An embedding is a numerical representation of text in a multidimensional vector. Embeddings allow AIs to measure semantic similarity between texts and are fundamental to RAG search.
What is an Embedding?
An embedding transforms text into a series of numbers (a vector) that captures its meaning. Similar texts have similar embeddings.
How Embeddings Work
- Text is processed by an embedding model
- Model outputs a vector (e.g., 1536 dimensions)
- Vectors are compared by cosine similarity
- Similar content has vectors pointing in similar directions
Role in AEO
Embeddings are used in:
- RAG search: Finding relevant content
- Semantic similarity: Grouping related topics
- Content recommendation: Suggesting related content