Embeddings
Also known as: vector embeddings, text embeddings
Embeddings are numerical vector representations of text, images, audio, or other data — produced by ML models so that semantically similar items end up close together in vector space, enabling search, clustering, and retrieval.
Detailed explanation
An embedding is typically a fixed-length vector of floating-point numbers (e.g., 384, 768, or 1536 dimensions) generated by a model trained to map similar inputs to nearby points. Once data is embedded, downstream tasks can use distance metrics (cosine, dot product) to find similar items.
Embeddings power semantic search, recommendation, retrieval-augmented generation, deduplication, classification, and clustering. The choice of embedding model matters a lot for quality: domain-specific models often outperform general-purpose ones on specialized data.
Operationally, embeddings need a strategy for storage (vector database or vector-capable store), refresh (when the underlying model or data changes), and versioning (consumers must know which embedding version produced which vectors).