Others

How to Extract Values with Multiple Labels in Azure Document Intelligence

Maira Mehtab Sep 20, 2026 868 views

Question details

The user needs to know how to extract a single data value when it is associated with multiple different label names (such as tractor number, truck number, fleet ID, or unit number) using Azure AI Document Intelligence Custom Models.

Product: Azure AI Document Intelligence
Device & OS: not provided
Scenario: Training or configuring a custom document model to recognize varying textual labels that all refer to the same logical data field.
Observed behavior: The system must accurately identify and map differently labeled fields on various documents into a single standardized output value.

Before you start

Ensure you have an active Azure subscription, access to the Azure AI Document Intelligence Studio, and permissions to manage and train custom models within your workspace.

Solution 1Recommended

Seek Specialized Guidance on Microsoft Q&A

Since mapping dynamic label variations requires specific schema configurations and training strategies, consulting the Azure AI Services community is the most effective way to resolve complex custom model challenges.

Azure Document Intelligence requires precise dataset labeling to handle field variability. When multiple distinct labels map to the same logical field, the model training can become complex, often requiring nuanced tagging strategies or post-processing logic.

Navigate to Microsoft Q&A

Open your web browser and visit the official Microsoft Q&A support platform.

Select the correct tags

Filter the forums using the 'Azure AI Services' and 'Azure AI Document Intelligence' tags to ensure your question reaches the right product engineers and experts.

Post your specific labeling scenario

Create a detailed post describing your custom model's schema. List the various labels (e.g., tractor number, truck number, fleet ID) and provide examples of how they appear on your documents to get the most accurate advice.

Post-Processing Alternative: If the model struggles to map them natively, consider extracting them as separate optional fields and using your application's backend code to merge them into a single variable after the API response is received.

Free Microsoft Office alternative

Enhance Your Document Management with WPS Office

While Azure Document Intelligence provides specialized API tools for data extraction, WPS Office offers a free, lightweight, and highly compatible alternative to Microsoft Office for your everyday document creation, editing, and reviewing needs.

1. Visit the official website: Go to the WPS Office official website to access the latest version.
2. Download the software: Click the 'Free Download' button to get the lightweight installer for your operating system.
3. Install and explore: Run the installer, set up your workspace, and enjoy seamless compatibility with all your standard Office documents.

Fully compatible with Microsoft Word, Excel, and PowerPoint file formats.Seamless migration with a familiar, user-friendly interface.Lightweight software that consumes fewer system resources.Built-in AI tools for quick document summarization and smart content generation.

Frequently Asked Questions

How many samples do I need to train a custom model in Azure Document Intelligence?

You typically need a minimum of five sample documents of the same layout to train a custom template model. However, to handle highly variable labels like 'truck number' vs 'fleet ID', providing a larger and more diverse training dataset will significantly improve extraction accuracy.

Can Azure Document Intelligence handle handwritten labels?

Yes, Azure Document Intelligence is capable of extracting both printed and handwritten text. You must ensure that your training dataset includes clear examples of the handwritten label variations so the model learns to recognize them.

Is there a way to merge extracted fields post-processing?

Yes. If your custom model is trained to output distinct fields for 'tractor number' and 'unit number', you can easily write a script in your application logic to check which field is populated in the JSON response and map it to a single unified column in your database.

Office Quick Solutions