azure-ai-search-quantization.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Azure AI Search: Scalar Quantization and Stored Property \n",
    "\n",
    "In addition to launching [new storage limits](https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/announcing-cost-effective-rag-at-scale-with-azure-ai-search/ba-p/4104961), Azure AI Search introduces new features for optimizing price-performance: **Scalar Quantization**, **Binary Quantization** and a new **\"stored\" property** for vector fields.\n",
    "\n",
    "## Features\n",
    "\n",
    "### Scalar Quantization\n",
    "\n",
    "Scalar Quantization is a compression technique for vector data. It reduces storage costs and enhances search performance without significantly compromising result accuracy. This method transforms high-dimensional vectors into a lower-bit representation, resulting in substantial storage savings.\n",
    "\n",
    "### Binary Quantization\n",
    "\n",
    "Binary Quantization is another compression technique that converts vectors into binary codes. This method further reduces storage requirements and can improve search speed, especially in large datasets.\n",
    "\n",
    "### All-Of-The-Above\n",
    "\n",
    "All refers to the combined application of both Stored and Binary Quantization techniques. This approach maximizes storage efficiency and search performance by leveraging the strengths of both methods.\n",
    "\n",
    "### Stored Property\n",
    "\n",
    "The \"stored\" property provides flexibility in managing storage costs by selectively storing document properties in the index. When set to False, the property isn't stored, which reduces the overall index size and cost, while maintaining searchability.\n",
    "\n",
    "## Implementation\n",
    "\n",
    "The implementation process involves several steps:\n",
    "\n",
    "1. **Installation:** Install all necessary libraries to ensure the workspace has the latest versions of Azure Search Documents, OpenAI, and other dependencies.\n",
    "\n",
    "2. **Index Creation:** Create or update indexes using the Azure Search Documents Index Client. The scenarios specified include:\n",
    "   - `baseline`: The original index without optimizations.\n",
    "   - `stored`: The Stored Property is set to False.\n",
    "   - `scalar-quantization`: Applies Scalar Quantization compression.\n",
    "   - `binary-quantization`: Applies Binary Quantization compression.\n",
    "   - `all`: Combines Stored=False and Binary Quantization techniques.\n",
    "\n",
    "3. **Document Transformation and Uploading:** Transform the dataset into a list of dictionaries suitable for Azure AI Search ingestion. Upload these transformed documents in batches to the respective indexes created in the previous step."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Install required libraries\n",
    "This code block is used to install and upgrade the necessary Python libraries for the project:\n",
    "\n",
    "- `azure-search-documents`: This is the Azure Search Documents library, which provides the necessary tools to interact with the Azure Search service.\n",
    "\n",
    "- `openai`: This library is used to interact with the OpenAI API, which provides various AI models.\n",
    "\n",
    "- `python-dotenv`: This library is used to read key-value pairs from a .env file and set them as environment variables.\n",
    "\n",
    "- `azure-identity`: This library provides Azure Active Directory token authentication support across the Azure SDK.\n",
    "\n",
    "- `datasets`: This library provides a collection of datasets with easy-to-use APIs for downloading and using them for model training and evaluation.\n",
    "\n",
    "- `matplotlib`: This is a plotting library for Python and its numerical mathematics extension, NumPy. It provides an object-oriented API for embedding plots into applications.\n",
    "\n",
    "- `prettytable`: This library is used to easily display tabular data in a visually appealing ASCII table format.\n",
    "\n",
    "The `%pip install` command is a magic command in Jupyter notebooks, used to run pip commands directly from the notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting azure-search-documents\n",
      "  Using cached azure_search_documents-11.5.1-py3-none-any.whl.metadata (23 kB)\n",
      "Requirement already satisfied: azure-core>=1.28.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-search-documents) (1.30.2)\n",
      "Requirement already satisfied: azure-common>=1.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-search-documents) (1.1.28)\n",
      "Requirement already satisfied: isodate>=0.6.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-search-documents) (0.6.1)\n",
      "Requirement already satisfied: typing-extensions>=4.6.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-search-documents) (4.12.2)\n",
      "Requirement already satisfied: requests>=2.21.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-core>=1.28.0->azure-search-documents) (2.32.3)\n",
      "Requirement already satisfied: six>=1.11.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-core>=1.28.0->azure-search-documents) (1.16.0)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.28.0->azure-search-documents) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.28.0->azure-search-documents) (3.7)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.28.0->azure-search-documents) (2.2.2)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.28.0->azure-search-documents) (2024.7.4)\n",
      "Using cached azure_search_documents-11.5.1-py3-none-any.whl (297 kB)\n",
      "Installing collected packages: azure-search-documents\n",
      "Successfully installed azure-search-documents-11.5.1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: openai in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (1.37.1)\n",
      "Requirement already satisfied: anyio<5,>=3.5.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (4.4.0)\n",
      "Requirement already satisfied: distro<2,>=1.7.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (1.9.0)\n",
      "Requirement already satisfied: httpx<1,>=0.23.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (0.27.0)\n",
      "Requirement already satisfied: pydantic<3,>=1.9.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (2.8.2)\n",
      "Requirement already satisfied: sniffio in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (1.3.1)\n",
