You must store both the source text and its embedding, then filter by category before the vector search. What is required for efficient execution?
- A A vector index only
- B A vector index plus a range index on category, so the filter narrows candidates (correct answer)
- C Separate containers per category
- D Strong consistency
Shortcut: Filtered vector search benefits from indexing the filter property as well as the vector path, allowing the engine to restrict the candidate set before computing distances.