I have two questions about Decipher IDP.

[sy_lee]

Hello,

I have two questions regarding the use of Decipher IDP.

First, I have set the skip user setting for Verify within IDP. Is there a feature within Decipher IDP that can send notifications when Verify occurs? Or is it best to check the status via Batch ID and send an email?

Second, I want to activate the training in IDP, but it doesn't seem to be working. In Capture Models, it shows "Last Trained: n/a," but in Training Data, it says "51 training data items found." What could be the issue? I have set it to train every 50 items in Capture Model and also marked for training, but there is no change. Could anyone provide a solution?

Thank you.

Image)

[Nandhakumar · Accepted solution]

Hi @sy_lee ,

1. To notify the users when there is any batches to verify

There is no default notification feature available with decipher. 

2. Training Model

Can you confirm whether 51 training data items found from the dropdown are pointing to respective Document Type?

Thanks,

Nandha

[Ben.Lyons1]

Hi @sy_lee ,

I can confirm @Nandhakumar  is correct with respect to question 1. Notification of pending batches can only be done through a BP process at this time.

Training data is different to the ML capture models set up in the document type, I would suggest reviewing the types of machine learning information here.

I also would like to highlight that 50 documents is not enough to create a suitable ML capture model and suggest you familiarise yourself with our best practice guide before getting too far along.

Thanks

[sy_lee]

Upon further review of item 2, we have confirmed that the specified Document Type has not yet reached 50. Thank you.

[sy_lee]

I am aware that 50 documents are insufficient, but as the person managing the system, I needed to ensure the functionality was working correctly. Thank you for mentioning it again.

[Ben.Lyons1]

Hi @sy_lee ,

The 51 training items found relate to the number of region templates, which can be considered as 51 different document layouts. So this could be 51 layouts from 1000 trained documents, or 51 layouts from 51 trained documents. These are created and updated every time a batch completes data verification.

I would expect that at least 1 of those was for the document type you're referencing, if that's the one selected when processing the documents to test the system.

Other things to check would be:

1. Ensure ML has been enabled in the config file, Decipher IDP post-installation configuration
2. Batches have been fully processed i.e. marked as complete using the Decipher VBO. Decipher will not count documents towards the available documents if they are still waiting to be exported.

Thanks