Embedding (Vector Embedding)
A numerical representation of text, images, or other data as a vector. Similar items have similar vectors — the foundation of semantic search and RAG.
An embedding is a vector (list of numbers, typically 384-3072 dimensions) that represents the semantic meaning of text, images, audio, or other data. Modern embedding models map similar items to similar vectors, even when the surface forms differ.
For example, "dog" and "puppy" produce nearby vectors, even though no characters match. "Sunny weather" and "warm bright day" produce nearby vectors despite different words. This semantic similarity enables search that goes beyond keyword matching.
Embeddings are produced by models like OpenAI's text-embedding-3, Cohere Embed, or open-source alternatives (BGE, Nomic). They're stored in vector databases and queried via similarity search (cosine similarity, dot product). Embeddings are the backbone of semantic search, RAG systems, recommendation engines, and AI clustering.