Why after training, "Test" only looked at 2 pages of my test document, instead of looking at the entire document?

Question

In Azure AI Document Intelligence Studio, I am using Custom Extraction Model.

As a trustee from a board, my goal is to be able to analyze Check Journals in PDF Format.

The check journals are 10 to 20 pages in table format of check number, date, code, vendor name, check comment and check amount. If i want to analyze 12 months of check expenses, I would need to download 12 pdf files (each pdf file are between 5 to 20 pages long).

To train successfully my model, I created 7 samples of 1 page pdf file that show multiple scenarios. The training was successful.

Issues happened under TEST.

My question is about TEST. The pdf file has 10 pages. For some reason, the TEST only looked at 2 pages. I was hoping it would look at all the pages.

What did i do wrong? I looked at my configuration, but I can't seem to find the right option to click to make this work.

Please help. I can be reached at edhafrance@gmail.com

Thank you.

Ed

Answer 1

Possible solutions:

Increase training data: Include more multi-page documents in your training set.

Review model settings: Ensure the model is configured to process multi-page documents correctly.

Test with more documents: Test with multiple 10-page documents to ensure consistent results.