it.uniroma2.sag.kelp.data.representation.structure.similarity

Class VectorBasedStructureElementSimilarity

java.lang.Object

it.uniroma2.sag.kelp.data.representation.structure.similarity.VectorBasedStructureElementSimilarity

All Implemented Interfaces:

StructureElementSimilarityI

Direct Known Subclasses:

LexicalStructureElementSimilarity

public abstract class VectorBasedStructureElementSimilarity
extends Object
implements StructureElementSimilarityI

This class provides a useful abstraction of a StructureElement similarity based on a wordspace. Each StructureElement can be associated to a vector. Then the similarity between two structure elements is a the kernel operation between their corresponding vectors. If the structure elements have been enriched using a manipulator (i.e., LexicalStructureElementEnricher), the associated vectors are directly retrieved. Otherwise they are retrieved by a wordspace.

Author:

Danilo Croce, Simone Filice

Field Summary

Fields

Modifier and Type	Field and Description
static String	VECTOR_NAME

Constructor Summary

Constructors

Constructor and Description
VectorBasedStructureElementSimilarity()

Method Summary

Methods

Modifier and Type	Method and Description
protected Vector	getCorrespondingVector(StructureElement element) Returns the vector associated to element.
String	getEnrichmentName() Returns the identifier of the vectors associated to a StructureElement during the manipulation operation performed by a Manipulator (i.e.
Kernel	getKernel() Returns the kernel used in comparing two vectors
protected float	getSimilarity(Vector vector1, Vector vector2) Returns the similarity between vector1 and vector2 computed using the kernel function
WordspaceI	getWordspace() Returns the wordspace from which the vectors associated to a word must be retrieved
void	setEnrichmentName(String enrichmentName) Sets the identifier of the vectors associated to a StructureElement during the manipulation operation performed by a Manipulator (i.e.
void	setKernel(Kernel kernel) Sets the kernel to be used in comparing two vectors
void	setWordspace(WordspaceI wordspace) Sets the wordspace from which the vectors associated to a word must be retrieved

Methods inherited from class java.lang.Object

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

Methods inherited from interface it.uniroma2.sag.kelp.data.representation.structure.similarity.StructureElementSimilarityI

sim

Field Detail

VECTOR_NAME

public static final String VECTOR_NAME

See Also:

Constant Field Values

Constructor Detail

VectorBasedStructureElementSimilarity

public VectorBasedStructureElementSimilarity()

Method Detail

getKernel

public Kernel getKernel()

Returns the kernel used in comparing two vectors

Returns:

the kernel used in comparing two vectors

setKernel

public void setKernel(Kernel kernel)

Sets the kernel to be used in comparing two vectors

NOTE: the kernel cache mechanism cannot be enabled. Thus the cache is automatically disabled to kernel

Parameters:

kernel - the kernel to be used in comparing two vectors

getWordspace

public WordspaceI getWordspace()

Returns the wordspace from which the vectors associated to a word must be retrieved

Returns:

the wordspace from which the vectors associated to a word must be retrieved

setWordspace

public void setWordspace(WordspaceI wordspace)

Sets the wordspace from which the vectors associated to a word must be retrieved

Parameters:

wordspace - the wordspace from which the vectors associated to a word must be retrieved

getEnrichmentName

public String getEnrichmentName()

Returns the identifier of the vectors associated to a StructureElement during the manipulation operation performed by a Manipulator (i.e. LexicalStructureElementEnricher)

Returns:

the enrichmentName

setEnrichmentName

public void setEnrichmentName(String enrichmentName)

Sets the identifier of the vectors associated to a StructureElement during the manipulation operation performed by a Manipulator (i.e. LexicalStructureElementEnricher) This manipulation optimize the similarity evaluation cause it avoids retrieving vectors from the wordspaces

Parameters:

enrichmentName - the enrichmentName to set

getCorrespondingVector

protected Vector getCorrespondingVector(StructureElement element)

Returns the vector associated to element. It firstly tries to check whether a vector as been attached as additional information to element. If no vector has been attached it retrieves the vector associated to the textual representation of element (applying element.getTextFromData()) in the wordspace.

Parameters:

element - the structure element whose corresponding vector must be retrieved

Returns:

the vector associated to element

getSimilarity

protected float getSimilarity(Vector vector1,
                  Vector vector2)

Returns the similarity between vector1 and vector2 computed using the kernel function

Parameters:

vector1 - the first vector to be compared

vector2 - the second vector to be compared

Returns:

the similarity between vector1 and vector2