Embedding

An embedding is a vector, a fixed-length list of numbers, produced by a model to capture the meaning of an item such as a sentence, product, or picture. Items with similar meaning get vectors that are close together.

Why it matters

Embeddings enable semantic search, recommendations, clustering, and duplicate detection, matching by meaning instead of exact keywords. They are the foundation of RAG systems and are stored in a vector database.

Example

Simplified three-dimensional vectors (real ones have hundreds or thousands of dimensions):

"server down"   -> [0.81, 0.12, 0.55]
"site outage"   -> [0.79, 0.15, 0.52]
"chocolate cake"-> [0.05, 0.92, 0.10]

Cosine similarity between the first two is high, so a search for one can find the other.

Common mistakes

• Mixing vectors from different embedding models in one index; they are not comparable.
• Forgetting to re-embed content after changing the model.
• Assuming embeddings capture exact facts or numbers reliably.

Related

• Vector database
• Retrieval-augmented generation
• Natural language processing
• Neural network

This page is general information written by our editorial team and updated when documentation changes. Servers and providers differ, so test changes safely first. See our editorial policy and disclaimer.