Package com.gengoai.hermes.similarity

Class EmbeddingSimilarity

java.lang.Object

com.gengoai.hermes.similarity.EmbeddingSimilarity

All Implemented Interfaces:

HStringSimilarity, Serializable

public class EmbeddingSimilarity
extends Object
implements HStringSimilarity, Serializable

Implementation of a HStringSimilarity that calculates similarity based on the similarity between the HStrings in embedding space. This expects that embeddings have been assigned to the HString or tokens.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor	Description
EmbeddingSimilarity(@NonNull Similarity measure)	Instantiates a new EmbeddingSimilarity.

Method Summary

Modifier and Type	Method	Description
double	calculate(@NonNull HString first, @NonNull HString second)	Calculates the similarity between the two given HString
void	fit(@NonNull DocumentCollection corpus)	In certain cases a HStringSimilarity needs to collect corpus level statistics to determine similarity.

Methods inherited from class java.lang.Object

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

Constructor Detail

EmbeddingSimilarity

public EmbeddingSimilarity(@NonNull
                           @NonNull Similarity measure)

Instantiates a new EmbeddingSimilarity.

Parameters:

measure - the similarity measure to use

Method Detail

calculate

public double calculate(@NonNull
                        @NonNull HString first,
                        @NonNull
                        @NonNull HString second)

Description copied from interface: HStringSimilarity

Calculates the similarity between the two given HString

Specified by:

calculate in interface HStringSimilarity

Parameters:

first - the first HString

second - the second HString

Returns:

the similarity between first and second

fit

public void fit(@NonNull
                @NonNull DocumentCollection corpus)

Description copied from interface: HStringSimilarity

In certain cases a HStringSimilarity needs to collect corpus level statistics to determine similarity. The fit method allows implementations to perform this logic at a corpus level.

Specified by:

fit in interface HStringSimilarity

Parameters:

corpus - the corpus to fit the similarity measure to