1Z0-184-25 Question 39
Select 2Your media streaming platform runs on Oracle Cloud Infrastructure and stores metadata for millions of movies in an Oracle Autonomous Database, including high-dimensional embeddings representing each movie� content. You want to find the top 10 most similar movies to a user� preference vector quickly. After deciding to use a vector index on the embeddings column for approximate nearest neighbor (ANN) queries, which two actions must you take to ensure your queries run efficiently?
- A
Create the embeddings column using a supported vector data type that defines the dimension for all rows.
- B
Create a vector index on the embeddings column with 'INDEXTYPE IS VECTOR' and specify the dimension in your CREATE INDEX statement.
- C
Convert the embeddings column to a CLOB data type and rely on an Oracle Text index for semantic similarity searches.
- D
Exclude rows with null or incomplete embeddings by preprocessing them in a staging table before loading into your main table.
- E
Use specialized approximate nearest neighbor query functions or operators that reference your vector index when retrieving the top similar results.
Show answer and explanation
Correct answers: B, E
Explanation
To efficiently run approximate nearest neighbor searches on high-dimensional embeddings in Oracle Autonomous Database, you need both a properly created vector index and specialized query operators or functions referencing that index. Consult Oracle documentation for your specific database version and review best practices for storing embeddings (such as using consistent dimensions and supported data types) and for invoking the vector index during query execution.
- A. Incorrect.
Option 1: While storing embeddings in a supported vector data type is best practice (particularly in newer Oracle Database versions that allow vector-specific columns), this alone does not guarantee efficient ANN queries. You could also store embeddings in certain numeric formats or arrays. The key requirement is that dimensions remain consistent, but the question specifically asks about the two actions that ensure efficient ANN queries, which involve indexing and correct query usage.
- B. Correct.
Option 2: Correct. Creating a vector index with 'INDEXTYPE IS VECTOR' (or the appropriate syntax in your environment) is essential for enabling faster approximate nearest neighbor searches. Specifying the correct dimension parameter ensures the index functions properly with your embeddings.
- C. Incorrect.
Option 3: Incorrect. Oracle Text indexes are suited for textual data, not high-dimensional numeric vectors. Using a text index for numeric embeddings would not accelerate the kind of similarity queries needed for this scenario.
- D. Incorrect.
Option 4: Incorrect. Filtering null embeddings is a good data hygiene step, but it is not strictly required to achieve efficient ANN queries. The index creation process can typically ignore or handle null values; this step alone does not address the main requirement for vector-based querying.
- E. Correct.
Option 5: Correct. Even with a properly created vector index, you must use specialized approximate nearest neighbor search functions or operators in your SQL queries. These functions leverage the index and allow you to retrieve results based on vector similarity.