Curate and enrich asset metadata

When assets are ingested into the catalog from AWS Glue Data Catalog, they come with technical metadata like column names, data types, and S3 locations. This information is useful for engineers who already know what they're looking for, but it doesn't help someone who is trying to understand what a table represents, whether it's trustworthy, or how it should be used. Metadata curation is the process of adding that business context so that tables become understandable and discoverable by everyone in the organization. You do this from the Browse page in the admin portal. You can add this context manually, or use AI to automatically generate business names, descriptions, and glossary term associations for tables and their columns. See Use AI to generate metadata suggestions for details.

All metadata curation happens on the asset detail page. To get there, choose Browse in the left navigation pane, then choose the table you want to curate. The asset detail page shows the table's business name and description at the top (both editable), along with a Generate suggestions button for AI-generated metadata. The Overview tab is where you manage the summary, glossary terms, README documentation, and metadata forms. The Schema tab is where you curate individual columns, including their business names, descriptions, glossary terms, metadata forms, and README documentation.

As you make changes, the asset shows a Changes pending badge next to the table name. These changes are saved as a new draft revision but are not discoverable to anyone until you re-publish the asset. See Publish data for details.

Edit the business name

Each table has a technical name that comes from the AWS Glue Data Catalog (for example, policies) and a business name that you can set to something more descriptive (for example, "Insurance Policy Management"). The business name appears in search results, making it easier for people to identify the table without knowing the technical name. You can also use AI to generate business names for the table and all of its columns at once.

To edit the business name

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Next to Business name, choose the edit icon (pencil).

5. Enter the new business name in the text field.

6. Choose Save.

Edit the description

The description is a short text that explains what the table contains and what it is used for. It appears above the tabs on the asset detail page and is separate from the longer AI-generated summary on the Overview tab. You can also use AI to generate descriptions for the table and all of its columns at once.

To edit the description

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Below the business name, choose the edit icon (pencil) next to the description text.

5. Enter or update the description text.

6. Choose Save.

Add or edit the README

The README is a longer-form documentation section where you can provide detailed context about a table using Markdown syntax. This is useful for documenting things like refresh schedules, known data quality issues, usage guidelines, or anything else that helps someone understand and use the data correctly. The README supports up to 4,096 characters and includes a side-by-side preview so you can see how the formatted content will look.

To add or edit the README

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. On the Overview tab, find the README section.

5. Choose Edit.

6. Enter your documentation using Markdown syntax.

7. (Optional) Use the side-by-side preview to see how the formatted content will look.

8. Choose Save.

Associate glossary terms with a table

Glossary terms help people understand what a table represents by linking it to standardized business vocabulary. For example, tagging a table with "GDPR" tells someone that the table contains data subject to GDPR compliance requirements, without them having to read through the schema or documentation. Terms must be created in a glossary before they can be associated with a table. See Create and manage business glossaries for how to create glossaries and terms. You can also use AI to suggest glossary term associations based on the table's technical metadata.

To associate glossary terms with a table

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. On the Overview tab, in the Glossary Terms section, choose the edit icon. The Edit Glossary terms panel opens, showing any terms already associated with the asset.

5. Choose Add terms.

6. Search for the term you want to associate, or browse the available terms. You can filter by glossary using the dropdown.

7. Select the checkbox next to one or more terms.

8. Choose Add terms to confirm your selection.

9. Review the updated list of associated terms. To remove a term, choose the remove icon (trash) in the Action column.

Attach a metadata form to a table

Metadata forms let you capture structured information about a table using predefined fields. For example, a "Data Retention Policy" form might have fields for retention period, next review date, and whether the data contains PII. Forms must be created and enabled before they can be attached to a table. See Standardize metadata using metadata forms for how to create forms.

To attach a metadata form to a table

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. On the Overview tab, in the Metadata Forms section, choose Add Form. The Select a form panel opens.

5. Browse or search for the form you want to attach. Each form shows its name, description, and number of fields.

6. Select the radio button next to the form you want to attach.

7. Choose Next.

8. Fill in the form fields. Required fields are marked accordingly.

9. Choose Save.

The form appears in the Metadata Forms section on the Overview tab. You can expand or collapse individual forms, and use Collapse all to minimize all forms at once.

To manage how forms are displayed to users who find this asset in the catalog, choose the edit icon (pencil) next to the Metadata Forms section header. This opens the Manage metadata forms panel, where you can do the following:

• Reorder forms – Drag forms using the grip handle to change the order in which forms appear on the asset detail page. For example, if you want the "Data Ownership" form to appear before the " table" form, drag it above. Users who discover this asset in the catalog will see forms in the order you set here.

• Show expanded – Controls whether a form is expanded or collapsed when users view the asset detail page. When toggled on, users immediately see all the form's field names and values without choosing anything. When toggled off, users see only the form name as a collapsed header and must choose to expand it. Use this to highlight the most important forms while keeping less critical ones out of the way.

• Remove a form – Choose the delete icon (trash) to detach a form from the asset. This removes the form and all its field values from the asset. The form itself is not deleted and can be re-attached later.

Edit metadata form values

After a form is attached to a table, you can update its field values at any time.

To edit metadata form values

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. On the Overview tab, find the form you want to edit in the Metadata Forms section.

