgQe/bsgUQigbGxMR4/fiy0RxKJBCEhIdi2bRvu3bsHPT09AEBubi50dXUBAJqamqisrER5eTnKyspgbW0NVVVVJCUlwc/PD1VVVVBWVsZ7772H77//Hjk5OaiqqoKOjg7at2+P8vJy5Ofnw8jICP/9739hYmIijFLX1b5mZ2cL26moqBAy/OrLbGB/0r8X77N/UkZGBnl6etIPP/xAMTExZGpqSjt27BCev3fvHn333XcKo9/s7Sbv9X1RQTP5XOSbN2/SwoULyd/fn1xcXGju3LnCXDOimvWjNTQ0KCAgQCh0tmTJEmrevLkwF3Hx4sXk7OwsVMEnqlnz0dPTUxiFkhdyIapZHsXPz08YlZbvc2MLucidPXuWjh49KqwJWZ/6qosyxv4aeabMizJVCgsL6fjx47Ro0SJasGCBMMKXl5dHM2bMIH19ffLz86P4+Hhh9DAmJoa0tbUViqnJZDLau3cvjRkzhjw8PKhJkybCMpeVlZU0Y8YM8vb2btS+yzNu5NuTj/ps3ryZRCKRsEzX/fv3adOmTbR+/Xq6dOlSg6NlXOPh9fA6FbJKT0+nZs2aCfNn5TIzM0lHR0coYFVVVUUbN26kKVOm0OrVq+nKlSsNXsf5PKuRkpJCIpGItLS0SCqVkpaWFpmamlJ8fLyQoZKZmUkzZ84kPz8/6tmzp7AeOVHNaicikYjatm1LV69epYKCAho5ciQ1adJEWL6PqKYAnpmZmVAPZuLEidS+fXth1ZHq6mrKyckhDQ0NOnHiBBER6evr04oVKxSW6ZJKpbR48WIiqimcaGFhQcuXLyeiP9qgqqoq2rp1K/Xr148mT55MCQkJwlrp8m3VVlZWRtHR0bRq1SoaOnQoubm5kYGBAWloaAiZP3y+vDo80v2OMDQ0hLW1Naqrq+Hl5YVt27YpjFRbW1vD2tr6X9xD9k+T9/rWzl4AgLi4OCgrK+POnTuYNm0aOnfujBUrVuDnn3/G2bNn0atXLxgZGWHFihVISEjA2rVrYW5uDi8vL4jFYvTo0UOYzz958mTExcXhwIEDGDt2LFxdXYUeWDkXFxeUlJQgKysLANCyZUusXr0ajx8/hpOTE8aMGYMxY8ZAQ0MD/fv3x61bt3DmzBmEhoYKvcJAzWiFTCYTRhLkvfUymQxt27ZV+I4ymazOUbi3tYefsX9CdXU1YmNjceTIEcTFxaGkpARdunTB+PHjIZFIFH5fZWVlePTokcLjWVlZmDx5MpKTk2Fubg5VVVX88MMPCA8PR//+/TF//nyUlZUhKSkJ7u7ukMlkAAB3d3cUFxfj/v37cHV1BQAsXrwYkZGRaNGiBUJCQpCTk4Pw8HDk5eXB0NAQFhYWKC4uRlZWlsL88NrkbYm+vj6ys7ORmJgIU1NTod0YMWIEPDw84OzsDAAwNzfH2LFj6z0+VGuUk0eP/rrax5MamJdfVVWFJ0+e4Pz589i+fTu0tbUxYcIEtGzZEiKRCEVFRVi6dCmKiopw9epV5OXlITQ0FJMmTcJPP/0EMzMzmJmZwcnJSahhQ/8/On79+nVMnDgRRAQXFxcEBwdjx44dSE1NBQC4urqirKxMuL41lFVhaWmJ/Px8XL9+XWH+rUQiwZEjR6ChoQEiglgsxrhx4xo8LlRr9J4zs2pYWlri+vXrcHBwgFgsxldffYVNmzbB3d0dQE39hdDQUBQXF6NHjx5QUVHBxIkTkZ6ejtDQUDg7O6NJkyYIDAxEixYtAABTp07FDz/8gObNmwvbsbW1RWlpKa5fvw4vLy9YWFjgt99+Q3Z2NvT09KCkpITCwkLo6OigoKAAABASEoIVK1agqKgIgwYNwvr166Gjo4Nu3boBALS0tODr64uysjIAf9yriMVijBo1CqNGjarzO9duZ0pLS7F8+XKsWrUK7u7ucHV1xdSpU+Hm5gYbGxthHjqfL6/QvxTsM8ZeEfkcnIZGa8vKyig2NpbS0tJo27ZtNG3aNGG+T/fu3UlLS4vGjRtHW7Zsobt379KTJ08oPj5eoWp4dHQ0+fj40Pr164moZo6/s7OzsF6oTCajiooK+vLLL8nKyoqIiO7cuUPm5ua0c+dO4XMKCwvJzc1NqLaZnJxMSkpKCvOqt2zZQu3btyddXV2yt7en6dOnU25ubqPnnslkMh69ZuwVO3ToEHl5eVH//v3piy++oK+//pqWLFlCCQkJVFhYSEuWLKELFy5QQEAAKSsrk729vbD2NFHNPMaIiAiFpfpmz55Nbdu2Ff79xRdfkKurq/Aa+e9aV1eXVq9eTTKZjFJSUkhTU5M2btwotA+bN28mJSUlYRnLAwcOkJubm5BhU1f7IH8sMzOTYmJiGrV0X+0RfVY/eZv8Mms213dcaz+WlpamMFf+9OnT1L59ewoKCqKBAwfSuHHjKCAggBwcHITR6KioKGratKnCyhq3bt0iR0dHWrlyJRERrV69mgwNDYnoj/OitLSUBg4cSH5+fkKVaSIiAwMDCg0NFf7dvHnzepevrP3diGqWX62dCfbs93v2PXyu/Xnff/896ejo0NOnT4mIaNu2beTn56cwUrxlyxaysrISVlRp2rSpcI9DVLOcrpmZmUIGRGJiItnY2NDGjRuJiGjv3r2kra1NkyZNEuZZh4aGkouLi5Cd8+TJE1q7di15eHiQhoYGtWjRolFLzck1Nkvv6dOnjV7ei/39eKSbsddI7ZHahkZC6Jme7Nrqm4Mj7znfu3cvJk6cCKBmnmFkZCTU1NQwZcoU6OjooFevXjh27BgcHR0VKua6u7vj8uXL+OyzzxAbG4v09HShpx+oyZbQ1NQUevjl1VvV1NSE6tO2trbQ0tJCdna28D11dHSgrKyM69evo6SkBFZWVlBTU0NmZiY8PDwAAKNHj0ZQUJDCyHZtJSUlSEhIQGxsLC5fvoyoqCgMGTIECxYsQJMmTf703FLGWONkZWVhypQpGDRoEBYvXgw1NTWFjJLCwkLMnj0bnp6eGDhwIDZs2ICoqCiMGzcO5ubmCAkJgaamJkJDQ3Hv3j38+uuviI2NxZkzZ5CWlobU1FRIpVKYm5ujoqICd+/ehYuLC6qqqqCqqgo7OzskJiZCJBLh/v37aNq0KTQ1NSESiZCVlYX9+/eDiBAXF4e2bdtCT08PJSUluHz5Mjp27CjM96VamTK1Rxobu/rGi2pfvMuo1mj0s21ydXU1UlJSYGhoCH19fVRWVj43r7b2cc3JycG1a9egr68PT09P7Nu3D5MmTUJ5eTm6du2KhQsXwtnZGU2bNkVlZSVOnz6NLVu2oHfv3rh//z7mzZuHGTNmICgoCACQl5eH9u3bC58vkUjg5uaGpKQkADV1cR4/fozHjx9DX18fANCkSROkpqaiXbt2QuXo5cuXo7KyEhkZGcjNzYWRkRGUlJRw+fJlFBcXQ0dHR7jO1z7X5N+trhHL+kbz+Tz7aywtLSEWi5GYmIjWrVtj27ZtkEgkWLVqFc6ePYvbt2/j4cOHEIvFyM3NhYODA+zt7YVzAqip5m1ra4uLFy9i5MiRAAADAwPY2NggPj4eAIQ6Dr/++isKCwtx48YNZGZmYsGCBXB1dUV1dbWQfTFo0CAYGBjUu8/yKuPPauy5oKGh8RJHiP3dOOhm7DVSO1iWyWTIz8+Hvr7+cw1qQ0FkdHQ0vv76ayQkJMDDwwMfffQRunXrBiUlJdy5cwczZ87EyJEjERERgatXr+LmzZuIj48X0pxsbW2FAh213b17FxMnToSTkxNmzZoFDw8PzJkzB+np6ZDJZDA2NoaxsTEOHTqEKVOmQCqV4unTp9i6dSu6d+8uLJmiqqqK33//HUVFRdDR0QEA9OvXD+bm5hCJRFBTU8OTJ0+eS3uXB