Documents tab in Document Splitter skills

Manage documents in a Document Splitter skill — upload, organize into sets, label, and configure documents for training and testing.

The Documents tab lists every document a Document Splitter skill uses for setup, training, and testing. From here you can upload files, organize them into sets, label them, and run tests against specific subsets.

Document sets

Documents are organized into document sets. Every skill has an All Documents set (created automatically) that contains every uploaded document. You can also create custom sets — each is a subset of All Documents, and a single document can belong to multiple custom sets.

When you create a new activity, it uses All Documents by default. Custom sets are useful when one skill applies different rules to different document variants — see the Classify By Text and Image activity.

The Documents tab title reflects the current selection: it reads All Documents for the default set, or Documents: <Set name> for a custom set.

Create a custom document set

1. Open the Documents tab

   On the skill, switch to the Documents tab.

2. Add a new set

   Click the add button in the list of document sets.

3. Name the set

   Provide a name for the custom set.

4. Add documents

   Add the documents you want the set to contain. The new set appears alongside All Documents and can be edited at any time.

Test sets

You can also create a custom set strictly for testing — for example, a balanced set of documents from a specific client. A test set isn’t linked to any activity.

Create a test set

1. Add a new set

   On the Documents tab, click the add button in the list of document sets. If the documents you want are already uploaded, select them and choose Create Set from Selection on the shortcut menu.

2. Name the test set

   Provide a name that identifies the set’s purpose.

3. Run the test

   Select the set, and then choose Test Skill Using This Set on the shortcut menu.

To test against an ad-hoc selection instead, choose Test Skill on the shortcut menu of one or more selected documents.

When editing a skill that was created in Vantage, that skill’s document set also appears here.

Document processing

You can upload documents at any point. Any new upload is processed immediately by the current flow on the Activities tab. After processing, the predicted labeling appears on the Fields tab, and reference labeling is generated automatically from it.

Reference labeling generated this way is unconfirmed until you copy predicted labeling to reference (shortcut menu on the document). Documents with unconfirmed labeling cannot be used for training and are excluded from test results. To confirm labeling, review the document on the Fields tab.

View options

• List view — shows documents as a list with upload date, the sets each document is in, and a Reference Fields column showing whether the document is labeled.
• Thumbnail view — shows documents as thumbnails. Use the preview button for the full image; rotate from the more menu.

Toolbar commands

Command	What it does
Upload	Upload documents to the selected set. Each upload is also added to All Documents. You can also drag files in from File Explorer.
Import Labeled Documents From Folder (in the Upload drop-down)	Upload pre-labeled documents to All Documents. The source folder can come from a skill’s training set, manually reviewed transactions, or ABBYY FlexiCapture. For details, see Importing labeled documents from a folder.
Divide Set	Split one set into two by ratio — for example, to create separate training and testing collections.
Add from Set	Add to the current set every document from another set that isn’t already in the current set.
Add to Set	Add the selected documents to a different set. Disabled when the skill has only one set.
Delete	Delete the selected documents from the current set, or from every set. From within All Documents, deletion always removes from every set.
Rotate	Change the orientation of the selected pages: rotate 90° left, 90° right, or 180°.

A set with fewer than 10 documents cannot be divided.

Limits

Don’t upload more than 10,000 documents to a single skill — beyond that, the skill may become unstable.

Related topics

• Document Splitter activities
• Label documents in Advanced Designer
• Document Splitter skill settings in Advanced Designer