You can create a document class subscription for vector indexing. After you create the
subscription, a vector index is automatically created for any new documents that are added to the
object store, which belong to the subscribed document class. However, you need to run a custom sweep
job to create a vector index for existing documents that were added to the object store before the
subscription was created.
Before you begin
For all existing document content that you want to be vector indexed, make sure that you
have the following permissions:
About this task
Configure and run a custom job sweep for vector indexing to add a vector index for existing
document content.Note: For better throughput and performance, do not run the summary generation
custom job sweep and the vector indexing custom job sweep at the same time.
Procedure
- Start the New Custom Job wizard in the Administration Console for
Content Platform Engine (ACCE) as an object
store administrator:
- In the domain navigation pane, select the object store.
- In the object store navigation pane, select the folder.
- Click New.
You can configure the new custom job
sweep to configure vector indexing for existing documents.
- Enter the name and description for the custom job sweep for vector
indexing.
You can name the sweep as Custom AI Indexing
Job.
- Specify the Custom Sweep subclass as Custom Sweep
Job.
- Specify the Sweep Mode as
Normal.
- Select the Enabled property to enable to custom sweep to run.
Click Next.
- Specify the Target Class for which you enabled persistent storage
of extracted text.
- Specify the Filter Expression to filter objects of the selected
target class type.
For example, you can include a filter
VersionStatus=1 for document classes that are
Released.
- Optional: Select Enabled if you want to include
subclasses and if you want to record failures.
- Specify the Sweep Action as Gen AI vector Indexing
Request Job Sweep Action.
- Optional: Specify the start and end dates for the custom job
sweep.
- Complete the wizard.