9wymQzV1dVQUVHBxo0bERISAj09PWhoaMDGxgYeHh6YMWMGOnfujCZNmvylvwdjrHHU1NSQkZGBPn36CJ1stdszXV1dSCQSlJaWYuTIkTA1NYWNjQ2SkpKwfft2vPfee7CxscFvv/2G+fPnQyaTwd3dHUOGDMGaNWtw6dIlSKVSmJmZQVNTE7du3YKLi4uwjdatWyMpKQklJSVwd3eHj48P5s2bhx9++AGFhYVo164dtLW1kZ6eDgDw8vLCkSNHhKkwdRVOysrKws2bN+Hu7g5DQ8MGU5jZi3Ehq8YVsqqrw6H28WN/jfx3bGJiAkNDQ8TFxaF169awtrbGtm3bUFJSAk9PTwQHB8PNzQ0SiUT427Vs2RIXL14Ugt8mTZrAw8MDV69eFT5fS0sLxsbGwlKG8mVGmzVrhr59++LWrVvo0KEDXFxcACgGzPKAW94x8+y9H3e0vNk46GbsNVFWVob9+/dj586dSExMhIqKCtzc3BAQEIDBgwcL820qKytx5coVJCQkoGnTpujVq5fwGbdu3UJoaCicnJywcOFCnD59GkFBQdizZw969OiBW7duITs7G0uWLIFIJIKHhwfWrl0Le3t7ZGVlwcPDA82bN0d1dTWKi4sV9u/7779HcXEx5syZA1tbWwA1NykVFRXIzs6GRCKBra0tTpw4gZCQEPj4+OD8+fMQiUQYPXq0sEbp9OnTYWhoqBAQz507V2FbysrK9d7g1r5h6dKlCw4fPgwPDw80bdq0zhsVxtirZ2hoCHd3d3zyySfo3LkzlJWVYWhoiHbt2sHW1hZGRkawsLCAubm50NkGAP7+/jh9+jRu3LgBGxsbLFiwAJqamti1axf09PRw48YNbN68GYmJifjggw8glUqhpaWFgwcPon///rh9+zacnJzg7u6O9evXIzs7G7a2tli1ahW2b9+OrKws+Pv7o2PHjkJARUTQ1dVVyJyJjo7GlStXcPHiRSQkJOD+/fuorq6Gubk59u/fD0NDQw56/qLU1FTY2NhAU1MTRkZGyM3NhZaWFk6ePAkDAwNYWlri/v37mDVrFqKjo6Gnp4eRI0figw8+AFDzd9u7dy8yMjJw4MABNGvWDJ988glWrFiBjRs3CvNsp0yZgnnz5qG8vBxaWlqwsLBAQUEBCgoKoKenB5lMBmtrazRp0gQpKSno0qULtLW1cfjwYUydOhUikQhEhKtXr2LMmDFo0qQJdHR0YGRkhHPnzsHPz084F+bPnw8bGxscPnwYEokEH330EaytraGlpQWgJnjq1KmTcAzKy8tx8eJFXLlyBbGxsbh27Rru37+PsrIyTJo0CcuXL3+uw5n9veR/O319fVhZWQmrpVhaWsLOzk7o+JErLS1FdnY2zMzM4OXlhb1796K4uBi6uroQi8Vo1qwZ1qxZI7xeU1MTI0aMENY7t7S0hK6uLh49eoSAgAAEBAS8cB+55sPbiX/ZjL0GKisrsXLlSuzatQtdu3bF+PHj0bx5c8TFxSE5ORlFRUUwNDRETk4ORo0ahfj4eEgkEigpKWHLli04dOgQZDIZ9uzZg7y8PKxcuRKGhoYYNGgQqqursXDhQnTv3h0FBQXQ19dHVVUV1NTUUFVVBVtbW+jq6iI5ORmBgYGQSCTQ09PDzZs34ePjIwS/5eXlUFdXR25uLmxtbbFx40ZkZmYCAG7cuAGJRAJTU1N06tQJAQEBSEtLg729PRYtWiTcDAFAcHBwncfg2bSphm5w5RckW1tboQOAMfbvOnDgACIiInDmzBkYGRmhpKQEc+bMQevWrfHTTz/Bz88PMTExKC4uFtIcTUxMIJPJUFRUhKKiIqioqMDW1lYYfdy3bx8ePHiAxMREAICNjQ0mT56Mzz//HGpqajA1NUVkZCSCgoKgpaUlpIEbGRlh+vTpz+1j7QJVtY0dOxYmJibw8vLCtGnT4ObmBltbW+jr63Ow/TfhQlZcyOp1oq2tDQcHB1y+fBkAMGDAAOzYsQOTJk3CsmXLYGFhgeTkZBw6dAiampr4+OOP0bx5czx48ACPHj0SOu0++OADODs7K0wN7N69u7AdFRUVNG3aFOfOncODBw9gampabzvE3nL/7BRyxlhdfvrpJxKJRLR27VqhqEddhg4dSkFBQUKhj9zcXJJKpUIxMx8fH2HZB7mDBw+Subk5paSk0C+//EJSqZR+//13IvqjIIyLiwuNHz+eSkpKiIjI19eXpk+fTkR/FHG5fPkyde/enaRSKVlYWFDPnj0pIiKCwsLCKCYmRthWy5YtFQrS1EVeNIYx9nbKzs6mnJwcOnnyJDk6OlJYWBidPHmSlJWVhWVyiIj27NlDJiYmdOfOHaqoqKBFixaRvr4+TZo0iQYNGkRDhgyhjz76SFjuSy45OZny8vIa3Ad5UbPGFOp60dKJ7O/HhazYv6m6upq+/vprsrS0JKKa9uLnn38mLy8v8vX1JXNzczIwMKBu3boJ9ziVlZWUlJTUqM+XF7ojqll2cOTIkZSamvpqvgx7I/BIN2Ovgf/85z9o3bo1JkyYoPB4enq6MMLTsmVLREVF4fjx4ygoKMDevXuRlZWFtLQ0HDhwAOPHj4e6ujpKSkpQXFwspLcZGxsLcyBbtWoFFRUVREZGwsvLC0pKSjh+/DhSUlJgZ2eHoqIiqKurQyqVCvOR5Kl2np6e2LBhA06fPg0TExP4+PgopIkCgJOTE4Ca+d8AUFFRAWVl5ed6dHleEmNvN1NTUwA16eOtWrXCjRs38Nlnn6G6uhphYWEIDw+HiooK1q9fj7Zt2woZKzNnzoS6ujpOnz4NJycnDBo0CK6ursI8cTl7e3vhv+n/R5ieHT16maJmnNL7z+NCVuzfpKSkBBMTE2RmZuLRo0cwMTFB9+7d4eXlhbNnz0JPTw8eHh4K54OysrJwn1NbXSPXtc+JiRMnCgVs2buLrzKMvQaSk5Px4YcfoqysDGpqahCJROjRoweioqJQXl4Of39/fPzxxzA0NISrqyuMjIxgbm4OJycnREREwM/PD0BNhfHff/9dmCsH1BRoUVVVhaamJgwMDDB+/Hh8+eWXyM3NhZubGxISEhAQEIBr166hoqICAPDhhx8iKSlJuJDI090sLS2FdUrlahf8sLS0BAAUFRUBAFRVVf+R48cYez1kZmZCV1dXKDz0yy+/4JdffsHMmTNhZGQEAAgICMBXX32FS5cuoW3btoiIiBDer6ysjLCwMISFhTV6m7z29ZuFuJAVe0106NABx48fF6azAEDTpk3Rv39/hdfRC4ooctvDGoODbsb+ZeXl5TAxMUF+fr5Cw75+/XpYWVlh1apV+N///ofs7GyYmJhg8uTJWL16dZ0Xgf79+yMmJgZhYWHYtGkTKisrsX37dmFkGgAmT56MZs2aITw8HCdPnsTHH3+MPn36oHfv3sIcpcDAQAQGBta5vzKZDMAfF5naFxsNDQ3ExMQIRdMYY+8O+XzY5s2bIz8/H8nJySgrK0NwcDAmTZoEkUgEY2NjmJqaYt68eUIgVRf5El71LYHI3lxcyIq9LszNzWFubv7c40QE4I9zlefZs78DtyqM/cuUlZXRvHlzXLt2DY8fPwZQE9jKR41dXV3x9OlTqKmpwcXFBWfOnEFRUZFwEbhz5w7WrVuH9PR0+Pv7Y/78+bh37x5at24NGxsb5Obm4uuvvxbSJ9XU1DBgwABcunQJ169fR0hICHbv3o2mTZsqpL5VVVXVub8vugnmgJuxd5O2tjamTZsGQ0NDuLm5YcGCBYiKikJ4eLgwQujp6Yn4+HghXbyyslLoyKtNLBbXOTWFvT3khawSEhIA1BSykslkmDRpkrAU5c2bN7Fq1Srs27cPABQKWcl98MEHiIyMFAIleSEredFOeSGrjIwMPHjwAADqPOfYu0l+3tTW0LKsjP1ZPNLN2L9MLBajT58+CA4OxtGjRzFu3DgoKSkJFwJjY2OkpqYiPz8fYWFhiIyMRKdOndCvXz+kpKTgxo0bsLCwENKhevXqBWdnZ9y9exd2dnawsbF5bpunTp1Cbm4unj59igsXLiAmJgarV69WSJ3jOY6MsZc1YMAADBgw4LnHZTIZxGIxrK2tERkZicLCQhgbG/Myf+8wdXV1NG/eHEePHgVQ08H8zTffYM6cORg8eDDS0tJQVlYGb29vYVlJLy8vXL9+XWFOf7NmzdCsWbPnPr+6uhpAzTXW1tYWeXl5KC8vB8Aj2ewPHFyzf4qI6uriYYz9owoKCjBixAicOHECy5YtQ/fu3WFgYID09HTs378fp06dQnh4ONq3b4/8/HysW7cOZ86cgbGxMQICAtCpUydhZLwxoqOjsWzZMty7dw+enp4YMmQIevbsyfPVGGN/mXw+rHz5HPljSkpKuH37NjIyMtChQwdubxh27dqFYcOG4cGDBzAxMQEAPHz4sN5CVvXhJZgYY687DroZe00UFxdj+vTpiI2NRXV1NVJTU6GkpCQUUatvjnV9Gir8wTcojDHG/m1ZWVm4ceMG/P39Gyy8+aJCVowx9rrjoJux18ytW7eQnZ0NqVQKqVRa7+tkMhlkMtlLLYvDGGOMve6eLWTFGGNvOg66GXuN1NWb/2y1cMYYY+xtwaPYjLF3AQfdjL2G+CaEMcYYY4yxtwMPnTH2GuKAmzHGGGOMsbcDB92MMcYYY4wxxtgrwkE3Y4wxxhhjjDH2inDQzRhjjDHGGGOMvSIcdDPGGGOMMcYYY68IB92MMcYYY4wxxtgrwkE3Y4wxxhhjjDH2inDQzRhjjLGXJhKJcOjQoX97NxhjjLHXHgfdjDHG2Btq1KhREIlECAkJee65SZMmQSQSYdSoUY36rKioKIhEIhQUFDTq9dnZ2ejevftL7C1jjDH2buKgmzHGGHuDWVpaYvfu3SgtLRUeKysrw86dO2FlZfW3b6+iogIAYGpqCjU1tb/98xljjLG3DQfdjDHG2BvMy8sLlpaWOHDggPDYgQMHYGVlBU9PT+ExmUyGpUuXwtraGurq6nB3d8e+ffsAAKmpqejUqRMAQF9fX2GEvGPHjpg8eTKmTp0KIyMjvPfeewCeTy/PzMzEkCFDYGBgAE1NTXh7eyM2NhYAcPXqVXTq1Ana2trQ0dFBy5YtERcX9yoPC2OMMfbaUP63d4Axxhhjf83o0aOxdetWDBs2DADw7bffIjg4GFFRUcJrli5dih9++AHr16+Hvb09oqOjMXz4cBgbG6Ndu3bYv38/goKCcOvWLejo6EBdXV1473fffYcJEybg7NmzdW6/uLgYHTp0gEQiweHDh2FqaorLly9DJpMBAIYNGwZPT0+sW7cOYrEY8fHxUFFReXUHhDHGGHuNcNDNGGOMveGGDx+OmTNnIi0tDQBw9uxZ7N69Wwi6y8vLsWTJEpw4cQK+vr4AABsbG5w5cwYbNmxAhw4dYGBgAAAwMTGBnp6ewufb29tj+fLl9W5/586dyMnJwaVLl4TPsbOzE55PT0/Hp59+iubNmwufxxhjjL0rOOhmjDHG3nDGxsbo2bMntm3bBiJCz549YWRkJDx/584dlJSUoGvXrgrvq6ioUEhBr0/Lli0bfD4+Ph6enp5CwP2s6dOnY+zYsfj+++8REBCADz74ALa2to34Zowxxtibj4Nuxhhj7C0wevRoTJ48GQCwZs0aheeKi4sBAEePHoVEIlF4rjHF0DQ1NRt8vnYqel0WLFiAoUOH4ujRo/jf//6H+fPnY/fu3ejXr98Lt80YY4y96biQGmOMMfYWCAwMREVFBSorK4ViZ3LOzs5QU1NDeno67OzsFP5naWkJAFBVVQUAVFdXv/S2W7Rogfj4eOTn59f7GgcHB0ybNg2//vor+vfvj61bt770dhhjjLE3EQfdjDHG2FtALBbjxo0bSEpKglgsVnhOW1sbYWFhmDZtGr777jvcvXsXly9fxjfffIPvvvsOANCsWTOIRCJERkYiJydHGB1vjCFDhsDU1BR9+/bF2bNnce/ePezfvx/nz59HaWkpJk+ejKioKKSlpeHs2bO4dOkSnJyc/tbvzxhjjL2uOOhmjDHG3hI6OjrQ0dGp87nFixdj7ty5WLp0KZycnBAYGIijR4/C2toaACCRSLBw4ULMmDEDTZs2FVLVG0NVVRW//vorTExM0KNHD7i5uWHZsmUQi8UQi8XIy8vDhx9+CAcHBwwcOBDdu3fHwoUL/5bvzBhjjL3uRERE//ZOMMYYY4wxxhhjbyMe6WaMMcYYY4wxxl4RDroZY4wxxhhjjLFXhINuxhhjjDHGGGPsFeGgmzHGGGOMMcYYe0U46GaMMcYYY4wxxl4RDroZY4wxxhhjjLFXhINuxhhjjDHGGGPsFeGgmzHGGGOMMcYYe0U46GaMMcYYY4wxxl4RDroZY4wxxhhjjLFXhINuxhhjjDHGGGPsFeGgmzHGGGOMMcYYe0X+D6Hn1uwpaxJcAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "keys = list(finals_gpt_before_metrics.keys())\n",
    "\n",
    "metrics_before_values = list(metrics_before_finetuning.values())\n",
    "metrics_after_values = list(metrics_after_finetuning.values())\n",
    "\n",
    "# Plotting\n",
    "plt.figure(figsize=(10, 6))\n",
    "\n",
    "# Plotting bar graphs\n",
    "bar_width = 0.35\n",
    "x = range(len(keys))  # X-axis positions for bars\n",
    "plt.bar(x, metrics_before_values, width=bar_width, label='Metrics Before Finetuning')\n",
    "plt.bar([pos + bar_width for pos in x], metrics_after_values, width=bar_width, label='Metrics After Finetuning')\n",
    "\n",
    "plt.title('Comparison of Metrics Before and After Finetuning')\n",
    "plt.xlabel('Metrics')\n",
    "plt.ylabel('Scores')\n",
    "plt.xticks([pos + bar_width/2 for pos in x], keys, rotation=10)  # Setting keys as x-axis labels\n",
    "\n",
    "plt.legend()\n",
    "plt.grid(False)\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e04ada88-30ab-4598-be00-adde8bc30e30",
   "metadata": {},
   "source": [
    "# Comparing the performance of model before and after finetuning using GPT based evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "f9e8aa6e-0832-4f69-83a6-1991bb80651e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "gpt_before_values = list(finals_gpt_before_metrics.values())\n",
