Content Understanding - Json Issue for invoice Data Extraction

Question

Hi Team,

we noticed Document understanding - json response for Invoice template - it extract billing address and value address. however, this valueAddress is missing in content understanding.

  "valueAddress": {

              "houseNumber": "601",

              "road": "4th AVE",

              "city": "Seattle",

              "state": "WA",

              "countryRegion": "US",

              "streetAddress": "601 4th AVE"

            }

it would be nice if you could enable this option in content understanding.

Answer 1

The issue you're encountering with the missing valueAddress in the Content Understanding response for invoice data extraction may stem from differences in the capabilities or configurations of the Document Understanding and Content Understanding services. While Document Understanding can extract structured address components, Content Understanding may not currently support the same level of detail for address extraction.

To address this, you might consider reviewing the configuration settings for your Content Understanding analyzers to ensure they are set up to capture the necessary details. Additionally, you can explore the possibility of customizing your analyzers to include specific fields like valueAddress if the platform allows for such modifications.

For further assistance, you may want to check the latest documentation or reach out to Microsoft support for guidance on feature requests or enhancements regarding address extraction capabilities in Content Understanding.

---

References:

• Document Intelligence invoice model
• What is a Content understanding analyzer?

Answer 2

Hi Sarath Kumar

Thank you for reaching out on the Microsoft Q&A.

Content Understanding and Document Intelligence use different extraction schemas:

• Document Intelligence (Invoice Model) Designed for structured document parsing, so it returns detailed address components like houseNumber, road, city, state, countryRegion, and streetAddress.
• Content Understanding Focuses on semantic analysis and generalized content extraction. It does not currently support granular address decomposition (valueAddress) and instead provides addresses as plain text or high-level entities.

Current Limitation

There is no configuration option in Content Understanding to enable valueAddress breakdown today. This is by design because CU prioritizes semantic understanding over structured field extraction.

Recommended Workarounds

• Hybrid Approach
  • Use Content Understanding for classification and summarization tasks.
  • Use Document Intelligence prebuilt-invoice or custom extraction models for structured fields like valueAddress.
• Post-Processing
  • If CU returns the full address as text, apply an address parser (e.g., Azure Maps Address Parser or regex logic) to split into components.

Feature Request

Your suggestion is valid and aligns with invoice workflows. To request this enhancement:

• Submit feedback via Azure Feedback Portal: https://feedback.azure.com
• Tag under Azure AI Document Intelligence / Content Understanding

References:

• Azure Document Intelligence Overview
• Troubleshooting Discrepancies in Document Extraction Results
• Document Intelligence Invoice Model
• Migration Guide for SDK

I Hope this helps. Do let me know if you have any further queries.

---

If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

Thank you!