edu.ucla.sspace.similarity

Interface SimilarityFunction

All Known Implementing Classes:

AbstractSymmetricSimilarityFunction, AverageCommonFeatureRank, CosineSimilarity, DotProduct, EuclideanSimilarity, GaussianKernel, JaccardIndex, KendallsTau, KLDivergence, LinSimilarity, OneSimilarity, PearsonCorrelation, PolynomialKernel, SpearmanRankCorrelation, TanimotoCoefficient

public interface SimilarityFunction

An Interface for any similarity metric between two Vectors. Methods should provide efficient support for sparse vectors if an implementation is feasible as these metrics are heavily used in some algorithms. Metrics should note via isSymmetric() whether or not they are symmetic, whereby method.sim(A,B) == method.sim(B,A) always holds true for any A,B. If the metric takes in any parameters, such as weights, they can use the setParams(double...) method, which can take in any number of double arguments. Methods should also provide a no argument constructor that has reasonable default values, or loads values from system properties.

Author:

Keith Stevens

Method Summary

Methods

Modifier and Type	Method and Description
boolean	isSymmetric() Returns true if sim(A,B) == sim(B,A) is true for any A, B.
void	setParams(double... arguments) Sets an double parameters, such as weights, for this SimilarityFunction.
double	sim(DoubleVector v1, DoubleVector v2) Returns the similarity between v1 and v2.
double	sim(IntegerVector v1, IntegerVector v2) Returns the similarity between v1 and v2.
double	sim(Vector v1, Vector v2) Returns the similarity between v1 and v2.

Method Detail

setParams

void setParams(double... arguments)

Sets an double parameters, such as weights, for this SimilarityFunction.

isSymmetric

boolean isSymmetric()

Returns true if sim(A,B) == sim(B,A) is true for any A, B.

sim

double sim(DoubleVector v1,
         DoubleVector v2)

Returns the similarity between v1 and v2. If isSymmetric() is false, the ordering does matter.

sim

double sim(IntegerVector v1,
         IntegerVector v2)

Returns the similarity between v1 and v2. If isSymmetric() is false, the ordering does matter.

sim

double sim(Vector v1,
         Vector v2)

Returns the similarity between v1 and v2. If isSymmetric() is false, the ordering does matter.