    "gpt_after_values = list(finals_gpt_after_metrics.values())\n",
    "\n",
    "# Plotting\n",
    "plt.figure(figsize=(10, 6))\n",
    "\n",
    "# Plotting bar graphs\n",
    "bar_width = 0.35\n",
    "x = range(len(keys))  # X-axis positions for bars\n",
    "plt.bar(x, gpt_before_values, width=bar_width, label='GPT Metrics Before Finetuning')\n",
    "plt.bar([pos + bar_width for pos in x], gpt_after_values, width=bar_width, label='GPT Metrics After Finetuning')\n",
    "\n",
    "plt.title('Comparison of GPT Metrics Before and After Finetuning')\n",
    "plt.xlabel('Metrics')\n",
    "plt.ylabel('Scores')\n",
    "plt.xticks([pos + bar_width/2 for pos in x], keys, rotation=10)  # Setting keys as x-axis labels\n",
    "\n",
    "plt.legend()\n",
    "plt.grid(False)\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a51274eb-00c9-48ad-91b1-c965bab8830e",
   "metadata": {},
   "source": [
    "## Converting the ground_truth and before and after finetuning extractions in the desirable excel format for human level evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "731b6fa6-681a-4f75-8c5e-534e761f0f9e",
   "metadata": {},
   "outputs": [],
   "source": [
    "urls=list(true_data.keys())\n",
    "columns=[\"Indicators\",\"Indicators_before\",\"Indicators_after\",\"H_Indicator_before\",\"H_Indicator_after\",\n",
    "         \"Geographic Coverage\",\"Geographic_Coverage_before\",\"Geographic_Coverage_after\",\"H_Geographic_Coverage_before\",\"H_Geographic_Coverage_after\",\n",
    "         \"Format\",\"Format_before\",\"Format_after\",\"H_Format_before\",\"H_Format_after\",\n",
    "         \"Spatial Resolution\",\"Spatial Resolution (Standard)\",\"Spatial_Resolution_before\",\"Spatial_Resolution_after\",\"H_Spatial_Resolution_before\",\"H_Spatial_Resolution_after\",\n",
    "         \"Temporal Resolution\",\"Temporal_Resolution_before\",\"Temporal_Resolution_after\",\"H_Temporal_Resolution_before\",\"H_Temporal_Resolution_after\",\n",
    "         \"Temporal Extent\",\"Temporal_Extent_before\",\"Temporal_Extent_after\",\"H_Temporal_Extent_before\",\"H_Temporal_Extent_after\"]\n",
    "human_evaluation=pd.DataFrame(columns=columns)\n",
    "for each_url in urls:\n",
    "    tru={\"urls\":each_url}\n",
    "    true_details=true_data[each_url]\n",
    "    tru.update(true_details)\n",
    "    before_details=before_finetuned[each_url]\n",
    "    before_details={i+str(\"_before\"):v for i,v in before_details.items()}\n",
    "    tru.update(before_details)\n",
    "    after_details=after_finetuned[each_url]\n",
    "    after_details={i+str(\"_after\"):v for i,v in after_details.items()}\n",
    "    tru.update(after_details)\n",
    "    new_data = pd.DataFrame([tru])\n",
    "    human_evaluation = pd.concat([human_evaluation, new_data], ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e42dce7-5db4-451e-8ae2-2e4679924ffa",
   "metadata": {},
   "source": [
    "# Lets look at the dataset before dumping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "abce54fe-c6c2-4447-a30a-43674cadada7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Indicators</th>\n",
       "      <th>Indicators_before</th>\n",
       "      <th>Indicators_after</th>\n",
       "      <th>H_Indicator_before</th>\n",
       "      <th>H_Indicator_after</th>\n",
       "      <th>Geographic Coverage</th>\n",
       "      <th>Geographic_Coverage_before</th>\n",
       "      <th>Geographic_Coverage_after</th>\n",
       "      <th>H_Geographic_Coverage_before</th>\n",
       "      <th>H_Geographic_Coverage_after</th>\n",
       "      <th>...</th>\n",
       "      <th>Temporal_Resolution_before</th>\n",
       "      <th>Temporal_Resolution_after</th>\n",
