org.simmetrics.metrics

Class CosineSimilarity<T>

java.lang.Object

org.simmetrics.metrics.CosineSimilarity<T>

Type Parameters:

T - type of the token

All Implemented Interfaces:

Distance<com.google.common.collect.Multiset<T>>, Metric<com.google.common.collect.Multiset<T>>, MultisetDistance<T>, MultisetMetric<T>

public final class CosineSimilarity<T>
extends Object
implements MultisetMetric<T>, MultisetDistance<T>

Calculates the cosine similarity over two multisets. The similarity is defined as the cosine of the angle between the multisets expressed as sparse vectors.

similarity(a,b) = a·b / (||a|| * ||b||)
distance(a,b) = 1 - similarity(a,b)

The cosine similarity is identical to the Tanimoto coefficient, but unlike Tanimoto the occurrence (cardinality) of an entry is taken into account. E.g. [hello, world] and [hello, world, hello, world] would be identical when compared with Tanimoto but are dissimilar when the cosine similarity is used.

This class is immutable and thread-safe.

See Also:

TanimotoCoefficient, Wikipedia Cosine similarity

Constructor Summary

Constructors

Constructor and Description
CosineSimilarity()

Method Summary

Modifier and Type	Method and Description
float	compare(com.google.common.collect.Multiset<T> a, com.google.common.collect.Multiset<T> b) Measures the similarity between multisets a and b.
float	distance(com.google.common.collect.Multiset<T> a, com.google.common.collect.Multiset<T> b) Measures the distance between multisets a and b.
String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

CosineSimilarity

public CosineSimilarity()

Method Detail

compare

public float compare(com.google.common.collect.Multiset<T> a,
                     com.google.common.collect.Multiset<T> b)

Description copied from interface: MultisetMetric

Measures the similarity between multisets a and b. The measurement results in a value between 0 and 1 inclusive. A value of zero indicates that the sets are dissimilar, a value of 1 indicates they are similar.

Results are undefined if a and b are based on different equivalence relations (as HashMultiset and TreeMultiset are).

Specified by:

compare in interface Metric<com.google.common.collect.Multiset<T>>

Specified by:

compare in interface MultisetMetric<T>

Parameters:

a - multisets a to compare

b - multisets b to compare

Returns:

a value between 0 and 1 inclusive indicating similarity

distance

public float distance(com.google.common.collect.Multiset<T> a,
                      com.google.common.collect.Multiset<T> b)

Description copied from interface: MultisetDistance

Measures the distance between multisets a and b. The measurement results in a non-negative value. A value of 0.0 indicates that a and b are similar.

Results are undefined if a and b are based on different equivalence relations (as HashMultiset and TreeMultiset are).

Specified by:

distance in interface Distance<com.google.common.collect.Multiset<T>>

Specified by:

distance in interface MultisetDistance<T>

Parameters:

a - list a to compare

b - list b to compare

Returns:

a non-negative value

toString

public String toString()

Overrides:

toString in class Object