DBMS_VECTOR

The DBMS_VECTOR package simplifies common operations with Oracle AI Vector Search, such as extracting chunks or embeddings from user data, generating text for a given prompt or an image, creating a vector index, or reporting on index accuracy.

This table lists the DBMS_VECTOR subprograms and briefly describes them.

Table 12-1 DBMS_VECTOR Package Subprograms

Subprogram	Description
ONNX Model Related Procedures: These procedures enable you to load an ONNX model into Oracle AI Database and drop the ONNX model.
LOAD_ONNX_MODEL	Loads an ONNX model into the database
LOAD_ONNX_MODEL_CLOUD	Loads an ONNX model from object storage into the database
DROP_ONNX_MODEL Procedure	Drops the ONNX model
INMEMORY_ONNX_MODEL	Controls the in-memory sharing of initializers for an ONNX model with external initializers within the database
Chainable Utility (UTL) Functions: These functions are a set of modular and flexible functions within vector utility PL/SQL packages. You can chain these together to automate end-to-end data transformation and similarity search operations.
UTL_TO_EMBEDDING and UTL_TO_EMBEDDINGS	Converts text or an image to one or more vector embeddings
UTL_TO_GENERATE_TEXT	Generates text for a prompt (input string) or an image
Credential Helper Procedures: These procedures enable you to securely manage authentication credentials in the database. You require these credentials to enable access to third-party service providers for making REST calls.
CREATE_CREDENTIAL	Creates a credential name
DROP_CREDENTIAL	Drops an existing credential name
Data Access Functions: These functions enable you to retrieve data, create index, and perform simple similarity search operations.
CREATE_INDEX	Creates a vector index
REBUILD_INDEX	Rebuilds a vector index
GET_INDEX_STATUS	Describes the status of a vector index creation
ENABLE_CHECKPOINT	Enables the Checkpoint feature for a vector index user and index name
DISABLE_CHECKPOINT	Disables the Checkpoint feature for a vector index user and index name
INDEX_VECTOR_MEMORY_ADVISOR	Determines the vector memory size that is needed for a vector index
QUERY	Performs a similarity search query
UTL_TO_RERANK	Reorders search results for a more relevant output
Accuracy Reporting Function: These functions enable you to determine the accuracy of existing search indexes and to capture accuracy values achieved by approximate searches performed by past workloads.
INDEX_ACCURACY_QUERY	Verifies the accuracy of a vector index
INDEX_ACCURACY_REPORT	Captures accuracy values achieved by approximate searches

Note:

DBMS_VECTOR is a lightweight package that does not support text processing or summarization operations. Therefore, the UTL_TO_TEXT and UTL_TO_SUMMARY chainable utility functions and all the chunker helper procedures are available only in the advanced DBMS_VECTOR_CHAIN package.