1Z0-184-25 Question 68
Select 2Your company is building a product recommendation engine using text embeddings in Oracle Autonomous Database with vector search capabilities. You generate product description embeddings using Oracle AI Services for language processing, store them in a vector column, and create an index for similarity searches. However, the query results are returning irrelevant items. Which two actions should you take to improve the accuracy and performance of your similarity search?
- A
A. Ensure that the vector column� dimension matches the dimension of the embeddings, and recreate the index to reflect this dimension.
- B
B. Increase the column size to hold more text data rather than focusing on the vector column definitions.
- C
C. Use approximate nearest neighbor (ANN) indexing and validate that the distance function aligns with your embedding model (e.g., cosine similarity).
- D
D. Disable indexing on the vector column and rely only on a standard B-tree index for faster lookups.
Show answer and explanation
Correct answers: A, C
Explanation
To perform effective similarity searches in Oracle Autonomous Database, you must align your vector column definitions with the model� embedding dimensions and use an indexing strategy optimized for vector queries, such as approximate nearest neighbor (ANN) indexing. Ensuring the correct distance function (e.g., cosine) matches your embedding generation process is also key. For more details on configuring and optimizing vector search, refer to the Oracle Database 23c documentation on vector data types and indexing best practices.
- A. Correct.
A. Correct. If the dimension of the stored vectors does not match the model� embedding dimension, similarity calculations become inaccurate. After correcting the dimension, you should recreate the index so that the engine properly accounts for the new dimension and can deliver relevant results.
- B. Incorrect.
B. Incorrect. Increasing the column size for text data does not directly impact vector similarity search. Vector search accuracy and performance hinge on using correct vector dimensions, indexing strategies, and distance functions rather than the raw text storage parameters.
- C. Correct.
C. Correct. Approximate nearest neighbor (ANN) indexing is often used to improve query performance when dealing with high-dimensional vector data. Ensuring that the chosen distance function (e.g., cosine or Euclidean) matches how you generated embeddings is crucial for accurate similarity search results.
- D. Incorrect.
D. Incorrect. Disabling the vector index and relying on a B-tree index for vector data will significantly degrade similarity search accuracy and performance, as B-tree indexes are not designed to handle nearest neighbor queries for high-dimensional vectors.