1Z0-184-25 Question 107
Single answerYou are designing a semantic search solution for product descriptions in an Oracle Database environment running on Oracle Cloud Infrastructure (OCI). Your security requirements dictate that both the generation of vector embeddings and retrieval queries must remain entirely within the database. Which approach best addresses these requirements for generating and storing vector embeddings inside Oracle Database?
- A
Use Oracle Machine Learning� embedded Python functionality within the database to generate embeddings and store them in a vector column.
- B
Export product descriptions to OCI Data Science, generate embeddings externally, then re-import the vectors into a standard VARCHAR2 column.
- C
Enable Oracle Text indexing on the product descriptions and rely on CONTAINS queries for semantic similarity scoring.
- D
Create PL/SQL procedures that call external operating system scripts to generate embeddings and store them in a CLOB column.
Show answer and explanation
Correct answer: A
Explanation
Oracle Machine Learning� embedded Python functionality natively integrates Python scripts within the database engine, enabling in-database data processing. Starting with Oracle Database 23c, a vector data type can be used to store embeddings for advanced similarity searches. By avoiding external data movement, you maintain security and compliance while benefiting from native vector operations. For more information, consult the Oracle Database 23c documentation on vector data types and Oracle Machine Learning� in-database Python capabilities.
- A. Correct.
Correct. Oracle Machine Learning� embedded Python interface allows you to define scripts that run inside the database, preserving data locality and security. You can generate embeddings and store them in a vector column (introduced in Oracle Database 23c) while remaining entirely within the database environment.
- B. Incorrect.
Incorrect. While running embedding generation on OCI Data Science is a valid approach for some workflows, it does not meet the requirement of generating embeddings fully inside the database. Additionally, storing vectors in a VARCHAR2 column is less efficient and does not leverage Oracle� dedicated vector data type.
- C. Incorrect.
Incorrect. Oracle Text is powerful for text indexing and keyword-based search, but it does not natively generate or store vector embeddings. CONTAINS queries do not provide semantic similarity in the same way that vector-based searches do.
- D. Incorrect.
Incorrect. Offloading embedding generation to external OS scripts breaks the requirement of processing strictly within the database. Moreover, storing embeddings in a CLOB column is suboptimal compared to the dedicated vector type for efficient vector operations.