Hello,
With my team we are using Document Understanding framework to Classify and Extract informations from pdf.
We start a big execution with around 8000 pdf and we got this error on some of them : “Classify Document Scope - Classify: Argument can’t be empty
Parameter name: documentText”
As you can see on the picture in debug mode we are facing to a strange case, the argument is not empty and the Classify Scope say the reverse.
Does the same always happen on the specific documents? Any chance you could share a sample of such document that still reproduces the issue, maybe together with a sample zip of a dummy project?
Hi @loginerror,
Unfortunately, for privacy reason I’m not able to share doc with you.
I tried to solve by upgrading my dependencies, but no effect.
I tried another way to solve my problem.
When I let only one classifier inside the Classify Scope, it works and the execution pass to the next activity.
But combining several Classifier is not possible
How can we reproduce the issue, what documents are you using/what is particular to the documents leading to the documents? Are there native/scanned PDFs?
For your information, I opened a ticket at the support team and we identify that the problem is coming from the fact that on some documents combining several classifier will result with this error.
Support Answer :
"Hello Mourad,
An issue was identified testing with the sample workflow you have provided.
An internal ticket was opened with the development team for further investigation. For now, there is no ETA provided.
Best regards! @cristian.oprescu
"
As you can see on the picture I reproduced the issue with the sample project provided by the support.
