Iterations
An iteration is a part of the Predictive Coding process. Potentially relevant documents are found through training and grouped in an iteration set.
Iterations are created manually or automatically for a specific workflow. When an iteration is created, the system trains the workflow’s value.
Whereas training covers all documents of a matter, the different document sets created for iterations are limited to the universe documents. This means that all sets are filtered by the current universe.
For each iteration, there is exactly one training. When training for an iteration runs, another iteration for the same workflow cannot be created. As a result of training, documents are categorized as to whether they are relevant for the review workflow or not.
Iteration processing is based on:
- Iteration input
- The input consists of all universe documents that are tagged with the value to be trained. The documents are used as reference for training.
- Iteration output
- The output consists of:
- all documents that are newly suggested for the field value after training and
- all documents that are already tagged with the value, but are not tagged to the target review state valid for them, i.e. documents that are potentially relevant for the workflow.
- Iteration set
- The iteration set is a document set based on the iteration output. These documents can be batched for manual review. The iteration set is represented by an Iteration line on the Review Workflows page.