@Override
@Deprecated
public double getInformationLoss(final Node node) {
check(node);
metric.evaluate(node, currentGroupify);
return node.getInformationLoss().getValue();
}
@Override
public void check(final Node node) {
// Store snapshot from last check
if (stateMachine.getLastNode() != null) {
history.store(
stateMachine.getLastNode(), currentGroupify, stateMachine.getLastTransition().snapshot);
}
// Transition
final Transition transition = stateMachine.transition(node);
// Switch groupifies
final IHashGroupify temp = lastGroupify;
lastGroupify = currentGroupify;
currentGroupify = temp;
currentGroupify.clear();
// Apply transition
switch (transition.type) {
case UNOPTIMIZED:
currentGroupify =
transformer.apply(transition.projection, node.getTransformation(), currentGroupify);
break;
case ROLLUP:
currentGroupify =
transformer.applyRollup(
transition.projection, node.getTransformation(), lastGroupify, currentGroupify);
break;
case SNAPSHOT:
currentGroupify =
transformer.applySnapshot(
transition.projection,
node.getTransformation(),
currentGroupify,
transition.snapshot);
break;
}
// Mark as checked
node.setChecked();
// Propagate k-anonymity
node.setKAnonymous(currentGroupify.isKAnonymous());
// Propagate anonymity and information loss
if (currentGroupify.isAnonymous()) {
node.setAnonymous(true);
metric.evaluate(node, currentGroupify);
} else {
node.setInformationLoss(null);
node.setAnonymous(false);
}
}
@Override
@Deprecated
public Data transform(final Node node) {
// Apply transition and groupify
currentGroupify.clear();
currentGroupify = transformer.apply(0L, node.getTransformation(), currentGroupify);
// Determine outliers and set infoloss
if (!node.isChecked()) {
node.setChecked();
node.setAnonymous(currentGroupify.isAnonymous());
metric.evaluate(node, currentGroupify);
node.setTagged();
}
return getBuffer();
}
@Override
public Data transformAndMarkOutliers(final Node node) {
// Apply transition and groupify
currentGroupify.clear();
currentGroupify = transformer.apply(0L, node.getTransformation(), currentGroupify);
// Determine outliers and set infoloss
node.setAnonymous(currentGroupify.isAnonymous());
if (!node.isChecked()) {
node.setChecked();
metric.evaluate(node, currentGroupify);
node.setTagged();
}
// Find outliers
if (config.getAbsoluteMaxOutliers() != 0) {
currentGroupify.markOutliers(transformer.getBuffer());
}
// Return the buffer
return getBuffer();
}
/**
* Entry point.
*
* @param args the arguments
*/
public static void main(String[] args) throws IOException {
// Define data
Data data = getData();
// Define attribute types
data.getDefinition().setAttributeType("age", getHierarchyAge());
data.getDefinition().setAttributeType("zipcode", getHierarchyZipcode());
data.getDefinition().setAttributeType("disease1", AttributeType.SENSITIVE_ATTRIBUTE);
data.getDefinition().setAttributeType("disease2", AttributeType.SENSITIVE_ATTRIBUTE);
// Create an instance of the anonymizer
ARXAnonymizer anonymizer = new ARXAnonymizer();
ARXConfiguration config = ARXConfiguration.create();
config.addPrivacyModel(new KAnonymity(3));
config.addPrivacyModel(
new HierarchicalDistanceTCloseness("disease1", 0.6d, getHierarchyDisease()));
config.addPrivacyModel(new RecursiveCLDiversity("disease2", 3d, 2));
config.setMaxOutliers(0d);
config.setQualityModel(Metric.createEntropyMetric());
// Now anonymize
ARXResult result = anonymizer.anonymize(data, config);
// Print info
printResult(result, data);
// Process results
if (result.getGlobalOptimum() != null) {
System.out.println(" - Transformed data:");
Iterator<String[]> transformed = result.getOutput(false).iterator();
while (transformed.hasNext()) {
System.out.print(" ");
System.out.println(Arrays.toString(transformed.next()));
}
}
}
/**
* Returns the test cases.
