Vector Search allows for the discovery of approximations based on the relative distance between embedded vector data, which is usually the result of an NLP model (Bag of Words, Word2Vec, GloVe, etc). The approximate distance between vector embeddings is often computed with a cosine-based algorithm.
Vector Search allows for the discovery of approximations based on the relative distance between data that has been "vectorized" or written as arrays of numbers. This vectorized data is known as a "vector embedding" and is usually the result of a natural language processing (NLP) model such as Bag of Words, Word2Vec, or GloVe. The approximate distance between vector embeddings is often computed with a cosine-based algorithm.
Common Vector Search algorithms:
- Approximate Nearest Neighbor (ANN)
- Dot Product
- K Nearest Neighbor (KNN)
Databases with Vector Search capability:
