Skip to Main Content

JSTOR Seeklight: Generating, reviewing, and editing metadata [Tier 3]

  • Updated

On this page:

Generating metadata

JSTOR Seeklight uses an AI-compatible, Dublin Core based metadata schema. At this time, metadata can only be generated for this schema. See Recommended mappings for the AI-Compatible Dublin Core template for more information.

There are two ways you can generate metadata with JSTOR Seeklight:

Starting with uploaded files

  1. From the JSTOR Stewardship Home page, select the link to Upload media and generate metadata with JSTOR AI.
  2. Choose the files you’d like to upload. 
    • At this time, JSTOR Seeklight supports JPG/JPEG and PDF files.
    • If you have images files (JPG/JPEG) you’d like to group together to create a compound object, select the Group by filename button. See Compound objects for more information.
  3. Optionally, add additional context to help the generation process.
  4. Select Upload. You'll be prompted to create a new project for your generated metadata or add the generated metadata to an existing project.
    • If adding metadata to an existing project, the project must have been created with the AI-compatible metadata schema template.
  5. Select Generate metadata. The system will begin processing your upload in the same tab.
    • Important: Do not close this tab until the upload is finished, or the upload will be stopped and unable to complete.
  6. Once your files have uploaded, you'll be automatically redirected to the Activity Monitor on the project page. The Generating tab will display the status of metadata generation for your uploaded files.
  7. From here, once metadata has been successfully generated, you can review and edit your metadata on the Metadata Records tab.

Starting with a new or existing project

To create a new JSTOR Seeklight-compatible project:

  1. Create a new project.
  2. Select the AI-compatible Dublin Core template as your metadata schema. You must select this template to use JSTOR Seeklight.

To add metadata to your new or existing AI-compatible project:

  1. In your project, open the Add drop-down menu and select Upload & Generate Metadata.
  2. Choose the files you’d like to upload.
    • At this time, JSTOR Seeklight supports JPG/JPEG and PDF files.
    • If you have images files (JPG/JPEG) you’d like to group together to create a compound object, select the Group by filename button. See Compound objects for more information.
  3. Optionally, add additional context to help the generation process.
  4. Select Upload. Your metadata generation job will run in the same tab.
    • Important: Do not close this tab until the upload is finished, or the upload will be stopped and unable to complete.
  5. Once it’s finished, you'll be automatically redirected to the project page where you can review and edit your metadata.
Note: At this time, existing item metadata cannot be re-generated with JSTOR Seeklight.

Providing additional context to help the generation process

When generating metadata with JSTOR Seeklight, there is an optional user input box where you may enter additional context to help inform the metadata generation process. This additional context is a form of prompt augmentation, which means the information you provide is added to the tool’s base prompt to enhance the output it provides.

Here are some things to keep in mind when providing additional context for JSTOR Seeklight:

  • Writing in full sentences with specific details will help the most with metadata output. 
  • You can provide information that is not in the uploaded files, or that you want represented in the metadata, such as the files’ subject matter, topic, or core content. 
  • The tool will perform best with context that is relevant to all files in your upload. For example:
    • These are all paintings by the artist Jenny Saville.
    • All of these files are Master’s theses from Brooklyn College.

Reviewing and editing your metadata

Your generated metadata is contained within a project. On the Metadata Records tab, you're able to view, filter, sort, and mark metadata as reviewed. You can also review and edit metadata at the item level.

  • When you mark an item as reviewed, the "Reviewed" tag will be applied to the item.
  • To mark multiple items in your project as reviewed at once, select one or more items and then select Mark as Reviewed.
  • You can remove the "Reviewed" tag by selecting one or more items Reviewed for any item(s) that have the "Reviewed" tag currently.

When you edit an item’s generated metadata, the "Edited" tag will be automatically applied to the item, and the confidence score for any edited field(s) will display as "Edited." For more information on editing item metadata, see Cataloging items.

Note: Tags are applied at the item level. If you edit one AI-generated metadata field on an item, the item will display the "Edited" tag even if you did not edit any other fields.

Items tags do not display on JSTOR. Tagged items do display the following text at the end of the Item Details panel on JSTOR: The collection contributor used AI to facilitate the creation of some metadata for this item.

Articles in this section