de.unidu.is.retrieval.pire.dt
Class AbstractDT

java.lang.Object
  de.unidu.is.retrieval.pire.dt.AbstractDT

All Implemented Interfaces:

DT

Direct Known Subclasses:

NameDT, NumberDT, TextDT, TextGermanDT

public abstract class AbstractDT

extends java.lang.Object

implements DT

An abstract class for IR datatypes.

Since:

2003-09-23

Version:

$Revision: 1.12 $, $Date: 2005/03/14 17:33:13 $

Author:

Henrik Nottelmann

Constructor Summary

AbstractDT()

Method Summary

void	addProbRules(Index index, java.lang.String queryID, java.lang.String subqueryID, java.lang.String operator, java.util.List addList) Computes probabilities of relevance based on the RSVs (probabilities of inference).
void	addRSVRules(Index index, java.lang.String queryID, java.lang.String subqueryID, java.lang.String operator, double weight, java.lang.Object value, java.util.List addList) Adds and evaluates a query condition.
void	addToIndex(Index index, java.lang.String docID, java.lang.String operator, java.lang.Object value) Add the document content of the specified index.
void	computeIndex(Index index, java.lang.String operator) Computes the indexing weights for the specified index and the operator.
java.lang.String	convertOperator(java.lang.String operator) Converts the operator name into an identifier.
protected abstract Filter	getFilter(java.lang.String operator) Returns a filter for converting a document value into tokens/token frequency tuples.
java.util.Iterator	getIndexTokens(Index index, java.lang.String operator) Returns an iterator over all tokens for whom the indexing weight moments have to be computed.
protected java.lang.String	getProbsTemplate(Index index, java.lang.String queryID, java.lang.String subqueryID, java.lang.String operator) Returns a template for computing probabilities of relevance.
protected abstract Filter	getQueryFilter(java.lang.String operator) Returns a filter for converting a condition comparison value into tokens/token frequency tuples.
void	removeIndex(Index index, java.lang.String operator) Removes the index.
boolean	storedRSVs(java.lang.String operator) Tests whether the RSVs are already stored in the weight table, or if they are computed from outside.

Methods inherited from class java.lang.Object

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

Constructor Detail

AbstractDT

public AbstractDT()

Method Detail

getFilter

protected abstract Filter getFilter(java.lang.String operator)

Returns a filter for converting a document value into tokens/token frequency tuples.

Parameters:

operator - operator name

Returns:

filter

getQueryFilter

protected abstract Filter getQueryFilter(java.lang.String operator)

Returns a filter for converting a condition comparison value into tokens/token frequency tuples.

Parameters:

operator - operator name

Returns:

filter

convertOperator

public java.lang.String convertOperator(java.lang.String operator)

Converts the operator name into an identifier.

This implementation returns the operator name. Subclasses can override this behaviour.

Specified by:

convertOperator in interface DT

Parameters:

operator - operator name

Returns:

identifier for operator

addToIndex

public void addToIndex(Index index,
                       java.lang.String docID,
                       java.lang.String operator,
                       java.lang.Object value)

Add the document content of the specified index.

The specified value is applied to the filter corresponding to the operator filter, and the resulting tokens are added to the tf relation (together with the document id and the frequency of the token).

Specified by:

addToIndex in interface DT

Parameters:

index - underlying index

docID - document id

operator - operator name

value - attribute value

computeIndex

public void computeIndex(Index index,
                         java.lang.String operator)

Computes the indexing weights for the specified index and the operator.

This implementation uses a binary weighting scheme. Subclasses can use another weighting scheme by overriding this method.

Specified by:

computeIndex in interface DT

Parameters:

index - underlying index

operator - operator name

getIndexTokens

public java.util.Iterator getIndexTokens(Index index,
                                         java.lang.String operator)

Returns an iterator over all tokens for whom the indexing weight moments have to be computed.

The default implementation returns null, specifying that all tokens which stored in the index should be used. Other implementations can override this behaviour. TODO: really useful?

Specified by:

getIndexTokens in interface DT

Parameters:

index - underlying index

operator - operator name

Returns:

iterator over all tokens

removeIndex

public void removeIndex(Index index,
                        java.lang.String operator)

Removes the index.

Specified by:

removeIndex in interface DT

Parameters:

index - underlying index

operator - search operator name

storedRSVs

public boolean storedRSVs(java.lang.String operator)

Tests whether the RSVs are already stored in the weight table, or if they are computed from outside.

The default implemenation returns true. TODO: really useful?

Specified by:

storedRSVs in interface DT

Parameters:

operator - search operator

Returns:

true, if the RSVs are already stored

addRSVRules

public void addRSVRules(Index index,
                        java.lang.String queryID,
                        java.lang.String subqueryID,
                        java.lang.String operator,
                        double weight,
                        java.lang.Object value,
                        java.util.List addList)

Adds and evaluates a query condition.

This implementation adds a rule for using the indexing weight directly for the RSV.

Specified by:

addRSVRules in interface DT

Parameters:

index - underlying index

queryID - query id

subqueryID - subquery id

operator - search operator name

weight - condition weight

value - comparison value

addList - list to where the condition has to be added

getProbsTemplate

protected java.lang.String getProbsTemplate(Index index,
                                            java.lang.String queryID,
                                            java.lang.String subqueryID,
                                            java.lang.String operator)

Returns a template for computing probabilities of relevance.

The template string is an expression which contains the key PROB.

Parameters:

index - underlying index

queryID - query id

subqueryID - subquery id

operator - operator name

Returns:

template for computing probabilities of relevance

addProbRules

public void addProbRules(Index index,
                         java.lang.String queryID,
                         java.lang.String subqueryID,
                         java.lang.String operator,
                         java.util.List addList)

Computes probabilities of relevance based on the RSVs (probabilities of inference).

Specified by:

addProbRules in interface DT

Parameters:

index - underlying index

queryID - query id

subqueryID - subquery id

operator - search operator

addList - list to which rules can be appended