org.deidentifier.arx.metric.v2

Class MetricMDNMLoss

java.lang.Object

org.deidentifier.arx.metric.Metric<AbstractILMultiDimensional>

org.deidentifier.arx.metric.v2.AbstractMetricMultiDimensional

org.deidentifier.arx.metric.v2.MetricMDNMLoss

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

MetricMDNMLossPrecomputed

public class MetricMDNMLoss
extends AbstractMetricMultiDimensional

This class implements a variant of the Loss metric.

See Also:

Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from class org.deidentifier.arx.metric.Metric

Metric.AggregateFunction

Field Summary

Fields inherited from class org.deidentifier.arx.metric.v2.AbstractMetricMultiDimensional

k

Constructor Summary

Constructors

Constructor and Description
MetricMDNMLoss() Default constructor which treats all transformation methods equally.
MetricMDNMLoss(double gsFactor, Metric.AggregateFunction function) A constructor that allows to define a factor weighting generalization and suppression.
MetricMDNMLoss(Metric.AggregateFunction function) Default constructor which treats all transformation methods equally.

Method Summary

Modifier and Type	Method and Description
MetricConfiguration	getConfiguration() Returns the configuration of this metric.
double	getGeneralizationFactor() Returns the factor used weight generalized values.
double	getGeneralizationSuppressionFactor() Returns the factor weighting generalization and suppression.
protected ILMultiDimensionalWithBound	getInformationLossInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node, org.deidentifier.arx.framework.check.groupify.HashGroupify g) Evaluates the metric for the given node.
protected ILMultiDimensionalWithBound	getInformationLossInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node, org.deidentifier.arx.framework.check.groupify.HashGroupifyEntry entry) Returns the information loss that would be induced by suppressing the given entry.
protected AbstractILMultiDimensional	getLowerBoundInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node) Returns a lower bound for the information loss for the given node.
protected AbstractILMultiDimensional	getLowerBoundInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node, org.deidentifier.arx.framework.check.groupify.HashGroupify g) Returns a lower bound for the information loss for the given node.
java.lang.String	getName() Returns the name of metric.
ILScore	getScore(org.deidentifier.arx.framework.lattice.Transformation<?> node, org.deidentifier.arx.framework.check.groupify.HashGroupify groupify) Calculates the score.
protected DomainShare[]	getShares() For subclasses.
double	getSuppressionFactor() Returns the factor used to weight suppressed values.
protected void	initializeInternal(org.deidentifier.arx.framework.data.DataManager manager, DataDefinition definition, org.deidentifier.arx.framework.data.Data input, org.deidentifier.arx.framework.data.GeneralizationHierarchy[] hierarchies, ARXConfiguration config) Implement this to initialize the metric.
boolean	isAbleToHandleMicroaggregation() Returns whether this metric handles microaggregation
boolean	isGSFactorSupported() Returns whether a generalization/suppression factor is supported
boolean	isScoreFunctionSupported() Returns whether the metric provides a score function
protected double	normalizeAggregated(double aggregate) Normalizes the aggregate.
protected double	normalizeGeneralized(double aggregate, int dimension) Normalizes the aggregate.
ElementData	render(ARXConfiguration config) Renders the privacy model
java.lang.String	toString() Returns the name of metric.

Methods inherited from class org.deidentifier.arx.metric.v2.AbstractMetricMultiDimensional

createInformationLoss, createMaxInformationLoss, createMinInformationLoss, getAggregateFunction, getAggregationFunctionsGeneralized, getAggregationFunctionsNonGeneralized, getAggregationIndicesGeneralized, getAggregationIndicesNonGeneralized, getAggregationInformation, getDimensions, getDimensionsAggregated, getDimensionsGeneralized, initialize, setMax, setMin

