Vector similarity

Vector similarity quantifies the semantic or feature-based closeness of two data points (vectors) in a high-dimensional space; it is the core engine for modern AI applications.

This technology is the mathematical process of measuring distance between vector embeddings: numerical representations of complex data like text, images, or audio. Core algorithms include Cosine Similarity (focusing on vector angle) and Euclidean Distance (measuring absolute spatial gap). Vector similarity powers critical AI functions, notably Retrieval-Augmented Generation (RAG) for LLMs, semantic search (finding meaning over keywords), and recommendation systems. For example, it ensures a search for 'budget airfare' returns results for 'cheap flights' by confirming their vectors are highly aligned in the 768-dimension space common to many models.

https://en.wikipedia.org/wiki/Similarity_measure