       "      <th>H_Temporal_Resolution_before</th>\n",
       "      <th>H_Temporal_Resolution_after</th>\n",
       "      <th>Temporal Extent</th>\n",
       "      <th>Temporal_Extent_before</th>\n",
       "      <th>Temporal_Extent_after</th>\n",
       "      <th>H_Temporal_Extent_before</th>\n",
       "      <th>H_Temporal_Extent_after</th>\n",
       "      <th>urls</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Extreme Heat</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Global</td>\n",
       "      <td>global</td>\n",
       "      <td>Idaho, United States</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>varies</td>\n",
       "      <td>Varies - Multiple Datasets Included</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2018-07-09 to Present</td>\n",
       "      <td>present</td>\n",
       "      <td>2018-09-02 to Present</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.5067/ECOSTRESS/ECO2LSTE.001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Food Availability,Human Dimensions</td>\n",
       "      <td>Food Availability</td>\n",
       "      <td>Food Availability</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Central America, Caribbean Islands, Central As...</td>\n",
       "      <td>global</td>\n",
       "      <td>Global</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>yearly</td>\n",
       "      <td>Annual</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2009-01-01 to 2019-12-31</td>\n",
       "      <td>2009-2019</td>\n",
       "      <td>2009-01-01 to 2019-12-31</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://doi.org/10.7927/cx02-2587</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Disasters</td>\n",
       "      <td>Human Dimensions</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Northwest Territories and Northern Alberta, Ca...</td>\n",
       "      <td>Northwest Territories, Canada</td>\n",
       "      <td>Northwest Territories, Canada</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>N/A</td>\n",
       "      <td>Not Applicable</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2014-05-01 to 2015-10-01</td>\n",
       "      <td>varies</td>\n",
       "      <td>2014-06-09 to 2014-09-05</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.3334/ORNLDAAC/1694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Climate Change,Water Availability</td>\n",
       "      <td>Water Availability</td>\n",
       "      <td>Climate Change</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Global</td>\n",
       "      <td>global</td>\n",
       "      <td>Global</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>monthly</td>\n",
       "      <td>Monthly - &lt; Annual</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2002-04-04 to Present</td>\n",
       "      <td>present</td>\n",
       "      <td>2002-04-01 to 2017-06-30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://dx.doi.org/10.5067/TEMSC-3JC63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>Health &amp; Air Quality</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>United States</td>\n",
       "      <td>contiguous United States</td>\n",
       "      <td>United States</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>daily</td>\n",
       "      <td>Daily - &lt; Weekly</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2000-01-01 to 2016-12-31</td>\n",
       "      <td>2000-01-01 to 2016-12-31</td>\n",
       "      <td>2000-01-01 to 2016-12-31</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>https://doi.org/10.7927/0rvr-4538</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                           Indicators     Indicators_before  \\\n",
       "0                        Extreme Heat        Climate Change   \n",
       "1  Food Availability,Human Dimensions     Food Availability   \n",