Methods inherited from class org.deidentifier.arx.metric.Metric

createAECSMetric, createAECSMetric, createAmbiguityMetric, createClassificationMetric, createClassificationMetric, createDiscernabilityMetric, createDiscernabilityMetric, createEntropyBasedInformationLossMetric, createEntropyBasedInformationLossMetric, createEntropyMetric, createEntropyMetric, createEntropyMetric, createEntropyMetric, createEntropyMetric, createEntropyMetric, createHeightMetric, createHeightMetric, createInstanceOfHighestScore, createInstanceOfLowestScore, createKLDivergenceMetric, createLossMetric, createLossMetric, createLossMetric, createLossMetric, createMetric, createNormalizedEntropyMetric, createNormalizedEntropyMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecisionMetric, createPrecomputedEntropyMetric, createPrecomputedEntropyMetric, createPrecomputedEntropyMetric, createPrecomputedEntropyMetric, createPrecomputedEntropyMetric, createPrecomputedEntropyMetric, createPrecomputedLossMetric, createPrecomputedLossMetric, createPrecomputedLossMetric, createPrecomputedLossMetric, createPrecomputedNormalizedEntropyMetric, createPrecomputedNormalizedEntropyMetric, createPublisherPayoutMetric, createPublisherPayoutMetric, createStaticMetric, createStaticMetric, getDescription, getDescription, getInformationLoss, getInformationLoss, getLowerBound, getLowerBound, getNumRecords, getSubset, initialize, isAbleToHandleClusteredMicroaggregation, isIndependent, isMonotonic, isMonotonicWithGeneralization, isMonotonicWithSuppression, isMultiDimensional, isPrecomputed, isWeighted, list, round

Methods inherited from class java.lang.Object

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

Constructor Detail

MetricMDNMLoss

public MetricMDNMLoss()

Default constructor which treats all transformation methods equally.

MetricMDNMLoss

public MetricMDNMLoss(Metric.AggregateFunction function)

Default constructor which treats all transformation methods equally.

Parameters:

function -

MetricMDNMLoss

public MetricMDNMLoss(double gsFactor,
                      Metric.AggregateFunction function)

A constructor that allows to define a factor weighting generalization and suppression.

Parameters:

gsFactor - A factor [0,1] weighting generalization and suppression. The default value is 0.5, which means that generalization and suppression will be treated equally. A factor of 0 will favor suppression, and a factor of 1 will favor generalization. The values in between can be used for balancing both methods.

function -

Method Detail

getConfiguration

public MetricConfiguration getConfiguration()

Returns the configuration of this metric.

Overrides:

getConfiguration in class Metric<AbstractILMultiDimensional>

Returns:

getGeneralizationFactor

public double getGeneralizationFactor()

Description copied from class: Metric

Returns the factor used weight generalized values.

Overrides:

getGeneralizationFactor in class Metric<AbstractILMultiDimensional>

Returns:

getGeneralizationSuppressionFactor

public double getGeneralizationSuppressionFactor()

Description copied from class: Metric

Returns the factor weighting generalization and suppression.

Overrides:

getGeneralizationSuppressionFactor in class Metric<AbstractILMultiDimensional>

Returns:

A factor [0,1] weighting generalization and suppression. The default value is 0.5, which means that generalization and suppression will be treated equally. A factor of 0 will favor suppression, and a factor of 1 will favor generalization. The values in between can be used for balancing both methods.

getName

public java.lang.String getName()

Description copied from class: Metric

Returns the name of metric.

Overrides:

getName in class Metric<AbstractILMultiDimensional>

Returns:

getScore

public ILScore getScore(org.deidentifier.arx.framework.lattice.Transformation<?> node,
                        org.deidentifier.arx.framework.check.groupify.HashGroupify groupify)

Description copied from class: Metric

Calculates the score. Note: All score functions are expected to return a score value divided by the sensitivity of the score function.

Overrides:

getScore in class Metric<AbstractILMultiDimensional>

Returns:

getSuppressionFactor

public double getSuppressionFactor()

Description copied from class: Metric

Returns the factor used to weight suppressed values.

