Vector database
Database that stores content (e.g. text chunks) as vector embeddings for similarity search; used by RAG systems to retrieve relevant passages.
Definition
A vector database (or embedding store) stores representations of content as vectors (embeddings) in a high-dimensional space. RAG systems embed queries and document chunks into the same space; retrieval is done by similarity search (e.g. nearest neighbors). Before ingestion, content is typically chunked (e.g. by heading or token window). How you structure and chunk your web content affects whether it is stored as coherent, retrievable units.
Relevance to GEO
Your pages are often scraped, chunked, embedded, and stored in vector databases that power AI answers. Self-contained sections, clear headings, and JSON-LD improve the quality of chunks and the likelihood of accurate retrieval and citation.