Power Platform Community Forum Thread Details

I created a simple Forms Processing model to read a registration form, then save the field values to SharePoint. Some of the forms are coming in with handwritten input, mix of handwritten and digital, and others are all digital. Digital meaning the user filled out the form on their computer, then submitted it.

Fields:
User_Name
User_Address

Some of the form fields being processed are picking up the form field and the surrounding text.
Example: I'm wanting to capture User_Address, the forms processing engine picked up address value and some text from neighboring fields.

Question:
On a form by form basis, is it possible to see how the engine processed the form and correct it?

Categories:

AI Builder

Hi,

Thanks for posting this question. There’s a number of things you can do:

For each extracted field, the AI model also returns the associated confidence score. You can put a condition that if a confidence score is lower that 65%, for example, you can trigger an email alert for somebody to review the extracted data. You can then edit the AI model to add that document to the training set to improve the model performance as described here: https://docs.microsoft.com/en-us/ai-builder/improve-form-processing-performance
If you are looking for a side-by-side viewer to review the extracted data through a validation station you can leverage the new Document Automation solution: https://docs.microsoft.com/en-us/ai-builder/doc-automation
If you are seeing neighboring fields being picked up by the User_Address field, tag those adjacent fields as new separate fields in case you haven’t done so already. The AI model will then better learn the limits for each field.

Hope this helps. Let us know if we can further assist!