      "Requirement already satisfied: tqdm>4 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (4.66.4)\n",
      "Requirement already satisfied: typing-extensions<5,>=4.7 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from openai) (4.12.2)\n",
      "Requirement already satisfied: idna>=2.8 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from anyio<5,>=3.5.0->openai) (3.7)\n",
      "Requirement already satisfied: certifi in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from httpx<1,>=0.23.0->openai) (2024.7.4)\n",
      "Requirement already satisfied: httpcore==1.* in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from httpx<1,>=0.23.0->openai) (1.0.5)\n",
      "Requirement already satisfied: h11<0.15,>=0.13 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from httpcore==1.*->httpx<1,>=0.23.0->openai) (0.14.0)\n",
      "Requirement already satisfied: annotated-types>=0.4.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from pydantic<3,>=1.9.0->openai) (0.7.0)\n",
      "Requirement already satisfied: pydantic-core==2.20.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from pydantic<3,>=1.9.0->openai) (2.20.1)\n",
      "Requirement already satisfied: colorama in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from tqdm>4->openai) (0.4.6)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: python-dotenv in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (1.0.1)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: azure-identity in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (1.17.1)\n",
      "Requirement already satisfied: azure-core>=1.23.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-identity) (1.30.2)\n",
      "Requirement already satisfied: cryptography>=2.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-identity) (43.0.0)\n",
      "Requirement already satisfied: msal>=1.24.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-identity) (1.30.0)\n",
      "Requirement already satisfied: msal-extensions>=0.3.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-identity) (1.2.0)\n",
      "Requirement already satisfied: typing-extensions>=4.0.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-identity) (4.12.2)\n",
      "Requirement already satisfied: requests>=2.21.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-core>=1.23.0->azure-identity) (2.32.3)\n",
      "Requirement already satisfied: six>=1.11.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from azure-core>=1.23.0->azure-identity) (1.16.0)\n",
      "Requirement already satisfied: cffi>=1.12 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from cryptography>=2.5->azure-identity) (1.16.0)\n",
      "Requirement already satisfied: PyJWT<3,>=1.0.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from PyJWT[crypto]<3,>=1.0.0->msal>=1.24.0->azure-identity) (2.8.0)\n",
      "Requirement already satisfied: portalocker<3,>=1.4 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from msal-extensions>=0.3.0->azure-identity) (2.10.1)\n",
      "Requirement already satisfied: pycparser in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from cffi>=1.12->cryptography>=2.5->azure-identity) (2.22)\n",
      "Requirement already satisfied: pywin32>=226 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from portalocker<3,>=1.4->msal-extensions>=0.3.0->azure-identity) (306)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.23.0->azure-identity) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.23.0->azure-identity) (3.7)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.23.0->azure-identity) (2.2.2)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.21.0->azure-core>=1.23.0->azure-identity) (2024.7.4)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n",
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: datasets in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (2.20.0)\n",
      "Requirement already satisfied: filelock in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (3.15.4)\n",
      "Requirement already satisfied: numpy>=1.17 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (1.26.4)\n",
      "Requirement already satisfied: pyarrow>=15.0.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (17.0.0)\n",
      "Requirement already satisfied: pyarrow-hotfix in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (0.6)\n",
      "Requirement already satisfied: dill<0.3.9,>=0.3.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (0.3.8)\n",
      "Requirement already satisfied: pandas in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (2.2.2)\n",
      "Requirement already satisfied: requests>=2.32.2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (2.32.3)\n",
      "Requirement already satisfied: tqdm>=4.66.3 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (4.66.4)\n",
      "Requirement already satisfied: xxhash in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (3.4.1)\n",
      "Requirement already satisfied: multiprocess in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (0.70.16)\n",
      "Requirement already satisfied: fsspec<=2024.5.0,>=2023.1.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from fsspec[http]<=2024.5.0,>=2023.1.0->datasets) (2024.5.0)\n",
      "Requirement already satisfied: aiohttp in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (3.10.0)\n",
      "Requirement already satisfied: huggingface-hub>=0.21.2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (0.24.5)\n",
      "Requirement already satisfied: packaging in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (24.1)\n",
      "Requirement already satisfied: pyyaml>=5.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from datasets) (6.0.1)\n",
      "Requirement already satisfied: aiohappyeyeballs>=2.3.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (2.3.2)\n",
      "Requirement already satisfied: aiosignal>=1.1.2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (1.3.1)\n",
      "Requirement already satisfied: attrs>=17.3.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (23.2.0)\n",
      "Requirement already satisfied: frozenlist>=1.1.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (1.4.1)\n",
      "Requirement already satisfied: multidict<7.0,>=4.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (6.0.5)\n",
      "Requirement already satisfied: yarl<2.0,>=1.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from aiohttp->datasets) (1.9.4)\n",
