org.lenskit.mf.svd

Interface BiasedMFKernel

All Known Implementing Classes:

DomainClampingKernel, DotProductKernel

@DefaultImplementation(value=DotProductKernel.class)
public interface BiasedMFKernel

A kernel for biased matrix factorization. This function combines a user-item bias (baseline score) and the user- and item-factor vectors to make a final score.

Note that not all kernels are compatible with all model build strategies.

Kernels should be serializable and shareable.

Since:

2.1

Method Summary

Modifier and Type	Method and Description
double	apply(double bias, org.apache.commons.math3.linear.RealVector user, org.apache.commons.math3.linear.RealVector item) Apply the kernel function.

Method Detail

apply

double apply(double bias,
             @Nonnull
             org.apache.commons.math3.linear.RealVector user,
             @Nonnull
             org.apache.commons.math3.linear.RealVector item)

Apply the kernel function.

Parameters:

bias - The combined user-item bias term (the baseline score, usually).

user - The user-factor vector.

item - The item-factor vector.

Returns:

The kernel function value (combined score).

Throws:

IllegalArgumentException - if the user and item vectors have different lengths.