AWS Certified Data Engineer - Associate DEA-C01Demo

By Amazon · Question Mode

QuestionQ366

Data Store Management

A company is building a product recommendation system using Amazon OpenSearch Service. The system must perform к-nearest neighbors (k-NN) vector searches across 10 million product embeddings with 768-dimensional vectors. It must preserve high recall accuracy and allow incremental updates without reindexing as new products are added daily. The system must also support complex filtering by product categories and inventory status.

Which vector index type meets these requirements?

  • A FAISS Inverted File Index (IVF) with an nlist value of 1024 and an nprobes value of 10
  • B Lucene Hierarchical Navigable Small Worlds (HNSW) index with an M value of 16 and an efConstruction value of 200
  • C Exact k-NN search that uses a Painless script scoring
  • D Faiss index with binary quantization and an nlist value of 4096
Explanation

A Lucene HNSW index provides scalable approximate k-NN search with high recall, while allowing new documents to be indexed incrementally. OpenSearch supports efficient k-NN filtering with the Lucene engine’s HNSW implementation, enabling category and inventory constraints to be applied during vector search rather than only after retrieval.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!