Vector Database

A vector database stores data as high-dimensional numerical vectors (embeddings) generated by machine learning models, and supports fast similarity search across these vectors. Rather than exact keyword matching, vector databases find records that are semantically similar - 'conceptually close' in the embedding space.

Vector databases are the infrastructure layer behind AI-powered search and recommendation systems. Searching a property dataset with 'spacious open-plan apartment near park' returns listings where the semantic content of the description matches the query, even without an exact keyword match. Pinecone, Weaviate, and pgvector (PostgreSQL extension) are popular vector database options.

The growth of large language models has made vector databases a standard component of AI application architectures. Retrieval-Augmented Generation (RAG) systems use vector databases to find relevant context from a knowledge base before generating a response. For data products, combining vector search over dataset records with structured filters (price, location, type) creates powerful hybrid search experiences.