The Schema Mapping page displays all attributes from your input datasets, and allows you to map them to the attributes in the unified schema.

Creating a Unified Dataset

To create a unified dataset:

1. Select the Schema Mapping page. The left-hand side of the page displays all of the source attributes from each uploaded dataset. The right-hand side allows you to create a unified dataset. Tamr uses the following default name for the unified dataset: [project name]_unified_dataset.
2. Optionally, edit the unified dataset's name in the text box or leave the default.
3. Choose Create Unified Dataset. The screen changes to show that there are no unified attributes. You can begin adding them and building your unified schema.

Naming Unified Attributes

Unified attributes must have names that conform to the following requirements:

• Names cannot contain the . (period) character
• Names must be unique: they cannot match (case-insensitively) any other unified attribute in your project
• Names cannot be empty, and cannot contain a leading or trailing whitespace
• Names cannot match (case-insensitively) the following Tamr-generated attribute names:
• tamr_id
• origin_source_name
• origin_entity_id
• clusterId
• originSourceId
• originEntityId
• sourceId
• entityId

As a best practice, use a consistent style for your attribute names. For example, always use "camelCase" or "snake_case".

Creating a Unified Attribute

To create a unified attribute:

1. Select Create a new unified attribute on the right-hand side, or
2. Select one or more unmapped attributes on the left side and choose Bootstrap. Tamr launches the attribute bootstrapping process. Bootstrapping performs these steps:

• Groups together input attributes with the same name.
• Assigns input attributes to unified attributes.
• Uses the same name for a unified attribute as the name of the input attribute. You can change the default name.

Mapping Attributes Manually

After you create a unified attribute you can map additional input attributes to by dragging and dropping them.

Tip: Use the Filter icon to show only Unmapped input attributes.

If there is an input attribute that you choose not to map, hover over the Unmapped label, and select Do not map this attribute. You can also select this option from the Map drop-down menu.

You can map multiple input attributes using the Map drop-down menu.

After you map a few attributes manually, you can use Tamr machine learning to map additional attributes for you.

Editing Unified Attributes

If an input attribute is incorrectly mapped to a unified attribute, hover over the list of mapped input attributes, and select the attribute from the list.

You can also unmap input dataset attributes from the left-hand panel. Hovering over the name of the unified attribute reveals the option to toggle mapping for the attribute.

Changing the Unified Attribute Type

Another important aspect of the unified derived attribute is deciding whether Tamr will treat it as a number or a string type value. The attribute's type affects the approaches Tamr uses for machine learning.

To toggle the attribute type, choose the More icon (three dots) to the right of the unified attribute.

Deleting a Unified Attribute

To delete a unified attribute, select it and choose Remove.

Updating a Unified Dataset

To update a unified dataset:

1. Navigate to the Schema Mapping tab of any project.
2. Choose Update Unified Dataset.