How to import multiple documents into DU at once

Help Document Understanding
1

I have 10 documents. We would like to have them imported into the DU in batches at night, and users can start checking the classification from ActionCenter after they arrive at work in the morning. The work after that is not necessarily sequential, as many people will be involved: some will classify and extract one at a time, while others will just run classification first and then run extraction in batches.
With the current DU framework, it looks like we have to finish the extraction for one document before we can process the next document.

  1. Is this flow feasible?
  2. If feasible, is there a way?
2

Hi @ohta ,

I dont think with the DU Framework this is achievable, if you decide to step out of the DU Framework this can be acheived.

Steps-

  1. For Each file classify and extract if confidence score passes threshold. Add the items to a list seperately for Classification and Extraction where the items dont pass the threshold.
  2. Have a two other loops to traverse through the classification and move to extraction. and have a similar validatioin for the extraction confidence
  3. Have another loop for extraction only and export.

You can incorporate the RE Framework to handle the errors better.

3

Hi @DanRagh

Thank you for your answer.
After all, that is the way to create it.
It seems to me that I could manage to use Orchestration Proces as another method, but what do you think?

4

Hi @ohta

You can definetly try that out, I believe just an RE framework will work fine too as you will be enabling the persistance.

Please do let us know it works out for you.

Regards.

5

You can use parallel for each instead of for each in the DU framework. If you use the framework its a longe running process.