5. Choose the edit icon (pencil) next to the form name.

6. Update the field values.

7. Choose Save.

Curate column metadata

The Schema tab on the asset detail page shows all columns in the table. Each column row has inline edit controls for the business name, description, glossary terms, and metadata forms. The Actions dropdown on each row provides access to additional options including viewing column details and editing the column's README. This lets you add business context at the column level, which is especially useful for tables with cryptic column names or columns that need specific definitions.

Schema tab columns

Column	What it shows
Name	The column's business name (editable) and technical name. The business name is what people see in the catalog. The technical name comes from the Glue Data Catalog.
Data Type	The column's data type from the Glue Data Catalog (for example, string, int, double).
Description	A text description of what the column contains (editable).
Glossary Terms	Business glossary terms associated with this column (editable).
Metadata Forms	Metadata forms attached to this column.
Actions	Opens a dropdown with View Details (to see all column metadata in a panel) and Edit Readme (to add or edit the column's README documentation).

Edit a column's business name

To edit a column's business name

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Choose the Schema tab.

5. In the Name column, find the column you want to rename and choose the edit icon (pencil).

6. Enter the new business name in the text field. The technical name remains unchanged below it.

7. Choose the checkmark icon to save, or choose the cancel icon to discard.

To generate business names for all columns at once, see Use AI to generate metadata suggestions.

Edit a column's description

To edit a column's description

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Choose the Schema tab.

5. In the Description column, find the column you want to edit and choose the edit icon (pencil).

6. Enter or update the description.

7. Choose the checkmark icon to save, or choose the cancel icon to discard.

Associate glossary terms with a column

To associate glossary terms with a column

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Choose the Schema tab.

5. In the Glossary Terms column, find the column you want to edit and choose the glossary terms cell or the edit icon (pencil). The Edit Glossary terms panel opens.

6. Choose Add terms.

7. Search for the term you want to associate, or browse the available terms.

8. Select the checkbox next to one or more terms.

9. Choose Add terms to confirm your selection.

Attach a metadata form to a column

To attach a metadata form to a column

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Choose the Schema tab.

5. In the Metadata Forms column, find the column you want and choose Add metadata form. The Select a form panel opens.

6. Browse or search for the form you want to attach.

7. Select the checkbox next to the form.

8. Choose Next.

9. Fill in the form fields.

10. Choose Save.

Column actions

Each column row on the Schema tab has an Actions dropdown in the last column. The dropdown provides two options.

View Details opens a side panel showing all metadata for the column, including its business name, technical name, description, glossary terms, metadata forms, and README.

Edit Readme opens a modal where you can write or update the column's README documentation using Markdown. The editor includes a side-by-side preview, character count (0/4096), and toolbar with undo, redo, and copy controls.

To edit a column's README

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to edit.

4. Choose the Schema tab.

5. In the Actions column for the column you want to edit, choose the dropdown, then choose Edit Readme.

6. Enter or update the README content using Markdown formatting.

7. Choose Save Readme.

Use AI to generate metadata suggestions

Instead of manually writing business names and descriptions for every table and column, you can use AI to generate suggestions based on the table's technical metadata. This is especially useful when you have many tables to curate or when column names are abbreviated or cryptic.

AI can generate the following types of metadata.

• Business name for the table and each of its columns. For example, a column named policy_id might get the suggested business name "Policy ID," and a column named effective_date might become "Effective Date."

• Summary and column descriptions for the table and each of its columns. For columns, these are plain-language explanations of what the column contains. For the table itself, the generated description becomes the Summary on the Overview tab, a longer overview describing what the table contains and how its columns relate to each other. The summary can only be created through AI generation, but you can edit it afterward.

• Glossary terms for the table and its columns. The system suggests associations between columns and existing glossary terms based on the column's name and description. For example, a column named category might get a suggested association with the "Technology" glossary term.

Generate suggestions

To generate AI suggestions

1. Navigate to Amazon SageMaker Unified Studio and choose Domain management to open the admin portal.

2. In the left navigation pane under Catalog, choose Browse.

3. Choose the asset you want to curate.

4. Choose Generate suggestions in the top-right area of the page.

5. In the Generate suggestions dialog, select the content types you want to generate. The options are Business name, Summary and column descriptions, and Glossary terms. All three are selected by default. Deselect any types you do not want to generate.

6. Choose Generate.

The system analyzes the table's technical metadata and generates suggestions. This may take a minute or more depending on the number of columns.

Review and accept suggestions

When generation completes, a banner appears on the asset detail page showing the total number of suggestions. The banner has Reject all and Accept all buttons for bulk actions.

You can also review suggestions individually. On the Schema tab, suggested glossary terms appear inline in the Glossary Terms column with a sparkle icon to distinguish them from existing terms. Each suggestion has its own reject and accept buttons, so you can accept some suggestions and reject others for the same column.

Business names and descriptions that are generated are applied directly as pending changes. You can edit them using the same inline edit controls described in the earlier sections before re-publishing.

Tip

AI-generated suggestions are a starting point. Review them carefully and adjust the language to match your organization's terminology and standards before publishing.

Important

If you publish an asset without accepting or rejecting its AI-generated suggestions, the unreviewed metadata is not included in the published version. Always review and accept suggestions before publishing to ensure the generated business names, descriptions, and glossary terms are visible to other users.

Quotas and region availability

AI-generated metadata is powered by Amazon Bedrock. The following monthly quotas apply per domain:

• Business name generation: 50,000 invocations/month

• Business description and summary generation: 10,000 invocations/month

• Glossary term generation: 10,000 invocations/month

Contact AWS Support to request a quota increase.