Power Platform Community Forum Thread Details

We are using AI builder document processing model for our data extraction created the one collection so far. we are facing one issue while tagging for item table where the model is lacking to extract the information for below explain scenario.

We have table not on next page but on page 2 ,5 ,7 like this. we have tried tagging but the model was not allow to do tagging like this. Also the model accuracy score is degraded.

Please suggest some solution to handle this kind of scenario.

Thank you,

Divya Patel

Model Test.jpg

Your file is currently under scan for potential threats. Please wait while we review it for any viruses or malicious content.

Categories:

AI Builder

@DP-21050643-0

This is a common challenge with AI Builder Document Processing models. Multi-page tables that appear on non-consecutive pages (for example, pages 2, 5, and 7) are difficult for the model to interpret consistently, which can impact extraction accuracy.

A few things you can try:

During training, make sure the table is tagged across multiple sample documents so the model can learn the repeated structure better.
If the tables on different pages are independent sections, try tagging them as separate tables instead of one continuous table.
Keep the table structure and column headers as consistent as possible across documents, as AI Builder performs better with standardized layouts.
If your documents have very different formats or page patterns, consider creating separate models/collections for different document types.

In many cases, AI Builder can extract the table data page by page, and then Power Automate can be used afterward to merge or consolidate the extracted rows into a single output.

Vishnu WR

LinkedIn - My Tech Space

Please ✅ Does this answer your question if my post helped you solve your issue. This will help others find it more readily. It also closes the item. If the content was useful in other ways, please consider answering Yes to Was this reply helpful? or give it a Like ♥