      "Requirement already satisfied: typing-extensions>=3.7.4.3 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from huggingface-hub>=0.21.2->datasets) (4.12.2)\n",
      "Requirement already satisfied: charset-normalizer<4,>=2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.32.2->datasets) (3.3.2)\n",
      "Requirement already satisfied: idna<4,>=2.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.32.2->datasets) (3.7)\n",
      "Requirement already satisfied: urllib3<3,>=1.21.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.32.2->datasets) (2.2.2)\n",
      "Requirement already satisfied: certifi>=2017.4.17 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from requests>=2.32.2->datasets) (2024.7.4)\n",
      "Requirement already satisfied: colorama in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from tqdm>=4.66.3->datasets) (0.4.6)\n",
      "Requirement already satisfied: python-dateutil>=2.8.2 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from pandas->datasets) (2.9.0.post0)\n",
      "Requirement already satisfied: pytz>=2020.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from pandas->datasets) (2024.1)\n",
      "Requirement already satisfied: tzdata>=2022.7 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from pandas->datasets) (2024.1)\n",
      "Requirement already satisfied: six>=1.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from python-dateutil>=2.8.2->pandas->datasets) (1.16.0)\n",
      "Requirement already satisfied: matplotlib in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (3.9.1)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (1.2.1)\n",
      "Requirement already satisfied: cycler>=0.10 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (0.12.1)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (4.53.1)\n",
      "Requirement already satisfied: kiwisolver>=1.3.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (1.4.5)\n",
      "Requirement already satisfied: numpy>=1.23 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (1.26.4)\n",
      "Requirement already satisfied: packaging>=20.0 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (24.1)\n",
      "Requirement already satisfied: pillow>=8 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (10.4.0)\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (3.1.2)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from matplotlib) (2.9.0.post0)\n",
      "Requirement already satisfied: six>=1.5 in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from python-dateutil>=2.7->matplotlib) (1.16.0)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: prettytable in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (3.10.2)\n",
      "Requirement already satisfied: wcwidth in c:\\dev\\azure-ai-search-python-playground\\.venv\\lib\\site-packages (from prettytable) (0.2.13)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "[notice] A new release of pip is available: 24.0 -> 24.2\n",
      "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
     ]
    }
   ],
   "source": [
    "!pip install --upgrade azure-search-documents \n",
    "!pip install openai \n",
    "!pip install python-dotenv \n",
    "!pip install azure-identity\n",
    "!pip install datasets\n",
    "!pip install matplotlib\n",
    "!pip install prettytable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install --upgrade azure-search-documents ## I'm using 11.5.1\n",
    "%pip install openai \n",
    "%pip install python-dotenv \n",
    "%pip install azure-identity\n",
    "%pip install datasets\n",
    "%pip install matplotlib\n",
    "%pip install prettytable"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Initialize necessary imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import numpy as np\n",
    "import os\n",
    "import requests\n",
    "from azure.core.credentials import AzureKeyCredential\n",
    "from azure.identity import DefaultAzureCredential\n",
    "from azure.search.documents import SearchClient\n",
    "from azure.search.documents.indexes import SearchIndexClient\n",
    "from azure.search.documents.indexes.models import (\n",
    "    AzureOpenAIModelName,\n",
    "    AzureOpenAIVectorizer,\n",
    "    AzureOpenAIVectorizerParameters,\n",
    "    BinaryQuantizationCompression,\n",
    "    HnswAlgorithmConfiguration,\n",
    "    HnswParameters,\n",
    "    ScalarQuantizationCompression,\n",
    "    ScalarQuantizationParameters,\n",
    "    SearchField,\n",
    "    SearchFieldDataType,\n",
    "    SearchIndex,\n",
    "    SearchableField,\n",
    "    SimpleField,\n",
    "    VectorSearch,\n",
    "    VectorSearchAlgorithmKind,\n",
    "    VectorSearchProfile,\n",
    ")\n",
    "from azure.search.documents.models import VectorizableTextQuery, VectorizedQuery\n",
    "from dotenv import load_dotenv"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Environment Setup and Client Initialization\n",
    "\n",
    "This Python code block performs the following tasks:\n",
    "\n",
    "1. **Environment Variables:** Loads environment variables from a `.env` file using `load_dotenv()`.\n",
    "\n",
    "2. **OpenAI Client Setup:** Sets up the OpenAI client using environment variables for the endpoint, API key, and the name of the deployed model.\n",
    "\n",
    "3. **Search Service Client Setup:** Sets up the Search Service client using environment variables for the endpoint and API key.\n",
    "\n",
    "4. **Constants:** Defines constants such as the OpenAI API version and the index name.\n",
    "\n",
    "5. **Authentication:** Chooses the appropriate credential based on the authentication method (Azure Active Directory or API key).\n",
    "\n",
    "6. **Client Initialization:** Initializes the `SearchIndexClient` for creating indexes and the `SearchClient` for searching the index."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load environment variables from .env file\n",
    "load_dotenv()\n",
    "\n",
    "# Set up OpenAI client based on environment variables\n",
    "AZURE_OPENAI_ENDPOINT: str = os.getenv(\"AZURE_OPENAI_ENDPOINT\")\n",
    "AZURE_OPENAI_API_KEY: str = os.getenv(\"AZURE_OPENAI_API_KEY\")\n",
    "AZURE_OPENAI_EMBEDDING_DEPLOYED_MODEL_NAME: str = os.getenv(\"AZURE_OPENAI_EMBEDDING_DEPLOYED_MODEL_NAME\")\n",
    "AZURE_OPENAI_API_VERSION: str = \"2023-05-15\"\n",
    "\n",
    "# Set up Search Service client based on environment variables\n",
