org.terrier.matching.models.queryexpansion
Class QueryExpansionModel

java.lang.Object
  org.terrier.matching.models.queryexpansion.QueryExpansionModel

Direct Known Subclasses:

Bo1, Bo2, CS, CSCorrect, Information, KL, KLComplete, KLCorrect

public abstract class QueryExpansionModel
extends java.lang.Object

This class should be extended by the classes used for weighting terms and documents.

Properties:

• rocchio.beta - defaults to 0.4d
• parameter.free.expansion - defaults to true.

Author:

Gianni Amati, Ben He, Vassilis Plachouras

Field Summary

protected double	averageDocumentLength The average document length in the collection.
protected double	collectionLength The number of tokens in the collection.
protected double	documentFrequency The document frequency of a term.
protected double	EXPANSION_DOCUMENTS The number of top-ranked documents in the pseudo relevance set.
protected double	EXPANSION_TERMS The number of the most weighted terms from the pseudo relevance set to be added to the original query.
protected Idf	idf An instance of Idf, in order to compute the logs.
protected double	maxTermFrequency The maximum in-collection term frequencty of the terms in the pseudo relevance set.
protected long	numberOfDocuments The number of documents in the collection.
boolean	PARAMETER_FREE Boolean variable indicates whether to apply the parameter free query expansion.
double	ROCCHIO_BETA Rocchio's beta for query expansion.
protected double	totalDocumentLength The total length of the X top-retrieved documents.

Constructor Summary

QueryExpansionModel()
A default constructor for the class that initialises the idf attribute.

Method Summary

abstract java.lang.String	getInfo() Returns the name of the model.
void	initialise() Initialises the Rocchio's beta for query expansion.
abstract double	parameterFreeNormaliser() This method provides the contract for computing the normaliser of parameter-free query expansion.
abstract double	parameterFreeNormaliser(double _maxTermFrequency, double _collectionLength, double _totalDocumentLength) This method provides the contract for computing the normaliser of parameter-free query expansion.
abstract double	score(double withinDocumentFrequency, double termFrequency) This method provides the contract for implementing query expansion models.
abstract double	score(double withinDocumentFrequency, double termFrequency, double _totalDocumentLength, double _collectionLength, double _averageDocumentLength) This method provides the contract for implementing query expansion models.
void	setAverageDocumentLength(double _averageDocumentLength) Set the average document length.
void	setCollectionLength(double _collectionLength) Set the collection length.
void	setDocumentFrequency(double _documentFrequency) Set the document frequency.
void	setMaxTermFrequency(double _maxTermFrequency) This method sets the maximum of the term frequency values of query terms.
void	setNumberOfDocuments(long _numberOfDocuments)
void	setTotalDocumentLength(double _totalDocumentLength) Set the total document length.

Methods inherited from class java.lang.Object

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

Field Detail

averageDocumentLength

protected double averageDocumentLength

The average document length in the collection.

totalDocumentLength

protected double totalDocumentLength

The total length of the X top-retrieved documents. X is given by system setting.

collectionLength

protected double collectionLength

The number of tokens in the collection.

documentFrequency

protected double documentFrequency

The document frequency of a term.

idf

protected Idf idf

An instance of Idf, in order to compute the logs.

maxTermFrequency

protected double maxTermFrequency

The maximum in-collection term frequencty of the terms in the pseudo relevance set.

numberOfDocuments

protected long numberOfDocuments

The number of documents in the collection.

EXPANSION_DOCUMENTS

protected double EXPANSION_DOCUMENTS

The number of top-ranked documents in the pseudo relevance set.

EXPANSION_TERMS

protected double EXPANSION_TERMS

The number of the most weighted terms from the pseudo relevance set to be added to the original query. There can be overlap between the original query terms and the added terms from the pseudo relevance set.

ROCCHIO_BETA

public double ROCCHIO_BETA

Rocchio's beta for query expansion. Its default value is 0.4.

PARAMETER_FREE

public boolean PARAMETER_FREE

Boolean variable indicates whether to apply the parameter free query expansion.

Constructor Detail

QueryExpansionModel

public QueryExpansionModel()

A default constructor for the class that initialises the idf attribute.

Method Detail

initialise

public void initialise()

Initialises the Rocchio's beta for query expansion.

setNumberOfDocuments

public void setNumberOfDocuments(long _numberOfDocuments)

Parameters:

_numberOfDocuments - the numberOfDocuments to set

getInfo

public abstract java.lang.String getInfo()

Returns the name of the model. Creation date: (19/06/2003 12:09:55)

Returns:

java.lang.String

setAverageDocumentLength

public void setAverageDocumentLength(double _averageDocumentLength)

Set the average document length.

Parameters:

_averageDocumentLength - double The average document length.

setCollectionLength

public void setCollectionLength(double _collectionLength)

Set the collection length.

Parameters:

_collectionLength - double The number of tokens in the collection.

setDocumentFrequency

public void setDocumentFrequency(double _documentFrequency)

Set the document frequency.

Parameters:

_documentFrequency - double The document frequency of a term.

setTotalDocumentLength

public void setTotalDocumentLength(double _totalDocumentLength)

Set the total document length.

Parameters:

_totalDocumentLength - double The total document length.

setMaxTermFrequency

public void setMaxTermFrequency(double _maxTermFrequency)

This method sets the maximum of the term frequency values of query terms.

Parameters:

_maxTermFrequency -

parameterFreeNormaliser

public abstract double parameterFreeNormaliser()

This method provides the contract for computing the normaliser of parameter-free query expansion.

Returns:

The normaliser.

parameterFreeNormaliser

public abstract double parameterFreeNormaliser(double _maxTermFrequency,
                                               double _collectionLength,
                                               double _totalDocumentLength)

This method provides the contract for computing the normaliser of parameter-free query expansion.

Parameters:

_maxTermFrequency - The maximum of the in-collection term frequency of the terms in the pseudo relevance set.

_collectionLength - The number of tokens in the collections.

_totalDocumentLength - The sum of the length of the top-ranked documents.

Returns:

The normaliser.

score

public abstract double score(double withinDocumentFrequency,
                             double termFrequency)

This method provides the contract for implementing query expansion models.

Parameters:

withinDocumentFrequency - double The term frequency in the X top-retrieved documents.

termFrequency - double The term frequency in the collection.

Returns:

the score assigned to a document with the parameters, and other preset parameters

score

public abstract double score(double withinDocumentFrequency,
                             double termFrequency,
                             double _totalDocumentLength,
                             double _collectionLength,
                             double _averageDocumentLength)

This method provides the contract for implementing query expansion models. For some models, we have to set the beta and the documentFrequency of a term.

Parameters:

withinDocumentFrequency - double The term frequency in the X top-retrieved documents.

termFrequency - double The term frequency in the collection.

_totalDocumentLength - double The sum of length of the X top-retrieved documents.

_collectionLength - double The number of tokens in the whole collection.

_averageDocumentLength - double The average document length in the collection.

Returns:

double The score returned by the implemented model.