A search approach that retrieves results based on meaning rather than keyword overlap — typically uses vector embeddings to find documents semantically similar to a query.
Semantic Search converts queries and documents into dense vector embeddings (usually from a transformer model) and retrieves documents whose embeddings are closest to the query embedding via cosine similarity or dot product. It can find conceptually related documents that share no keywords — e.g., 'car maintenance tips' matches 'how to keep your vehicle running.' Powers most modern RAG systems, semantic enterprise search, and recommendation systems.
Building a semantic search over internal Confluence docs so 'how do I configure SSO?' surfaces docs about 'enabling single sign-on' even though no keywords overlap.
Semantic Search is the foundation of modern knowledge retrieval — keyword search fails when users don't know the exact terms documents use; semantic search bridges that gap.
Need help implementing this in your business?
Get Started