    "SEARCH_SERVICE_ENDPOINT = os.getenv(\"AZURE_SEARCH_SERVICE_ENDPOINT\")\n",
    "SEARCH_SERVICE_API_KEY = os.getenv(\"AZURE_SEARCH_ADMIN_KEY\")\n",
    "\n",
    "INDEX_NAME = \"dbpedia-1m\"\n",
    "\n",
    "# Authentication method flag\n",
    "use_aad_for_search = False  # Set based on your authentication method\n",
    "\n",
    "# Choose the correct credential based on your authentication method\n",
    "credential = (\n",
    "    DefaultAzureCredential()\n",
    "    if use_aad_for_search\n",
    "    else AzureKeyCredential(SEARCH_SERVICE_API_KEY)\n",
    ")\n",
    "\n",
    "# Initialize the SearchIndexClient for creating indexes\n",
    "index_client = SearchIndexClient(\n",
    "    endpoint=SEARCH_SERVICE_ENDPOINT, credential=credential\n",
    ")\n",
    "\n",
    "# Initialize the SearchClient\n",
    "search_client = SearchClient(endpoint=SEARCH_SERVICE_ENDPOINT, credential=credential, index_name=INDEX_NAME)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Processing and Conversion\n",
    "\n",
    "This Python code block loads a dataset, processes it by generating unique IDs and extracting relevant fields, and then converts it into a list of dictionaries for indexing. It uses multiprocessing and caching for efficient data processing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import uuid\n",
    "from datasets import load_dataset\n",
    "\n",
    "# Constants\n",
    "DATASET_NAME = \"KShivendu/dbpedia-entities-openai-1M\"\n",
    "BATCH_SIZE = 2000\n",
    "NUM_PROC = 32\n",
    "\n",
    "def process_batch(batch_data):\n",
    "    \"\"\"Processes a batch of data by generating unique IDs and extracting relevant fields.\"\"\"\n",
    "    return {\n",
    "        \"id\": [str(uuid.uuid4()) for _ in range(len(batch_data[\"title\"]))],  # Generate unique IDs using uuid4\n",
    "        \"title\": batch_data[\"title\"],\n",
    "        \"text\": batch_data[\"text\"],\n",
    "        \"embedding\": batch_data[\"openai\"],  # Ensure this field name matches your dataset schema\n",
    "    }\n",
    "\n",
    "# Load the entire dataset\n",
    "dataset = load_dataset(DATASET_NAME, split=\"train\")\n",
    "\n",
    "# Convert the dataset to a list of dictionaries for indexing\n",
    "# Enable caching to avoid reprocessing if the script is run multiple times\n",
    "documents_to_index = dataset.map(\n",
    "    process_batch,\n",
    "    batched=True,\n",
    "    batch_size=BATCH_SIZE,  # Adjust batch size based on your system's memory capacity and dataset size\n",
    "    num_proc=NUM_PROC,  # Adjust based on the number of available CPU cores\n",
    "    load_from_cache_file=True,\n",
    "    remove_columns=dataset.column_names,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Index Creation and Update\n",
    "\n",
    "This Python code block defines functions to create or update an Azure AI Search index with different scenarios. It sets up fields, vector search profiles, algorithms, compressions, and vectorizers based on the scenario, and then creates or updates the index accordingly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index for baseline created or updated successfully.\n",
      "Index for stored created or updated successfully.\n",
      "Index for scalar-quantization created or updated successfully.\n",
      "Index for binary-quantization created or updated successfully.\n",
      "Index for all created or updated successfully.\n"
     ]
    }
   ],
   "source": [
    "DIMENSIONS = 1536\n",
    "HNSW_PARAMETERS = {\"m\": 4, \"metric\": \"cosine\", \"ef_construction\": 400, \"ef_search\": 500}\n",
    "DEFAULT_OVERSAMPLING = 10\n",
    "QUANTIZED_DATA_TYPE = \"int8\"\n",
    "\n",
    "\n",
    "def create_fields(scenario):\n",
    "    profile_name = f\"profile-{scenario}\"\n",
    "    stored = True\n",
    "\n",
    "    if scenario == \"all\":\n",
    "        profile_name = \"profile-all\"\n",
    "        stored = False\n",
    "    elif scenario == \"binary-quantization\":\n",
    "        profile_name = \"profile-binary-quantization\"\n",
    "    elif scenario == \"scalar-quantization\":\n",
    "        profile_name = \"profile-scalar-quantization\"\n",
    "    elif scenario == \"stored\":\n",
    "        profile_name = \"profile-stored\"\n",
    "        stored = False\n",
    "\n",
    "    return [\n",
    "        SimpleField(name=\"id\", type=SearchFieldDataType.String, key=True),\n",
    "        SearchField(name=\"title\", type=SearchFieldDataType.String, searchable=True),\n",
    "        SearchField(name=\"text\", type=SearchFieldDataType.String, searchable=True),\n",
    "        SearchField(\n",
    "            name=\"embedding\",\n",
    "            type=\"Collection(Edm.Single)\",\n",
    "            vector_search_dimensions=DIMENSIONS,\n",
    "            vector_search_profile_name=profile_name,\n",
    "            stored=stored,\n",
    "        ),\n",
    "    ]\n",
    "\n",
    "\n",
    "def create_vector_search_profiles(scenario):\n",
    "    compression_name = None\n",
    "    if scenario == \"all\":\n",
    "        compression_name = \"compression-binary-quantization\"\n",
    "    elif scenario == \"binary-quantization\":\n",
    "        compression_name = \"compression-binary-quantization\"\n",
    "    elif scenario == \"scalar-quantization\":\n",
    "        compression_name = \"compression-scalar-quantization\"\n",
    "\n",
    "    if scenario == \"stored\":\n",
    "        return [\n",
    "            VectorSearchProfile(\n",
    "                name=\"profile-stored\",\n",
    "                algorithm_configuration_name=\"hnsw-stored\",\n",
    "                compression_name=None,\n",
    "                vectorizer_name=None,\n",
    "            )\n",
    "        ]\n",
    "\n",
    "    return [\n",
    "        VectorSearchProfile(\n",
    "            name=f\"profile-{scenario}\",\n",
    "            algorithm_configuration_name=f\"hnsw-{scenario}\",\n",
    "            compression_name=compression_name,\n",
    "            vectorizer_name=\"azure_openai_vectorizer\",\n",
    "        )\n",
    "    ]\n",
    "\n",
    "\n",
    "def create_algorithms(scenario):\n",
    "    return [\n",
    "        HnswAlgorithmConfiguration(\n",
    "            name=f\"hnsw-{scenario}\",\n",
    "            kind=VectorSearchAlgorithmKind.HNSW,\n",
    "            parameters=HnswParameters(**HNSW_PARAMETERS),\n",
    "        )\n",
