de.aitools.aq.algebra.vector.functions
Class CosineSimilarity

java.lang.Object
  de.aitools.aq.algebra.vector.functions.CosineSimilarity

All Implemented Interfaces:

Proximity<Vector>, Similarity<Vector>

public final class CosineSimilarity
extends java.lang.Object
implements Similarity<Vector>

A Similarity measure for Vectors.
The Cosine Similarity of two vectors is defined as the cosine of the angle between them.
It is calculated as the dot product of the both vectors divided by the multiplied euclidean norms of both.
The resulting similarity of a vector a and a vector b is between -1 (if a equals -b * k) and 1 (if a equals b * k), with k being a positive rational number (and thus not zero). A value of 0 signals that the two vectors are orthogonal or at least one of them is the null vector.
Proximity of this measure is equal to similarity. Normalized proximity is the proximity plus one and divided by two.

Version:

$Id: CosineSimilarity.java,v 1.3 2011/08/24 08:25:28 gollub Exp $

Author:

johannes.kiesel(/\t)uni-weimar.de

Constructor Summary

CosineSimilarity()
Construct a new CosineSimilarity.

Method Summary

double	computeNormalizedProximity(Vector v1, Vector v2) Compute a kind of proximity between two objects. The value returned by this method has to be between 0 (meaning they are as far away (according to the proximity measure) from each other as possible) and 1 (identical objects according to the proximity measure).
double	computeProximity(Vector v1, Vector v2) Compute a kind of proximity between two objects.
double	computeSimilarity(Vector v1, Vector v2) Compute the Similarity between two objects.

Methods inherited from class java.lang.Object

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

Constructor Detail

CosineSimilarity

public CosineSimilarity()

Construct a new CosineSimilarity.

Method Detail

computeSimilarity

public double computeSimilarity(Vector v1,
                                Vector v2)

Description copied from interface: Similarity

Compute the Similarity between two objects. A higher value signals higher similarity.

Specified by:

computeSimilarity in interface Similarity<Vector>

Parameters:

v1 - One object, to be compared to the...

v2 - ...second object

Returns:

Computed similarity

computeProximity

public double computeProximity(Vector v1,
                               Vector v2)

Description copied from interface: Proximity

Compute a kind of proximity between two objects. A higher value signals higher proximity.

Specified by:

computeProximity in interface Proximity<Vector>

Parameters:

v1 - One object, to be compared to the...

v2 - ...second object

Returns:

Computed proximity

computeNormalizedProximity

public double computeNormalizedProximity(Vector v1,
                                         Vector v2)

Description copied from interface: Proximity

Compute a kind of proximity between two objects.
The value returned by this method has to be between 0 (meaning they are as far away (according to the proximity measure) from each other as possible) and 1 (identical objects according to the proximity measure).

Specified by:

computeNormalizedProximity in interface Proximity<Vector>

Parameters:

v1 - One object, to be compared to the...

v2 - ...second object

Returns:

Computed normalized proximity