search

Documentation

Index

• func CosineSimilarity(X, Y *mat.Dense) (*mat.Dense, error)
• func MaximalMarginalRelevance(queryEmbedding *mat.Dense, embeddingList *mat.Dense, lambdaMult float64, k int) ([]int, error)

Functions

func CosineSimilarity

func CosineSimilarity(X, Y *mat.Dense) (*mat.Dense, error)

CosineSimilarity calculates the cosine similarity between two vectors. The vectors must be of the same length.

func MaximalMarginalRelevance

func MaximalMarginalRelevance(
	queryEmbedding *mat.Dense,
	embeddingList *mat.Dense,
	lambdaMult float64,
	k int,
) ([]int, error)

MaximalMarginalRelevance implements the Maximal Marginal Relevance algorithm. It takes a query embedding, a list of embeddings, a lambda multiplier, and a number of results to return. It returns a list of indices of the embeddings that are most relevant to the query. This is a relatively naive and unoptimized implementation of MMR. :-/ See https://www.cs.cmu.edu/~jgc/publication/The_Use_MMR_Diversity_Based_LTMIR_1998.pdf