    "    ]\n",
    "\n",
    "\n",
    "def create_compressions(scenario):\n",
    "    compressions = []\n",
    "    if scenario == \"scalar-quantization\":\n",
    "        compressions.append(\n",
    "            ScalarQuantizationCompression(\n",
    "                compression_name=\"compression-scalar-quantization\",\n",
    "                rerank_with_original_vectors=True,\n",
    "                default_oversampling=DEFAULT_OVERSAMPLING,\n",
    "                parameters=ScalarQuantizationParameters(\n",
    "                    quantized_data_type=QUANTIZED_DATA_TYPE\n",
    "                ),\n",
    "            )\n",
    "        )\n",
    "    elif scenario == \"binary-quantization\":\n",
    "        compressions.append(\n",
    "            BinaryQuantizationCompression(\n",
    "                compression_name=\"compression-binary-quantization\",\n",
    "                rerank_with_original_vectors=True,\n",
    "                default_oversampling=DEFAULT_OVERSAMPLING,\n",
    "            )\n",
    "        )\n",
    "    elif scenario == \"all\":\n",
    "        compressions.append(\n",
    "            ScalarQuantizationCompression(\n",
    "                compression_name=\"compression-scalar-quantization\",\n",
    "                rerank_with_original_vectors=True,\n",
    "                default_oversampling=DEFAULT_OVERSAMPLING,\n",
    "                parameters=ScalarQuantizationParameters(\n",
    "                    quantized_data_type=QUANTIZED_DATA_TYPE\n",
    "                ),\n",
    "            )\n",
    "        )\n",
    "        compressions.append(\n",
    "            BinaryQuantizationCompression(\n",
    "                compression_name=\"compression-binary-quantization\",\n",
    "                rerank_with_original_vectors=True,\n",
    "                default_oversampling=DEFAULT_OVERSAMPLING,\n",
    "            )\n",
    "        )\n",
    "    return compressions\n",
    "\n",
    "\n",
    "def create_vectorizers():\n",
    "    return [\n",
    "        AzureOpenAIVectorizer(\n",
    "            vectorizer_name=\"azure_openai_vectorizer\",\n",
    "            parameters=AzureOpenAIVectorizerParameters(\n",
    "                resource_url=AZURE_OPENAI_ENDPOINT,\n",
    "                deployment_name=AZURE_OPENAI_EMBEDDING_DEPLOYED_MODEL_NAME,\n",
    "                api_key=AZURE_OPENAI_API_KEY,\n",
    "                model_name=AZURE_OPENAI_EMBEDDING_DEPLOYED_MODEL_NAME,\n",
    "            ),\n",
    "        ),\n",
    "    ]\n",
    "\n",
    "\n",
    "def create_or_update_index(client, index_name, scenario):\n",
    "    fields = create_fields(scenario)\n",
    "    vector_search_profiles = create_vector_search_profiles(scenario)\n",
    "    algorithms = create_algorithms(scenario)\n",
    "    compressions = create_compressions(scenario)\n",
    "    vectorizers = create_vectorizers()\n",
    "\n",
    "    vector_search = VectorSearch(\n",
    "        profiles=vector_search_profiles,\n",
    "        algorithms=algorithms,\n",
    "        compressions=compressions,\n",
    "        vectorizers=vectorizers,\n",
    "    )\n",
    "\n",
    "    index = SearchIndex(\n",
    "        name=f\"{index_name}-{scenario}\", fields=fields, vector_search=vector_search\n",
    "    )\n",
    "    client.create_or_update_index(index)\n",
    "    print(f\"Index for {scenario} created or updated successfully.\")\n",
    "\n",
    "\n",
    "scenarios = [\"baseline\", \"stored\", \"scalar-quantization\", \"binary-quantization\", \"all\"]\n",
    "for scenario in scenarios:\n",
    "    create_or_update_index(index_client, INDEX_NAME, scenario)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Transformation\n",
    "\n",
    "This Python code block defines a function to transform a dataset into the format required by Azure AI Search. It then applies this function to all documents in the dataset and prints the first few transformed documents to verify correctness."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from azure.core.credentials import AzureKeyCredential\n",
    "from azure.search.documents import SearchClient\n",
    "\n",
    "# Function to transform your dataset into the format required by Azure Search\n",
    "def transform_to_search_document(record):\n",
    "    return {\n",
    "        \"id\": record[\"id\"],\n",
    "        \"title\": record[\"title\"],\n",
    "        \"text\": record[\"text\"],\n",
    "        \"embedding\": record[\"embedding\"]\n",
    "    }\n",
    "    \n",
    "# Transform all documents in the dataset\n",
    "transformed_documents = [transform_to_search_document(doc) for doc in documents_to_index]\n",
    "\n",
    "# Print the first few transformed documents to check correctness\n",
    "print(transformed_documents[:10])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Document Upload to Multiple Indexes with Concurrency\n",
    "\n",
    "This Python code block creates a search client and uploads documents to multiple Azure Search indexes using concurrency for efficiency. The process is repeated for each index with a different index name.\n",
    "\n",
    "1. **Search Client Creation:** The `create_search_client` function creates a `SearchClient` for a given index name.\n",
    "\n",
    "2. **Document Upload:** The `upload_documents_to_index` function uploads documents to the index. It defines an inner function `upload_batch` to upload a batch of documents and handle any errors.\n",
    "\n",
    "3. **Concurrency:** The `ThreadPoolExecutor` is used to upload multiple batches of documents concurrently. This is done to speed up the upload process by making use of multiple threads. The number of worker threads is defined by `NUM_WORKERS`. The documents are divided into batches of size `BATCH_SIZE`, and each batch is submitted to the executor for concurrent upload.\n",
    "\n",
    "4. **Execution for Multiple Indexes:** A search client is created for each index (e.g., \"dbpedia-1m-baseline\", \"dbpedia-1m-stored\", etc.), and the transformed documents are uploaded to each index in turn."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "import logging\n",
    "\n",
    "BATCH_SIZE = 1000\n",
    "NUM_WORKERS = 16\n",
    "\n",
    "def create_search_client(index_name):\n",
    "    return SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT, index_name=index_name, credential=credential\n",
    "    )\n",
