Hi @Sai Deepan.R ,
Please enable the below option,
Optional: Select the Generative AI-driven data extraction option to use the generative AI capabilities for extraction. For more information, see Document Automation data extraction using Generative AI.
Understand how the system improves extraction accuracy though user-provided changes in the Validator.
When a learning instance is created, the user has the option to enable this feature to send feedback to the learning instance based on user-provided changes in the Validator. In Document Automation, learning instances running in production mode can continuously "learn" whenever a user resizes or relocates the extraction region in the Validator
https://docs.automationanywhere.com/bundle/enterprise-v2019/page/enterprise-cloud/topics/iq-bot/native/idp-extraction.html#miy1682768247560__sec-validation-feedback
Please note that
- Feedback is only collected when the user changes the extraction region. If the user manually inputs text, the system does not collect feedback.
- The feedback file only contains data on the field location to improve extraction accuracy for subsequent documents
Hi @Tamil Arasu10 ,
Thanks for the reply and detailed documentation links. At the moment I have the optional Generative AI turned off.
We are doing a feasibility analysis and eventually will do a POC for client. I understand that validator helps the instance to understand the field position better. Eventually the accuracy would be better once once the instance learns from the validation task performed.
Can you please help me answer these questions
- One of the main question from a client is how many documents do we need to train before we go live? The end goal is avoid validator task for every single invoices uploaded, validated as part of the initial training.
- Feedback is only collected when the user changes the extraction region. If the user manually inputs text, the system does not collect feedback. - With each validator task I have been manually pointing to the right field and drawing the boxes. Will this help the instance to learn the field location and perform extraction on its own.
Hi @Sai Deepan.R
One of the main question from a client is how many documents do we need to train before we go live? The end goal is avoid validator task for every single invoices uploaded, validated as part of the initial training.
I'm not sure about the exact count ( no of documents ), i have seen the real the time each layout of invoices needs to be trained and whenever new layouts received that goes validation & during the validation was leaned 1st time itself.
Feedback is only collected when the user changes the extraction region. If the user manually inputs text, the system does not collect feedback. - With each validator task I have been manually pointing to the right field and drawing the boxes.(Yes, manually pointing will helps to learn )
Will this help the instance to learn the field location and perform extraction on its own.- Yes, Please try with few different layout invoices.