Hyphen not always being picked up by custom document processing model

(0) Share

Report

Posted on by AVFC

147

Hello

I'm using a structured and have also used un-structured custom document processing model to capture table data from a pdf but as you can see below image it sometimes misses out the hyphen that i need in the amount column. Really can't see why it's doing this? There's no consistency to it..

Categories:

AI Builder

All responses (3)

Answers (1)

Sort by

AVFC 147 on at

Like
a
(0)

Report
Copy link

Link copied!

If someone from MS can please advise as i need to get this over the line for a finance project.
samiak Microsoft Employee on at

Like
a
(1)

Report
Copy link

Link copied!

Hello,

You are facing a known issue related to the OCR engine. Our team is working to improve that capability.

As a workaround, you could make the logic conditional or use GPT data extraction method.

This post should help you building that:

Custom model can no longer handle hyphens ( - ) in... - Power Platform Community (microsoft.com)

Hope that helps,

Samia
Verified answer

AVFC 147 on at

Like
a
(0)

Report
Copy link

Link copied!

Hi Samia

Thanks for the quick response and suggestions provided. This has now been resolved bizarrely by playing around with the settings in print preview on the original Excel file. The OCR is now picking up the hyphen after quite a few tests. It's just really weird that it was picking up the hyphen some of the time before but as you say your team is looking into improving the algorithm.

Many Thanks

Rob