       "2                           Disasters      Human Dimensions   \n",
       "3   Climate Change,Water Availability    Water Availability   \n",
       "4                Health & Air Quality  Health & Air Quality   \n",
       "\n",
       "       Indicators_after H_Indicator_before H_Indicator_after  \\\n",
       "0        Climate Change                NaN               NaN   \n",
       "1     Food Availability                NaN               NaN   \n",
       "2        Climate Change                NaN               NaN   \n",
       "3        Climate Change                NaN               NaN   \n",
       "4  Health & Air Quality                NaN               NaN   \n",
       "\n",
       "                                 Geographic Coverage  \\\n",
       "0                                             Global   \n",
       "1  Central America, Caribbean Islands, Central As...   \n",
       "2  Northwest Territories and Northern Alberta, Ca...   \n",
       "3                                             Global   \n",
       "4                                      United States   \n",
       "\n",
       "      Geographic_Coverage_before      Geographic_Coverage_after  \\\n",
       "0                         global           Idaho, United States   \n",
       "1                         global                         Global   \n",
       "2  Northwest Territories, Canada  Northwest Territories, Canada   \n",
       "3                         global                         Global   \n",
       "4       contiguous United States                  United States   \n",
       "\n",
       "  H_Geographic_Coverage_before H_Geographic_Coverage_after  ...  \\\n",
       "0                          NaN                         NaN  ...   \n",
       "1                          NaN                         NaN  ...   \n",
       "2                          NaN                         NaN  ...   \n",
       "3                          NaN                         NaN  ...   \n",
       "4                          NaN                         NaN  ...   \n",
       "\n",
       "  Temporal_Resolution_before            Temporal_Resolution_after  \\\n",
       "0                     varies  Varies - Multiple Datasets Included   \n",
       "1                     yearly                               Annual   \n",
       "2                        N/A                       Not Applicable   \n",
       "3                    monthly                   Monthly - < Annual   \n",
       "4                      daily                     Daily - < Weekly   \n",
       "\n",
       "  H_Temporal_Resolution_before H_Temporal_Resolution_after  \\\n",
       "0                          NaN                         NaN   \n",
       "1                          NaN                         NaN   \n",
       "2                          NaN                         NaN   \n",
       "3                          NaN                         NaN   \n",
       "4                          NaN                         NaN   \n",
       "\n",
       "            Temporal Extent    Temporal_Extent_before  \\\n",
       "0     2018-07-09 to Present                   present   \n",
       "1  2009-01-01 to 2019-12-31                 2009-2019   \n",
       "2  2014-05-01 to 2015-10-01                    varies   \n",
       "3     2002-04-04 to Present                   present   \n",
       "4  2000-01-01 to 2016-12-31  2000-01-01 to 2016-12-31   \n",
       "\n",
       "      Temporal_Extent_after H_Temporal_Extent_before H_Temporal_Extent_after  \\\n",
       "0     2018-09-02 to Present                      NaN                     NaN   \n",
       "1  2009-01-01 to 2019-12-31                      NaN                     NaN   \n",
       "2  2014-06-09 to 2014-09-05                      NaN                     NaN   \n",
