แก้ไข

แชร์ผ่าน

What is Azure Language in Foundry Tools?

Azure Language is a cloud-based service that provides Natural Language Processing (NLP) features for understanding and analyzing text. Use this service to help build intelligent applications using the web-based Microsoft Foundry, REST APIs, and client libraries. For AI agent development, the service capabilities are also available as tools in Azure Language MCP server, which is available both as a remote server in the Microsoft Foundry Tool Catalog and as a local server for self-hosted environments.

Core capabilities

Recommended for new development

Core capabilities are the primary, actively evolving features of Azure Language. These features receive ongoing investment and feature updates, and are recommended for new development and long-term planning. If you are starting a new project or designing a future architecture, use core capabilities as the foundation for your natural language processing workflows.

Tip

Unsure which feature to use? See Which Azure Language core feature should I use to help you decide.

Microsoft Foundry enables you to use most of the following service features without the need to write code.

Language detection

Language detection evaluates text and detects a wide range of languages and variant dialects.

A screenshot of language detection in Foundry.

Custom NER

Custom named entity recognition (CNER) enables you to build custom AI models to extract custom entity categories (labels for words or phrases), using unstructured text that you provide.

A screenshot of a custom NER example.

Prebuilt NER

Prebuilt named entity recognition (NER) identifies different entries in text and categorizes them into predefined types.

A screenshot of named entity recognition in Foundry.

Personally identifiable information (PII) detection

Important

The Azure Language in Foundry Tools text personally identifiable information (PII) detection anonymization feature (synthetic replacement) is currently available in preview and licensed to you as part of your Azure subscription. Your use of this feature is subject to the terms applicable to Previews as described in the Supplemental Terms of Use for Microsoft Azure Previews and the Microsoft Products and Services Data Protection Addendum (DPA).

Personally identifiable information (PII) detection identifies entities in text and conversations (chat or transcripts) that are associated with individuals.

Conversation PII detection

A screenshot of conversation personally identifying information in Foundry.

Text PII detection

A screenshot of text personally identifying information in Foundry.

Text analytics for health

Text analytics for health extracts and labels relevant health information from unstructured text.

A screenshot of text analytics for health in Foundry.

Legacy capabilities

Supported for existing implementations

Legacy capabilities are established features that provide a stable, supported base for existing workloads and scenarios. These features are supported for existing implementations and established use cases.

Conversational language understanding

Conversational language understanding (CLU) enables users to build custom natural language understanding models to predict the overall intention of an incoming utterance and extract important information from it.

A screenshot of a conversational language understanding example.

Custom text classification

Custom text classification enables you to build custom AI models to classify unstructured text documents into custom classes you define. Creating a custom text classification project typically involves several different steps:

A screenshot of a custom text classification example.

Entity linking

Entity linking is a preconfigured feature that disambiguates the identity of entities (words or phrases) found in unstructured text and returns links to Wikipedia.

A screenshot of an entity linking example.

Key phrase extraction

Key phrase extraction is a preconfigured feature that evaluates and returns the main concepts in unstructured text, and returns them as a list.

A screenshot of key phrase extraction in Foundry.

Orchestration workflow

Orchestration workflow is a custom feature that enables you to connect conversational language understanding (CLU) AND custom question answering (CQA) applications.

A screenshot of an orchestration workflow example.

Question answering

Question answering is a custom feature that identifies the most suitable answer for user inputs. This feature is typically utilized to develop conversational client applications, including social media platforms, chat bots, and speech-enabled desktop applications.

A screenshot of a question answering example.

Sentiment analysis and opinion mining

Sentiment analysis and opinion mining preconfigured features that help you understand public perception of your brand or topic. These features analyze text to identify positive or negative sentiments and can link them to specific elements within the text.

A screenshot of sentiment analysis in Foundry.

Summarization

Summarization condenses information for text and conversations (chat and transcripts).

Conversation summarization

Conversation summarization recaps and segments long meetings into timestamped chapters.

A screenshot of conversation summarization in Foundry.

Call center summarization

Call center summarization summarizes customer issues and resolution.

A screenshot of call center summarization in Foundry.

