Also known as: vector search · neural search · hybrid search
Semantic search retrieves results based on meaning rather than keyword matching, using embeddings to find documents conceptually related to a query even when they share no literal terms.
Its advantage over lexical search is handling paraphrase, synonyms, and conceptual queries. Its weakness is precision on exact identifiers — product codes, error strings, CVE numbers — where literal matching is exactly what the user wants.
Hybrid search, combining lexical scoring such as BM25 with vector similarity and then re-ranking the merged results, consistently outperforms either approach alone and has become the default architecture for serious retrieval systems.
Embedding
An embedding is a numerical vector representation of text, images, or other data, positioned in a high-dimensional space such that semantically similar items sit close together.
Vector Database
A vector database is a data store optimised for indexing and querying high-dimensional embedding vectors, typically using approximate nearest neighbour algorithms to find the most semantically similar items quickly.
Retrieval-Augmented Generation
Retrieval-augmented generation (RAG) is a technique in which relevant documents are retrieved from an external knowledge source and supplied to a language model as context, so that its response is grounded in specific retrieved material rather than parametric memory alone.