       "3  2002-04-01 to 2017-06-30                      NaN                     NaN   \n",
       "4  2000-01-01 to 2016-12-31                      NaN                     NaN   \n",
       "\n",
       "                                                urls  \n",
       "0  https://dx.doi.org/10.5067/ECOSTRESS/ECO2LSTE.001  \n",
       "1                  https://doi.org/10.7927/cx02-2587  \n",
       "2           https://dx.doi.org/10.3334/ORNLDAAC/1694  \n",
       "3             https://dx.doi.org/10.5067/TEMSC-3JC63  \n",
       "4                  https://doi.org/10.7927/0rvr-4538  \n",
       "\n",
       "[5 rows x 32 columns]"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "human_evaluation.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0580a161-680b-456f-b69c-b54f43ec2416",
   "metadata": {},
   "source": [
    "# Dumping the excel sheet and human manually evaluates it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "42c0b3b8-0483-4d3d-9925-12b6353c76a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "human_evaluation.to_csv(\"human_evaluation_demo.csv\",index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5ceab52-36e9-42ec-9657-7963b3a5853f",
   "metadata": {},
   "source": [
    "# Loading updated human evaluation : \n",
    "**Note:** H_Indicator_before means Human evaluation of the field Indicator for before finetuing, similarly, H_Indicator_after means human evaluation of the field Indicator for after finetuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "8f0ae641-ebad-48c3-9ada-df83db7fcc94",
   "metadata": {},
   "outputs": [],
   "source": [
    "after_human_evaluation=pd.read_csv(\"../data/human_evaluation_done.csv\",encoding='latin1')\n",
    "before_columns=['H_Indicator_before',\"H_Geographic_Coverage_before\",'H_Format_before','H_Spatial_Resolution_before',\n",
    "         'H_Temporal_Resolution_before','H_Temporal_Extent_before']\n",
    "after_columns=['H_Indicator_after',\"H_Geographic_Coverage_after\",'H_Format_after','H_Spatial_Resolution_after','H_Temporal_Resolution_after',\n",
    "              'H_Temporal_Extent_after']\n",
    "human_eval_before=after_human_evaluation[before_columns]\n",
    "human_eval_after=after_human_evaluation[after_columns]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "10eee465-ae2b-4a89-9c47-97e00c5eb33b",
   "metadata": {},
   "source": [
    "# Loading the human evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "f42097f3-24b4-47cf-bf81-87a6067fcdad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>H_Indicator_before</th>\n",
       "      <th>H_Geographic_Coverage_before</th>\n",
       "      <th>H_Format_before</th>\n",
       "      <th>H_Spatial_Resolution_before</th>\n",
       "      <th>H_Temporal_Resolution_before</th>\n",
       "      <th>H_Temporal_Extent_before</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   H_Indicator_before  H_Geographic_Coverage_before  H_Format_before  \\\n",
       "0                   0                             1                0   \n",
       "1                   0                             1                0   \n",
       "2                   1                             1                1   \n",
       "3                   0                             1                0   \n",
       "4                   0                             1                0   \n",
       "\n",
       "   H_Spatial_Resolution_before  H_Temporal_Resolution_before  \\\n",
       "0                            1                             0   \n",
       "1                            0                             1   \n",
       "2                            0                             1   \n",
       "3                            0                             0   \n",
       "4                            1                             1   \n",
       "\n",
