Editar

Compartir a través de

Semantic operators in the Azure AI extension Preview

The Azure AI extension introduces Semantic Operators, a groundbreaking feature integrating advanced Generative AI (GenAI) functionality directly into PostgreSQL SQL. With Azure OpenAI chat completion APIs, these operators allow users to explore innovative ways of building GenAI applications directly within their databases.

Key features

The Semantic Operators provide users with four core SQL functions that use generative AI capabilities:

  • azure_ai.generate(): Generates text or structured output using (Large Language Models (LLMs).
  • azure_ai.is_true(): Evaluates the likelihood that a given statement is true.
  • azure_ai.extract(): Extracts structured features or entities from text.
  • azure_ai.rank(): Reranks a list of documents based on relevance to a given query.

Each function operates via AI Foundry endpoints registered using the azure_ai.set_setting function, ensuring seamless integration and user control.

Understanding semantic operators

Semantic Operators in the Azure AI extension are designed to simplify complex AI-driven tasks directly within your PostgreSQL database. These operators allow users to seamlessly integrate generative AI capabilities into their SQL workflows, enabling advanced text generation, truth evaluation, entity extraction, and document ranking. Each operator is optimized for ease of use and flexibility, allowing developers to build intelligent applications with minimal effort.

azure_ai.generate()

This function uses LLMs to generate text or structured output and supports custom parameters such as prompts, JSON schema, model selection, temperature, time-out, and retry logic.

Example usage:

SELECT azure_ai.generate(
  'Rewrite the following comment to be more polite: '
 comment_text
) AS polite_comment
FROM user_comments;

azure_ai.is_true()

This operator evaluates whether a given statement is likely true, returning a boolean value or NULL if inconclusive.

Example usage:

SELECT azure_ai.is_true(
  'The review talks about the product: '
 product_name
  ' Review: '
 review_text
) AS is_relevant_review
FROM product_reviews;

azure_ai.extract()

Extract structured features or entities from text based on user-defined labels.

Example usage:

SELECT azure_ai.extract(
  'Alice Smith traveled to Paris.',
 ARRAY['person', '___location', 'action']
);
-- Output: {"person": "Alice Smith", "___location": "Paris", "action": "travel"}

azure_ai.rank()

Reranks documents based on query relevance, supporting cross-encoder and GPT models.

Example usage:

SELECT azure_ai.rank(
  'How to Care for Indoor Succulents',
 ARRAY[
    'A complete guide to watering succulents.',
    'Best outdoor plants for shade.',
    'Soil mixtures for cacti and succulents.'
 ]
) AS ranked_documents;

Related content