es.uam.eps.ir.ranksys.nn.sim

Class VectorSimilarity

java.lang.Object

es.uam.eps.ir.ranksys.nn.sim.VectorSimilarity

All Implemented Interfaces:

Similarity

Direct Known Subclasses:

VectorCosineSimilarity, VectorJaccardSimilarity

public abstract class VectorSimilarity
extends Object
implements Similarity

Vector similarity. Based on the inner product of item/user profiles as vectors.

Author:

Saúl Vargas (saul.vargas@uam.es)

Field Summary

Fields

Modifier and Type	Field and Description
protected FastPreferenceData<?,?>	data User-item preferences.
protected boolean	dense If true, dense vectors are used to calculate similarities.
protected double[]	norm2Array Cached normalization for when dense is true.
protected it.unimi.dsi.fastutil.ints.Int2DoubleMap	norm2Map Cached normalization for when dense is false.

Constructor Summary

Constructors

Constructor and Description
VectorSimilarity(FastPreferenceData<?,?> data, boolean dense) Constructor.

Method Summary

Modifier and Type	Method and Description
protected abstract double	sim(double product, double norm2A, double norm2B) Calculates the similarity value.
Stream<Tuple2id>	similarElems(int idx1) Returns all the users/items having a similarity greater than 0, together with the value of the similarity.
IntToDoubleFunction	similarity(int idx1) Returns a function that returns the similarity with the user/item.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface es.uam.eps.ir.ranksys.nn.sim.Similarity

similarity

Field Detail

data

protected final FastPreferenceData<?,?> data

User-item preferences.

dense

protected final boolean dense

If true, dense vectors are used to calculate similarities.

norm2Map

protected final it.unimi.dsi.fastutil.ints.Int2DoubleMap norm2Map

Cached normalization for when dense is false.

norm2Array

protected final double[] norm2Array

Cached normalization for when dense is true.

Constructor Detail

VectorSimilarity

public VectorSimilarity(FastPreferenceData<?,?> data,
                        boolean dense)

Constructor. Uses maps for internal calculation.

Parameters:

data - preference data

dense - true for array-based calculations, false to map-based

Method Detail

similarity

public IntToDoubleFunction similarity(int idx1)

Description copied from interface: Similarity

Returns a function that returns the similarity with the user/item.

Specified by:

similarity in interface Similarity

Parameters:

idx1 - index of the user/item

Returns:

function that returns the similarity with the index

similarElems

public Stream<Tuple2id> similarElems(int idx1)

Description copied from interface: Similarity

Returns all the users/items having a similarity greater than 0, together with the value of the similarity.

Specified by:

similarElems in interface Similarity

Parameters:

idx1 - index of user/item

Returns:

stream of index-similarity pairs

sim

protected abstract double sim(double product,
                              double norm2A,
                              double norm2B)

Calculates the similarity value.

Parameters:

product - value of the inner product of vectors

norm2A - square of the norm of the first vector

norm2B - square of the norm of the second vector

Returns:

similarity value