Vector embeddings are numerical representations of data—such as words, images, or users—mapped into a continuous vector space. In this space, similar items are positioned closer together, enabling machines to understand relationships and patterns in the data.

Each item is represented as a dense vector of real numbers (e.g., 300-dimensional for text), learned from data using machine learning techniques. For example, in natural language processing, embeddings capture semantic meaning: the vectors for “king” and “queen” will be close, and the relationship between “man” and “woman” is mirrored in that between “king” and “queen”.

Embeddings are foundational in tasks like recommendation systems, search, and language modeling, because they translate complex inputs into structured, comparable formats.