Skip to main content

Upgrade index to vectorized version

POST 
/q/:qhi/search_update

Build a vectorized version of a given index object asynchronously. Callers can check status via GET /q/{qhi}/update_status.

The index is not committed to the fabric and exists only within the context of this service.

The update involves chunking the metadata and building a single vector embedding per chunk. This can be a length operation depending on the size of the metadata. A vector index will be built against these vectors for efficient semantic retrieval.

Note: The provided content id must be previously crawled.

Note: Vectorization only affects the ranking of the results not the content itself or the API for searching.

WARNING: Vectorization will make search queries significantly slower. If this is undesirable the caller can use the "text" query parameter in the search endpoint to force keyword search.

Request​

Responses​

successful response