What kind of data is a vector database optimized to store and query?
Vector database: stores numeric embedding vectors and finds the nearest ones to a query vector, enabling semantic search and recommendation.
2 / 5
What does approximate nearest neighbor (ANN) search trade off?
ANN: exact nearest-neighbor search is expensive in high dimensions, so ANN algorithms return approximately correct results far faster, trading a bit of recall for speed.
3 / 5
What is the HNSW index?
HNSW: Hierarchical Navigable Small World graphs are a popular ANN index providing high recall and low latency by greedily traversing a layered proximity graph.
4 / 5
Which distance metric is commonly used to compare embeddings?
Distance metrics: cosine similarity (angle between vectors) and Euclidean (L2) distance are standard for embeddings; the right choice depends on how the embedding model was trained.
5 / 5
What is metadata filtering in a vector database?
Metadata filtering: combines vector similarity with structured predicates so you can, for example, search only documents from a given user or category alongside semantic relevance.