Embedding

An embedding is a representation of text (or image) as a long vector of numbers, arranged so that content with similar meaning has similar vectors. "Dog" and "puppy" land close together, "dog" and "radiator" far apart - regardless of whether they share any letters. This is the foundation of semantic search: instead of searching by keywords, you compare meanings. Embeddings underpin RAG, recommendation systems, duplicate detection, and content clustering.