If your project is configured to use extraction online learning, the method of flagging and training documents has three possibilities.

1. Documents are flagged manually by Validation users

2. Documents are flagged via script

3. Documents are flagged automatically when the Automatic training after Validation option is selected

What happens with the dynamic knowledge base depends if the project is configured to make use of the Dynamic Knowledge Base. These are the possibilities:

1. This use case does not use the training data directly during production to improve extraction results.

   This is because the project is not configured to use the dynamic knowledge base during extraction but the workflow queue contains the Knowledge Base Learning Server.

   In this case, the Knowledge Base Learning Server is creating the dynamic knowledge base even if it is not used. This is because the dynamic knowledge base is created during training so it can be discovered if a document is required.

   The required documents will be added to the New Samples so that they can be imported into Project Builder. The only case where the training information is used is when the project is published after it is trained using the new training information.

2. This use case uses the training data directly during production to enhance extraction.

   The "Use dynamic knowledge base during extraction" option is selected the in Advanced Online Learning Options window and that the workflow queue contains the Knowledge Base Learning Server.

   In this use case, the Knowledge Base Learning Server creates the dynamic knowledge base and collects the documents for the New Samples. The Knowledge Base Learning Server makes use of the dynamic knowledge bases during extraction.