    "\n",
    "def upload_documents_to_index(client, documents):\n",
    "    def upload_batch(batch):\n",
    "        try:\n",
    "            result = client.merge_or_upload_documents(documents=batch)\n",
    "            if result.is_error:\n",
    "                for error in result.errors:\n",
    "                    logging.error(\n",
    "                        f\"Failed to upload document with ID: {error.key}, error: {error.error.message}\"\n",
    "                    )\n",
    "            else:\n",
    "                logging.info(\"Successfully uploaded batch\")\n",
    "        except Exception as e:\n",
    "            logging.error(f\"Failed to upload batch: {e}\")\n",
    "\n",
    "    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:\n",
    "        for i in range(0, len(documents), BATCH_SIZE):\n",
    "            batch = documents[i : i + BATCH_SIZE]\n",
    "            executor.submit(upload_batch, batch)\n",
    "\n",
    "index_name = \"dbpedia-1m-baseline\"\n",
    "client = create_search_client(index_name)\n",
    "upload_documents_to_index(client, transformed_documents)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "import logging\n",
    "\n",
    "BATCH_SIZE = 1000\n",
    "NUM_WORKERS = 16\n",
    "\n",
    "def create_search_client(index_name):\n",
    "    return SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT, index_name=index_name, credential=credential\n",
    "    )\n",
    "\n",
    "def upload_documents_to_index(client, documents):\n",
    "    def upload_batch(batch):\n",
    "        try:\n",
    "            result = client.merge_or_upload_documents(documents=batch)\n",
    "            if result.is_error:\n",
    "                for error in result.errors:\n",
    "                    logging.error(\n",
    "                        f\"Failed to upload document with ID: {error.key}, error: {error.error.message}\"\n",
    "                    )\n",
    "            else:\n",
    "                logging.info(\"Successfully uploaded batch\")\n",
    "        except Exception as e:\n",
    "            logging.error(f\"Failed to upload batch: {e}\")\n",
    "\n",
    "    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:\n",
    "        for i in range(0, len(documents), BATCH_SIZE):\n",
    "            batch = documents[i : i + BATCH_SIZE]\n",
    "            executor.submit(upload_batch, batch)\n",
    "\n",
    "index_name = \"dbpedia-1m-stored\"\n",
    "client = create_search_client(index_name)\n",
    "upload_documents_to_index(client, transformed_documents)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "import logging\n",
    "\n",
    "BATCH_SIZE = 1000\n",
    "NUM_WORKERS = 16\n",
    "\n",
    "def create_search_client(index_name):\n",
    "    return SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT, index_name=index_name, credential=credential\n",
    "    )\n",
    "\n",
    "def upload_documents_to_index(client, documents):\n",
    "    def upload_batch(batch):\n",
    "        try:\n",
    "            result = client.merge_or_upload_documents(documents=batch)\n",
    "            if result.is_error:\n",
    "                for error in result.errors:\n",
    "                    logging.error(\n",
    "                        f\"Failed to upload document with ID: {error.key}, error: {error.error.message}\"\n",
    "                    )\n",
    "            else:\n",
    "                logging.info(\"Successfully uploaded batch\")\n",
    "        except Exception as e:\n",
    "            logging.error(f\"Failed to upload batch: {e}\")\n",
    "\n",
    "    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:\n",
    "        for i in range(0, len(documents), BATCH_SIZE):\n",
    "            batch = documents[i : i + BATCH_SIZE]\n",
    "            executor.submit(upload_batch, batch)\n",
    "\n",
    "index_name = \"dbpedia-1m-scalar-quantization\"\n",
    "client = create_search_client(index_name)\n",
    "upload_documents_to_index(client, transformed_documents)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "import logging\n",
    "\n",
    "BATCH_SIZE = 1000\n",
    "NUM_WORKERS = 16\n",
    "\n",
    "def create_search_client(index_name):\n",
    "    return SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT, index_name=index_name, credential=credential\n",
    "    )\n",
    "\n",
    "def upload_documents_to_index(client, documents):\n",
    "    def upload_batch(batch):\n",
    "        try:\n",
    "            result = client.merge_or_upload_documents(documents=batch)\n",
    "            if result.is_error:\n",
    "                for error in result.errors:\n",
    "                    logging.error(\n",
    "                        f\"Failed to upload document with ID: {error.key}, error: {error.error.message}\"\n",
    "                    )\n",
    "            else:\n",
    "                logging.info(\"Successfully uploaded batch\")\n",
    "        except Exception as e:\n",
    "            logging.error(f\"Failed to upload batch: {e}\")\n",
    "\n",
    "    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:\n",
    "        for i in range(0, len(documents), BATCH_SIZE):\n",
    "            batch = documents[i : i + BATCH_SIZE]\n",
    "            executor.submit(upload_batch, batch)\n",
    "\n",
    "index_name = \"dbpedia-1m-binary-quantization\"\n",
    "client = create_search_client(index_name)\n",
    "upload_documents_to_index(client, transformed_documents)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "import logging\n",
    "\n",
    "BATCH_SIZE = 1000\n",
    "NUM_WORKERS = 16\n",
    "\n",
    "def create_search_client(index_name):\n",
    "    return SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT, index_name=index_name, credential=credential\n",
    "    )\n",
    "\n",
    "def upload_documents_to_index(client, documents):\n",
    "    def upload_batch(batch):\n",
    "        try:\n",
    "            result = client.merge_or_upload_documents(documents=batch)\n",
    "            if result.is_error:\n",
    "                for error in result.errors:\n",
    "                    logging.error(\n",
    "                        f\"Failed to upload document with ID: {error.key}, error: {error.error.message}\"\n",
    "                    )\n",
    "            else:\n",
    "                logging.info(\"Successfully uploaded batch\")\n",
    "        except Exception as e:\n",
    "            logging.error(f\"Failed to upload batch: {e}\")\n",
    "\n",
    "    with ThreadPoolExecutor(max_workers=NUM_WORKERS) as executor:\n",
    "        for i in range(0, len(documents), BATCH_SIZE):\n",
    "            batch = documents[i : i + BATCH_SIZE]\n",
    "            executor.submit(upload_batch, batch)\n",
    "\n",
    "index_name = \"dbpedia-1m-all\"\n",
    "client = create_search_client(index_name)\n",
    "upload_documents_to_index(client, transformed_documents)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Index Statistics Retrieval\n",
    "\n",
    "This Python code block retrieves and prints the statistics for multiple Azure Search indexes by iterating over a list of index names."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'document_count': 1000000, 'storage_size': 26601449427, 'vector_index_size': 6198717060}\n",
      "{'document_count': 1000000, 'storage_size': 13650457731, 'vector_index_size': 6198697620}\n",
      "{'document_count': 1000000, 'storage_size': 21992964504, 'vector_index_size': 1590725460}\n",
      "{'document_count': 1000000, 'storage_size': 20651723267, 'vector_index_size': 246717680}\n",
      "{'document_count': 1000000, 'storage_size': 7684412297, 'vector_index_size': 257431340}\n"
     ]
    }
   ],
   "source": [
    "index_names = [\n",
    "    \"dbpedia-1m-baseline\",\n",
    "    \"dbpedia-1m-stored\",\n",
    "    \"dbpedia-1m-scalar-quantization\",\n",
    "    \"dbpedia-1m-binary-quantization\",\n",
    "    \"dbpedia-1m-all\",\n",
    "]\n",
    "\n",
    "for index_name in index_names:\n",
    "    index_stats = index_client.get_index_statistics(index_name)\n",
    "    print(index_stats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Index Statistics Comparison\n",
    "\n",
    "The bar chart below presents a comparison of the storage size and vector index size for four different index configurations in Azure AI Search. Key takeways include:\n",
    "\n",
    "- `dbpedia-1m-baseline` represents the standard index size without optimizations.\n",
    "- `dbpedia-1m-stored` shows a reduced storage size, indicating that not storing certain properties can lead to cost savings.\n",
    "- `dbpedia-1m-scalar-quantization` demonstrates a significant reduction in the vector index size, thanks to scalar quantization compressing the vector data.\n",
    "- `dbpedia-1m-binary-quantization` shows further reduction in vector index size using binary quantization.\n",
    "- `dbpedia-1m-all` combines the benefits stored=False with binary quantization, resulting in the most substantial storage savings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Create a DataFrame to store the statistics\n",
    "df = pd.DataFrame()\n",
    "\n",
    "for index_name in index_names:\n",
    "    index_stats = index_client.get_index_statistics(index_name)\n",
    "    # Assuming index_stats is a dictionary, convert it to a DataFrame and transpose it\n",
    "    index_df = pd.DataFrame(index_stats, index=[0]).T\n",
    "    # Add the DataFrame to our main DataFrame, setting the column name to the index name\n",
    "    df[index_name] = index_df[0]\n",
    "\n",
    "# Filter the DataFrame to only include the 'storageSize' and 'vectorIndexSize' statistics\n",
    "df = df.loc[['storage_size', 'vector_index_size']]\n",
    "\n",
    "# Convert Bytes to Gigabytes\n",
    "df = df / (1024 ** 3)\n",
    "\n",
    "# Plot the DataFrame\n",
    "ax = df.plot(kind='bar', figsize=(12, 6))\n",
    "plt.title('Index Statistics Comparison')\n",
    "plt.ylabel('Value (GB)')\n",
    "plt.xlabel('Statistic')\n",
    "\n",
    "# Add labels to the top of the bars\n",
    "for p in ax.patches:\n",
    "    height = p.get_height()\n",
    "    ax.annotate(f'{height:.1f}', \n",
    "                (p.get_x() + p.get_width() / 2., height), \n",
    "                ha='center', va='center', \n",
    "                xytext=(0, 10), \n",
    "                textcoords='offset points')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Total Size by Index and Statistic\n",
    "\n",
    "The stacked bar chart below illustrates the total size of each index configuration, divided into `storage_size` and `vector_index_size`. Key takeaways include:\n",
    "\n",
    "- `dbpedia-1m-baseline` shows the largest total size, indicating the original unoptimized index.\n",
    "- `dbpedia-1m-stored` reduces the total index size due to optimizations in the storage properties.\n",
    "- `dbpedia-1m-scalar-quantization` further decreases the total size with an emphasis on compressing the vector index, illustrating the impact of scalar quantization.\n",
    "- `dbpedia-1m-both` achieves the lowest total size by combining storage property optimizations with scalar quantization, showcasing the most efficient use of storage.\n",
    "\n",
    "This visualization underscores the effectiveness of Azure AI Search's optimization features in reducing the overall index size, leading to potential cost savings while supporting efficient search capabilities.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Stacked bar chart\n",
    "df.transpose().plot(kind='bar', stacked=True)\n",
    "plt.title('Total Size by Index and Statistic')\n",
    "plt.ylabel('Value (GB)')\n",
    "plt.xlabel('Index')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Total Size by Index and Statistic\n",
    "\n",
    "The stacked bar chart below illustrates the total size of each index configuration, divided into `storage_size` and `vector_index_size`. Key takeaways include:\n",
    "\n",
    "- `dbpedia-1m-baseline` shows the largest total size, indicating the original unoptimized index.\n",
    "- `dbpedia-1m-stored` reduces the total index size due to optimizations in the storage properties.\n",
    "- `dbpedia-1m-scalar-quantization` further decreases the total size with an emphasis on compressing the vector index, illustrating the impact of scalar quantization.\n",
    "- `dbpedia-1m-binary-quantization` demonstrates a significant reduction in total size by applying binary quantization techniques.\n",
    "- `dbpedia-1m-all` achieves the lowest total size by combining multiple optimization strategies, showcasing the most efficient use of storage.\n",
    "\n",
    "This visualization underscores the effectiveness of Azure AI Search's optimization features in reducing the overall index size, leading to potential cost savings while supporting efficient search capabilities.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Search Relevance and Score Consistency\n",
    "\n",
    "The search results across different index configurations—`dbpedia-1m-baseline`, `dbpedia-1m-stored`, `dbpedia-1m-scalar-quantization`, `dbpedia-1m-binary-quantization`, and `dbpedia-1m-all`—showcase that the application of scalar quantization, binary quantization, and stored property optimizations does not adversely affect search relevance.\n",
    "\n",
    "The scores across the indexes remain consistent, underscoring that even with significant storage savings, the precision and quality of search results are maintained. Note that reranking and oversampling are recommended for mitigating the slight recall loss that comes with quantization."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results for index: dbpedia-1m-baseline\n",
      "+--------------------------------------+------------------+------------+\n",
      "|                  ID                  |      Title       |   Score    |\n",
      "+--------------------------------------+------------------+------------+\n",
      "| f9557902-e844-4239-9e8e-efb61ace958f |   Victor Yngve   | 0.88129556 |\n",
      "| 4c9ecd66-a189-4769-9843-b53e4b4152fd |    Martin Kay    | 0.87301606 |\n",
      "| 0ddbcd16-69ff-4bd3-b6b2-32262f0d68d6 |    Cliff Shaw    | 0.86885595 |\n",
      "| 2bce1c13-5bb4-424b-8523-c54f858d521a | Ronald Langacker | 0.86687255 |\n",
      "| b5a938a3-8dce-463d-a7df-b67add957112 |  Leonard Talmy   | 0.8652491  |\n",
      "+--------------------------------------+------------------+------------+\n",
      "\n",
      "\n",
      "Results for index: dbpedia-1m-stored\n",
      "+--------------------------------------+------------------+------------+\n",
      "|                  ID                  |      Title       |   Score    |\n",
      "+--------------------------------------+------------------+------------+\n",
      "| f9557902-e844-4239-9e8e-efb61ace958f |   Victor Yngve   | 0.88129556 |\n",
      "| 4c9ecd66-a189-4769-9843-b53e4b4152fd |    Martin Kay    | 0.87301606 |\n",
      "| 0ddbcd16-69ff-4bd3-b6b2-32262f0d68d6 |    Cliff Shaw    | 0.86885595 |\n",
      "| 2bce1c13-5bb4-424b-8523-c54f858d521a | Ronald Langacker | 0.86687255 |\n",
      "| b5a938a3-8dce-463d-a7df-b67add957112 |  Leonard Talmy   | 0.8652491  |\n",
      "+--------------------------------------+------------------+------------+\n",
      "\n",
      "\n",
      "Results for index: dbpedia-1m-scalar-quantization\n",
      "+--------------------------------------+------------------+------------+\n",
      "|                  ID                  |      Title       |   Score    |\n",
      "+--------------------------------------+------------------+------------+\n",
      "| f9557902-e844-4239-9e8e-efb61ace958f |   Victor Yngve   | 0.8812961  |\n",
      "| 4c9ecd66-a189-4769-9843-b53e4b4152fd |    Martin Kay    | 0.87301636 |\n",
      "| 0ddbcd16-69ff-4bd3-b6b2-32262f0d68d6 |    Cliff Shaw    |  0.868855  |\n",
      "| 2bce1c13-5bb4-424b-8523-c54f858d521a | Ronald Langacker | 0.86687165 |\n",
      "| b5a938a3-8dce-463d-a7df-b67add957112 |  Leonard Talmy   | 0.8652486  |\n",
      "+--------------------------------------+------------------+------------+\n",
      "\n",
      "\n",
      "Results for index: dbpedia-1m-binary-quantization\n",
      "+--------------------------------------+------------------+------------+\n",
      "|                  ID                  |      Title       |   Score    |\n",
      "+--------------------------------------+------------------+------------+\n",
      "| f9557902-e844-4239-9e8e-efb61ace958f |   Victor Yngve   | 0.8812961  |\n",
      "| 4c9ecd66-a189-4769-9843-b53e4b4152fd |    Martin Kay    | 0.87301636 |\n",
      "| 0ddbcd16-69ff-4bd3-b6b2-32262f0d68d6 |    Cliff Shaw    |  0.868855  |\n",
      "| 2bce1c13-5bb4-424b-8523-c54f858d521a | Ronald Langacker | 0.86687165 |\n",
      "| b5a938a3-8dce-463d-a7df-b67add957112 |  Leonard Talmy   | 0.8652486  |\n",
      "+--------------------------------------+------------------+------------+\n",
      "\n",
      "\n",
      "Results for index: dbpedia-1m-all\n",
      "+--------------------------------------+------------------+------------+\n",
      "|                  ID                  |      Title       |   Score    |\n",
      "+--------------------------------------+------------------+------------+\n",
      "| f9557902-e844-4239-9e8e-efb61ace958f |   Victor Yngve   | 0.8812961  |\n",
      "| 4c9ecd66-a189-4769-9843-b53e4b4152fd |    Martin Kay    | 0.87301636 |\n",
      "| 0ddbcd16-69ff-4bd3-b6b2-32262f0d68d6 |    Cliff Shaw    |  0.868855  |\n",
      "| 2bce1c13-5bb4-424b-8523-c54f858d521a | Ronald Langacker | 0.86687165 |\n",
      "| b5a938a3-8dce-463d-a7df-b67add957112 |  Leonard Talmy   | 0.8652486  |\n",
      "+--------------------------------------+------------------+------------+\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from prettytable import PrettyTable\n",
    "from azure.search.documents import SearchClient\n",
    "from azure.search.documents.models import VectorizableTextQuery\n",
    "from azure.core.credentials import AzureKeyCredential\n",
    "\n",
    "query = \"Who is considered one of the pioneers in computational linguistics?\"\n",
    "\n",
    "for index_name in index_names:\n",
    "    # Create a new SearchClient for each index\n",
    "    search_client = SearchClient(\n",
    "        endpoint=SEARCH_SERVICE_ENDPOINT,\n",
    "        index_name=index_name,\n",
    "        credential=AzureKeyCredential(SEARCH_SERVICE_API_KEY),\n",
    "    )\n",
    "    vector_query = VectorizableTextQuery(\n",
    "        text=query, k_nearest_neighbors=5, fields=\"embedding\"\n",
    "    )\n",
    "\n",
    "    results = search_client.search(\n",
    "        search_text=None,\n",
    "        vector_queries=[vector_query],\n",
    "        top=5,\n",
    "    )\n",
    "\n",
    "    # Create a PrettyTable instance\n",
    "    table = PrettyTable()\n",
    "    table.field_names = [\"ID\", \"Title\", \"Score\"]\n",
    "\n",
    "    for result in results:\n",
    "        table.add_row([result['id'], result['title'], result['@search.score']])\n",
    "\n",
    "    print(f\"Results for index: {index_name}\")\n",
    "    print(table)\n",
    "    print(\"\\n\")"
   ]
  }
 ],
 "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}