Azure AI Search Cheat Sheet

• Formerly known as Azure Cognitive Search
• A scalable cloud search service that lets you embed advanced search capabilities into your apps.
• It supports full-text, vector, and hybrid search approaches, making it ideal for both classic and AI-enhanced use cases.
• Natively integrates with Azure OpenAI, Foundry Models, and Azure Machine Learning.
• Supports third-party and open-source LLM integrations.

Key Features

1. Search Capabilities

• Full-Text Search: Utilizes inverted indexes for efficient keyword-based searches.
• Vector Search: Supports semantic search by matching vector representations of data.
• Hybrid Search: Combines full-text and vector search for enhanced relevance.
• Indexers: Automate data ingestion from sources like Azure Blob Storage and Cosmos DB, with support for scheduled or on-demand indexing

2. Intelligent Data Processing

• Extracts structured information from unstructured content using pre-defined cognitive capabilities like OCR, language detection, and key phrase extraction.
• Allows you to insert custom logic during indexing to meet specialized data transformation needs.
• Combines multiple skills to create step-by-step processing workflows before indexing the data.

3. Semantic Ranking

• Improve the accuracy of search results by reordering them based on how well the system understands the meaning of the query and documents.
• Create short summaries and emphasize important parts of the results to make them easier for users to understand.
• Automatically add related words to the user’s search to help find more matching results.

4. Other Relevant Search Features

• Location Awareness
  • Map-based filtering: Allows narrowing results using distance from a point or within defined boundaries like polygons.
• Filtering & Categorization
  
  • Navigation through facets: Enables UI-friendly filtering (e.g., by price or brand) with results organized into categories.
  • Customizable filters: Developers can also build advanced filtering logic using OData syntax to refine results.
• Improved User Experience
  • Offers suggestions as users type, improving speed and accuracy of input.
  • Surfaces actual documents, not just suggested terms, during partial input.
  • Broadens search scope by automatically including similar or related terms.
  • Emphasizes matched words in results, improving readability and context.
  • Easily reorder results by fields like date, price, or rating via search parameters.
  • Manage how many results are shown per page and apply limits to support performance needs.

4. Retrieval-Augmented Generation (RAG)

• Integration with LLMs: Combine Azure AI Search with LLMs like Azure OpenAI to build applications that generate responses grounded in enterprise data.
• Agentic Retrieval: Break down complex queries into sub-queries, retrieve relevant information, and synthesize comprehensive answers.

5. Security

• Role-Based Access Control: Manage access to search resources using Azure Active Directory roles.
• Private Endpoints: Secure communication between clients and the search service using private network connectivity.

Integration Options

• REST APIs
  Interact with Azure AI Search using standard RESTful APIs to manage indexes, upload documents, and run search queries.
• SDKs
  Easily integrate search capabilities into apps using SDKs for .NET, Java, Python, and JavaScript.
• Azure Portal
  Use the visual, browser-based interface to set up, configure, and monitor your search services without writing code.
• Data sources from Azure services
  Connect directly to Azure Blob Storage and Cosmos DB. You can also set up custom data sources to feed your search index.

Monitoring and Billing

• Usage Metrics: Track service usage, query volume, and indexing operations via Azure Monitor.
• Billing: Charges depend on the number of search units provisioned, influencing capacity and performance.
• Scaling: Adjust the number of replicas and partitions to meet performance and availability requirements

References:

https://learn.microsoft.com/en-us/azure/search/search-what-is-azure-search
https://learn.microsoft.com/en-us/azure/search/search-features-list
https://learn.microsoft.com/en-us/azure/search/semantic-search-overview
https://learn.microsoft.com/en-us/azure/search/retrieval-augmented-generation-overview