Unanswered

AI Builder Unstructured Document - Does not extract field

(0) Share

Report

Posted on by slypdog93

Hi,

I am using AI Builder to extract address data from an unstructured document. In my training collection I have 40 documents and have 6 tagged fields: Recipient, Address Line 1, Address Line 2, City, State, Zipcode. All fields are available in every document in my collection. I have trained the model and I'm now testing it with documents of the same layout as the training collection but new documents the model has not seen before.

My expectation was that AI Builder would return every field I have tagged but it is only returning Recipient and Address Line 1 consistently. The other fields are only returned at random even though they are present in every test document.

Is there a reason why AI Builder is not returning all fields? I would expect it would still return the field but with a low confidence score. Is there anything I can do to adjust my model to always return a value for every tagged field (I'm okay with a low confidence score but want it to at least attempt to find the field).

Categories:

AI Builder

I have the same question (0)

All responses (2)

Answers (0)

plarrue Moderator on at

Like (0)

Report

Hi @slypdog93 ,

Could you please share screenshots?

In the meantime, here's some interesting links.

Tag documents in a document processing model | Microsoft Learn

Fields couldn't be loaded for this document error in a document processing model - AI Builder | Microsoft Learn.

thanks,

Was this reply helpful? Yes No
takolota1 4,974 Moderator on at

Like (0)

Report

You could also try this new GPT data extraction method: https://powerusers.microsoft.com/t5/Power-Automate-Cookbook/Extract-Data-From-PDFs-and-Images-With-GPT/td-p/2201345

It doesn’t require specifying certain document areas, wordings, styles, data, etc. It just OCRs the file, converts it to a replica text (txt), and passes it to a GPT prompt where you can ask GPT to do whatever you want with the document data.

Even if you have some documents where the original model works better, you could set some criteria for your fields that may indicate a mistake & have it then run this method on it to double check if the custom model results agree with these results.

Was this reply helpful? Yes No