Overrides:

getSuppressionFactor in class Metric<AbstractILMultiDimensional>

Returns:

isAbleToHandleMicroaggregation

public boolean isAbleToHandleMicroaggregation()

Description copied from class: Metric

Returns whether this metric handles microaggregation

Overrides:

isAbleToHandleMicroaggregation in class Metric<AbstractILMultiDimensional>

Returns:

isGSFactorSupported

public boolean isGSFactorSupported()

Description copied from class: Metric

Returns whether a generalization/suppression factor is supported

Overrides:

isGSFactorSupported in class Metric<AbstractILMultiDimensional>

Returns:

isScoreFunctionSupported

public boolean isScoreFunctionSupported()

Description copied from class: Metric

Returns whether the metric provides a score function

Overrides:

isScoreFunctionSupported in class Metric<AbstractILMultiDimensional>

Returns:

render

public ElementData render(ARXConfiguration config)

Description copied from class: Metric

Renders the privacy model

Specified by:

render in class Metric<AbstractILMultiDimensional>

Returns:

toString

public java.lang.String toString()

Description copied from class: Metric

Returns the name of metric.

Overrides:

toString in class Metric<AbstractILMultiDimensional>

getInformationLossInternal

protected ILMultiDimensionalWithBound getInformationLossInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node,
                                                                 org.deidentifier.arx.framework.check.groupify.HashGroupify g)

Description copied from class: Metric

Evaluates the metric for the given node.

Specified by:

getInformationLossInternal in class Metric<AbstractILMultiDimensional>

Parameters:

node - The node for which to compute the information loss

g - The groupify operator of the previous check

Returns:

the double

getInformationLossInternal

protected ILMultiDimensionalWithBound getInformationLossInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node,
                                                                 org.deidentifier.arx.framework.check.groupify.HashGroupifyEntry entry)

Description copied from class: Metric

Returns the information loss that would be induced by suppressing the given entry. The loss is not necessarily consistent with the loss that is computed by getInformationLoss(node, groupify) but is guaranteed to be comparable for different entries from the same groupify operator.

Specified by:

getInformationLossInternal in class Metric<AbstractILMultiDimensional>

Returns:

getLowerBoundInternal

protected AbstractILMultiDimensional getLowerBoundInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node)

Description copied from class: Metric

Returns a lower bound for the information loss for the given node. This can be used to expose the results of monotonic shares of a metric, which can significantly speed-up the anonymization process. If no such metric exists, simply return null.

Specified by:

getLowerBoundInternal in class Metric<AbstractILMultiDimensional>

Returns:

getLowerBoundInternal

protected AbstractILMultiDimensional getLowerBoundInternal(org.deidentifier.arx.framework.lattice.Transformation<?> node,
                                                           org.deidentifier.arx.framework.check.groupify.HashGroupify g)

Description copied from class: Metric

Returns a lower bound for the information loss for the given node. This can be used to expose the results of monotonic shares of a metric, which can significantly speed-up the anonymization process. If no such metric exists, simply return null.

This variant of the method allows computing a monotonic share based on a groupified data representation. IMPORTANT NOTE: The groups may not have been classified correctly when the method is called, i.e., HashGroupifyEntry.isNotOutlier may not be set correctly!

Specified by:

getLowerBoundInternal in class Metric<AbstractILMultiDimensional>

Returns:

getShares

protected DomainShare[] getShares()

For subclasses.

Returns:

initializeInternal

protected void initializeInternal(org.deidentifier.arx.framework.data.DataManager manager,
                                  DataDefinition definition,
                                  org.deidentifier.arx.framework.data.Data input,
                                  org.deidentifier.arx.framework.data.GeneralizationHierarchy[] hierarchies,
                                  ARXConfiguration config)

Description copied from class: Metric

Implement this to initialize the metric.

Overrides:

initializeInternal in class AbstractMetricMultiDimensional

normalizeAggregated

protected double normalizeAggregated(double aggregate)

Normalizes the aggregate.

Parameters:

aggregate -

dimension -

Returns:

normalizeGeneralized

protected double normalizeGeneralized(double aggregate,
                                      int dimension)

Normalizes the aggregate.

Parameters:

aggregate -

dimension -

Returns: