Applications and use cases

Applications and use cases show where HNSW shows up in products: semantic search, recommendations, images, RAG, chunking, reranking, and freshness under business rules.
Created: Updated: 2 min read

These are the topics in this section, each on its own page with a stable path you can bookmark or share.

What topics are covered in this section?

Applications and use cases show where HNSW shows up in products: semantic search, recommendations, images, RAG, chunking, reranking, and freshness under business rules.

What product patterns rely on approximate neighbors?

Semantic search and RAG embed passages or chunks and retrieve context for people or generators. Recommendations use user and item embeddings with candidate generation, ranking, popularity skew, and fresh-item insertion. Image, face, duplicate, and text-to-image retrieval live in visual or joint embedding spaces. Cross-encoders and multi-vector document retrieval add second-stage ranking. Business-rule filtering and update freshness couple ANN to application policy.

HNSW is usually the retrieval engine inside these loops, not the whole product.

How does Weaviate support these patterns?

Weaviate provides vector, keyword, and hybrid search, filtered HNSW, multi-tenancy, and model integrations for embedding and reranking – matching the application glossary’s building blocks. Chunking and passage embedding remain application responsibilities that strongly affect recall. Part IX walks these use cases in curriculum form.

Application terms keep algorithm tuning tied to user outcomes.

Start with semantic search, RAG, candidate generation, and hybrid-related filtered pages, then Part IX chapters for your domain. For infrastructure under the app, see persistence and scaling hubs.

This section connects HNSW to product work. Next, open “What is retrieval-augmented generation (RAG)?” or Part IX’s semantic-search chapter, then chunking and reranking terms when quality issues sit above the index.