       "   H_Temporal_Extent_before  \n",
       "0                         0  \n",
       "1                         1  \n",
       "2                         0  \n",
       "3                         0  \n",
       "4                         0  "
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "human_eval_before.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "beefa272-5334-4df1-804a-d551447870e1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>H_Indicator_after</th>\n",
       "      <th>H_Geographic_Coverage_after</th>\n",
       "      <th>H_Format_after</th>\n",
       "      <th>H_Spatial_Resolution_after</th>\n",
       "      <th>H_Temporal_Resolution_after</th>\n",
       "      <th>H_Temporal_Extent_after</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   H_Indicator_after  H_Geographic_Coverage_after  H_Format_after  \\\n",
       "0                  0                            1               1   \n",
       "1                  1                            1               1   \n",
       "2                  1                            1               1   \n",
       "3                  0                            1               0   \n",
       "4                  0                            1               1   \n",
       "\n",
       "   H_Spatial_Resolution_after  H_Temporal_Resolution_after  \\\n",
       "0                           1                            1   \n",
       "1                           0                            1   \n",
       "2                           1                            1   \n",
       "3                           1                            0   \n",
       "4                           1                            1   \n",
       "\n",
       "   H_Temporal_Extent_after  \n",
       "0                        1  \n",
       "1                        1  \n",
       "2                        1  \n",
       "3                        0  \n",
       "4                        1  "
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "human_eval_after.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "43b82470-6cdf-49bf-bd45-bf712a55c3e5",
   "metadata": {},
   "source": [
    "# Preprocessing the human evaluation result in a desirable format for visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "e74bcab4-fefa-4dcb-b195-c274a79e14cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "before=human_eval_before.to_dict(orient='list')\n",
    "before={i.replace(\"H_\",\"\").replace(\"_before\",\"\"):j for i,j in before.items()}\n",
    "\n",
    "after=human_eval_after.to_dict(orient='list')\n",
    "after={i.replace(\"H_\",\"\").replace(\"_after\",\"\"):j for i,j in after.items()}\n",
    "\n",
    "human_before_metrics=get_metrics(before)\n",
    "human_after_metrics=get_metrics(after)\n",
    "\n",
    "keys = list(human_before_metrics.keys())\n",
    "metrics_before_values = list(human_before_metrics.values())\n",
    "metrics_after_values = list(human_after_metrics.values())\n",
    "\n",
    "# Plotting\n",
    "plt.figure(figsize=(10, 6))\n",
    "\n",
    "# Plotting bar graphs\n",
    "bar_width = 0.35\n",
    "x = range(len(keys))  # X-axis positions for bars\n",
    "plt.bar(x, metrics_before_values, width=bar_width, label='Human Evaluation of Features Before Finetuning')\n",
    "plt.bar([pos + bar_width for pos in x], metrics_after_values, width=bar_width, label='Human Evaluation of Features After Finetuning')\n",
    "\n",
    "plt.title('Comparison of Human Evaluated Metrics Before and After Finetuning')\n",
    "plt.xlabel('Metrics')\n",
    "plt.ylabel('Scores')\n",
    "plt.xticks([pos + bar_width/2 for pos in x], keys, rotation=10)  # Setting keys as x-axis labels\n",
    "\n",
    "plt.legend()\n",
    "plt.grid(False)\n",
    "plt.tight_layout()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}