Learned automation

Learned automation gives you the ability to speed up your workflow processes for classifying documents and applying metadata. It does this by using the information supplied by end-users and automatically creates "models" the system uses in the background to automatically classify the document and apply the appropriate metadata.

Document classification

The first phase in learned automation is document classification. Document classification allows you to automatically identify and categorize unique types of documents depending on their content. For example, you may want to differentiate between invoices and purchase orders.

 You can have a learned automation stencil in your workflow to do only document classification and not data extraction.

Data extraction

The second phase in learned automation is data extraction. A document must be classified before data extraction can take place. Once a document has been classified, data fields can quickly be located on the document, and data field values can automatically be captured.

An Administrator configures which data field is used to specify the document type when configuring learned automation. When the document type field and/or additional data fields are configured to do so, anytime the user provides or refines the data tagged for a document, the user's selections are automatically added to the identification and extraction models in the background. This is referred to as “training”.

When the document type field is displayed as unknown, the document classification model was unable to classify the document. The user should assign the document type. This document will then be added to the classification model.

Implementing learned automation

Administrators will need to complete the following tasks in order to implement learned automation:

  • Set up a workflow with a Learned Automation step. An OCR stencil still has to be included in a workflow that has a Learned Automation stencil.

  • For document classification, configure the Learned Automation stencil. See Learned Automation stencil for more information.

  • For data extraction, select the data fields that will be used for training purposes on the Training tab of the User Action step following the Learned Automation step. See Training for more information.

  • Access the Project Editor to manage form types and reset training if training was done incorrectly.