Example of additional functions

main.go

// Opaque type that models an index number used to identify a word.
type ItemIndex int

func (*ItemIndex i) IsPresent() {
  return i >= 0
}

// Opque type that models a fixed-length vector.
type Vector {
  vector []float32
}

// Map of structs with vectors and weights
type VectorWeightMap []struct {
	weight  float32
	vectors Vector
}

// API to decouple the K-nn interface that is needed for Weaviate from a concrete implementation.
type VectorIndex interface {
  // Return the number of items that is stored in the0
  GetNumberOfItems() (int, err)

  // Returns the length of the used vectors.
  GetVectorLength() (int, err)

  // Look up a word, return an index.
  // Check for presence of the index with index.IsPresent()
  WordToItemIndex(word string) (ItemIndex, err)

  // Based on an index, return the assosiated word.
  ItemIndexToWord(ItemIndex) (string, err)

  // Get the vector of an item index.
  GetVectorForItemIndex(item ItemIndex) (Vector, err)

  // Compute the distance between two items.
  GetDistance(a ItemIndex, b ItemIndex) (Vector, err)

  // Get the n nearest neighbours of item, examining k trees.
  // Returns an array of indices, and of distances between item and the n-nearest neighbors.
  GetNnsByItem(item ItemIndex, n int, k int) ([]ItemIndex, []float32, err)

  // Get the n nearest neighbours of item, examining k trees.
  // Returns an array of indices, and of distances between item and the n-nearest neighbors.
  GetNnsByVector(vector Vector, n int, k int) ([]ItemIndex, []float32, err),
  
  // Get the centroid based on provided vectors and the weight.
  // The a argument is used to choose a centroid calculation algorithm.
  // https://en.wikipedia.org/wiki/Centroid#Locating_the_centroid
  GetCentroidByVectorWeights(vectorWeightMap VectorWeightMap, a string) (Vector, err)
  
}