Unanswered

High Confidence Score but Incorrect Data Extraction

(0) Share

Report

Posted on by SY-10100932-0

I have trained the PDF model multiple times and achieved a good confidence score of 99%, but while testing the files, we’re still not getting accurate data. How can this be resolved?

Categories:

AI Builder

image.png

I have the same question (0)

All responses (2)

Answers (0)

stampcoin 4,413 Super User 2025 Season 2 on at

Like (0)

Report

High Confidence Score but Incorrect Data Extraction

@SY-10100932-0 Hi,

I would suggest that use more data for training;
and use standard pdf;
and provide more variant.

Use feedback, you can refer this link:

https://learn.microsoft.com/en-us/ai-builder/feedback-loop

good luck.

Enjoy Power Platform and have a great day 🚀 | My LinkedIn

If the answer helps, please consider 👍, Thanks.

Was this reply helpful? Yes No
developerAJ 3,778 on at

Like (0)

Report

High Confidence Score but Incorrect Data Extraction

The accuracy of the data depends on how well the model is trained with different formats and multiple documents for each type. It looks like you’ve already done that, but try increasing the number of documents. When it comes to implementation, ensure that your file closely matches the template used during training

Was this reply helpful? Yes No