Text summarization

Text summarization generates a summary, supporting two approaches:

  • Extractive summarization creates a summary by selecting key sentences from the document and preserving their original positions.

  • Abstractive summarization generates a summary by producing new, concise, and coherent sentences or phrases that aren't directly copied from the original document.

    A screenshot of text summarization in Foundry.

Available tools

Azure Language provides specialized tools that enable seamless integration between AI agents and language processing services through standardized protocols.

Azure Language MCP server

The MCP (Model Context Protocol) server creates a standardized bridge that connects AI agents directly to Azure Language services through industry-standard protocols. This integration enables developers to build sophisticated conversational applications with reliable natural language processing capabilities while ensuring enterprise-grade compliance, data protection, and processing accuracy throughout their AI workflows.

Azure Language provides both remote and local MCP server options:

  • Remote server: Available through Foundry Tool Catalog for cloud-hosted deployments.
  • Local server: Available for developers who prefer to host the server in their own environment.

For more information, see Azure Language MCP server.

Azure Language agents

Azure Language offers prebuilt agents that handle specific conversational AI scenarios with built-in governance, routing logic, and quality control mechanisms.

Azure Language Intent Routing agent

The Intent Routing agent intelligently manages conversation flows by understanding user intentions and delivering accurate responses in conversational AI applications. This agent uses predictable decision-making processes combined with controlled response generation to ensure consistent, reliable interactions that organizations can trust and monitor.

For more information, see Azure Language Intent Routing agent.

Azure Language Exact Question Answering agent

The Exact Question Answering agent provides reliable, word-for-word responses to your most important business questions. This agent automates frequently asked questions while maintaining human oversight and quality control to ensure accuracy and compliance.

For more information, see Azure Language Exact Question Answering agent.

Which core Language feature should I use?

This section helps you decide which core Language feature you should use for your application:

What do you want to do? Document format Your best solution Is this solution customizable?*
Detect and/or redact sensitive information such as PII and PHI. Unstructured text,
transcribed conversations		 PII detection
Extract categories of information without creating a custom model. Unstructured text The preconfigured NER feature
Extract categories of information using a model specific to your data. Unstructured text Custom NER ✓
Extract medical information from clinical/medical documents, without building a model. Unstructured text Text analytics for health

* If a feature is customizable, you can train an AI model using our tools to fit your data specifically. Otherwise a feature is preconfigured, meaning the AI models it uses can't be changed. You just send your data, and use the feature's output in your applications.

Which legacy Language feature should I use?

This section helps you decide which legacy Language feature you should use for your application:

What do you want to do? Document format Your best solution Is this solution customizable?*
Extract main topics and important phrases. Unstructured text Key phrase extraction
Determine the sentiment and opinions expressed in text. Unstructured text Sentiment analysis and opinion mining
Summarize long chunks of text or conversations. Unstructured text,
transcribed conversations.		 Summarization
Disambiguate entities and get links to Wikipedia. Unstructured text Entity linking
Classify documents into one or more categories. Unstructured text Custom text classification ✓
Detect the language that a text was written in. Unstructured text Language detection
Predict the intention of user inputs and extract information from them. Unstructured user inputs Conversational language understanding ✓
Connect apps from conversational language understanding and custom question answering. Unstructured user inputs Orchestration workflow ✓
Build a conversational application that responds to user inputs. Unstructured user inputs Question answering ✓

* If a feature is customizable, you can train an AI model using our tools to fit your data specifically. Otherwise a feature is preconfigured, meaning the AI models it uses can't be changed. You just send your data, and use the feature's output in your applications.

Tutorials

After you get started with Azure Language quickstarts, try our tutorials that show you how to solve various scenarios.

Code samples

You can find more code samples on GitHub for the following languages:

Deploy on premises using Docker containers

Use Language containers to deploy API features on-premises. These Docker containers enable you to bring the service closer to your data for compliance, security, or other operational reasons. The Language offers the following containers:

Responsible AI

An AI system includes not only the technology, but also the people who use it, the people affected by it, and the deployment environment. Read the following articles to learn about responsible AI use and deployment in your systems:

  • Last updated on