Using machine learning and generative AI in Amazon DataZone - Amazon DataZone

Using machine learning and generative AI in Amazon DataZone

Note

Powered by Amazon Bedrock: AWS implements automated abuse detection. Because the AI recommendations for descriptions functionality in Amazon DataZone is built on Amazon Bedrock, users inherit the controls implemented in Amazon Bedrock to enforce safety, security, and the responsible use of AI.

In the current release of Amazon DataZone, you can use the AI recommendations for descriptions functionality to automate data discovery and cataloging. Support for generative AI and machine learning in Amazon DataZone creates descriptions for assets and columns. You can use these descriptions to add business context for your data and recommend analysis for datasets, which can help boost data discovery results.

Powered by Amazon Bedrock's large language models, the AI recommendations for data asset descriptions in Amazon DataZone help you to ensure that your data is comprehensible and easily discoverable. The AI recommendations also suggest the most pertinent analytical applications for datasets. By reducing manual documentation tasks and advising on appropriate data usage, auto-generated descriptions can help you to enhance the trustworthiness of your data and minimize overlooking valuable data to accelerate informed decision making.

Important

In the current Amazon DataZone release, the AI recommendations for descriptions feature is only supported in the following regions:

  • US East (N. Virginia)

  • US West (Oregon)

  • Europe (Frankfurt)

  • Asia Pacific (Tokyo)

The following procedure describes how to generate AI recommendations for descriptions in Amazon DataZone:

  1. Navigate to the Amazon DataZone data portal URL, and then sign in using single sign-on (SSO) or your AWS credentials. If you’re an Amazon DataZone administrator, navigate to the Amazon DataZone console at https://console.aws.amazon.com/datazone and sign in with the AWS account where the domain was created, and then choose Open data portal.

  2. In the top navigation pane, choose Select project, and then choose the project that contains the asset for which you want to generate AI recommendations for descriptions.

  3. Navigate to the Data tab for the project.

  4. In the left navigation pane, choose Inventory data, and then choose the name of the asset for which you want to generate AI recommendations for descriptions for the asset.

  5. On the asset's details page, in the Business metadata tab, choose Generate descriptions.

  6. Once the descriptions are generated, you can either edit, accept, or reject them. Green icons are displayed next to each automatically generated metadata description for the data asset. In the Business metadata tab, you can choose the green icon next to the automatically generated Summary, and then choose Edit, Accept, or Reject to address the generated description. You can also choose Accept all or Reject all options that are displayed at the top of the page when the Business metadata tab is selected, and thus perform the selected action on all automatically generated descriptions.

    Or you can choose the Schema tab, and then address automatically generated descriptions individually by choosing the green icon for one column description at a time and then choosing Accept or Reject. In the Schema tab, you can also choose Accept all or Reject all and thus perform the selected action on all automatically generated descriptions.

  7. To publish the asset to the catalog with the generated descriptions, choose Publish asset, and then confirm this action by choosing Publish asset again in the Publish asset pop up window.

    Note

    If you don't accept or reject the generated descriptions for an asset, and then you publish this asset, this unreviewed automatically generated metadata is not included in the published data asset.