Similarity Search

A technique that retrieves the most relevant document segments by computing the distance or similarity—such as cosine similarity—between a query vector and vectors stored in a knowledge base. It serves as the core mechanism for semantic matching in RAG systems.