*
* @return
* @throws IOException
*/
@Parameters(name = "{index}:[{0}]")
public static Collection<Object[]> cases() throws IOException {
return Arrays.asList(
new Object[][] {
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
3,
0.01d,
ARXPopulationModel.create(Region.USA),
CellSizeEstimator.ZERO_TRUNCATED_POISSON)),
"occupation",
"./data/adult.csv",
130804.5332092598,
new int[] {0, 0, 1, 1, 0, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
1000,
0.01d,
ARXPopulationModel.create(Region.USA),
CellSizeEstimator.ZERO_TRUNCATED_POISSON)),
"occupation",
"./data/adult.csv",
151894.1394841501,
new int[] {0, 0, 1, 1, 1, 2, 1, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createAECSMetric())
.addPrivacyModel(
new KMap(
5,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
45.014925373134325,
new int[] {1, 0, 1, 2, 3, 2, 2, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new KMap(
3,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
23387.494246375998,
new int[] {0, 0, 1, 2, 3, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
5,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
28551.7222913157,
new int[] {1, 0, 1, 2, 3, 2, 2, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createAECSMetric())
.addPrivacyModel(
new KMap(
20,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
11.424242424242424,
new int[] {1, 0, 1, 1, 3, 2, 1, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new KMap(
7,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
17075.7181747451,
new int[] {0, 0, 1, 1, 2, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
3,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
15121.633326877098,
new int[] {0, 0, 1, 1, 1, 2, 1, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createAECSMetric())
.addPrivacyModel(
new KMap(
5,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
45.014925373134325,
new int[] {1, 0, 1, 2, 3, 2, 2, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new KMap(
2,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
23108.1673304724,
new int[] {1, 0, 1, 1, 3, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
10,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
30238.2081484441,
new int[] {0, 1, 1, 2, 3, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createAECSMetric())
.addPrivacyModel(
new KMap(
10,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
7.215311004784689,
new int[] {0, 0, 1, 1, 3, 2, 1, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new KMap(
5,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
17053.8743069776,
new int[] {0, 0, 1, 0, 2, 2, 2, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new KMap(
3,
DataSubset.create(
Data.create("./data/adult.csv", StandardCharsets.UTF_8, ';'),
Data.create(
"./data/adult_subset.csv", StandardCharsets.UTF_8, ';')))),
"occupation",
"./data/adult.csv",
15121.633326877098,
new int[] {0, 0, 1, 1, 1, 2, 1, 0},
false)
},
});
}
/**
* Update the view.
*
* @param part
*/
protected void doUpdate(ModelPart part) {
// Check model
if (model == null) {
return;
}
// Obtain definition
DataDefinition definition = model.getOutputDefinition();
if (definition == null) definition = model.getInputDefinition();
// Obtain relevant configuration objects;
ModelConfiguration config = null;
Metric<?> metric = null;
if (model.getOutputConfig() != null) {
config = model.getOutputConfig();
metric = config.getMetric();
// We don't need to update in many cases, if we are displaying an output configuration
if (part == ModelPart.ATTRIBUTE_TYPE
|| part == ModelPart.METRIC
|| part == ModelPart.ATTRIBUTE_WEIGHT
|| part == ModelPart.GS_FACTOR
|| part == ModelPart.MAX_OUTLIERS
|| part == ModelPart.DATA_TYPE
|| part == ModelPart.FINANCIAL_MODEL) {
return;
}
} else {
config = model.getInputConfig();
// TODO: This is such an ugly hack
metric = model.getMetricDescription().createInstance(model.getMetricConfiguration());
}
// Check
if (definition == null || config == null || model.getInputConfig().getInput() == null) {
reset();
return;
}
// Obtain handle
DataHandle data = model.getInputConfig().getInput().getHandle();
// Disable redrawing
root.setRedraw(false);
// Clear
roots.clear();
// Print basic properties
new Property(
Resources.getMessage("PropertiesView.9"),
new String[] {String.valueOf(data.getNumRows())}); // $NON-NLS-1$
new Property(
Resources.getMessage("PropertiesView.10"),
new String[] {
SWTUtil.getPrettyString(config.getAllowedOutliers() * 100d)
+ Resources.getMessage("PropertiesView.11")
}); //$NON-NLS-1$ //$NON-NLS-2$
// Utility measure
Property m =
new Property(
Resources.getMessage("PropertiesView.114"),
new String[] {metric.getDescription().getName()}); // $NON-NLS-1$
// Properties of the utility measure
if (metric.getAggregateFunction() != null) {
new Property(
m,
Resources.getMessage("PropertiesView.149"),
new String[] {metric.getAggregateFunction().toString()}); // $NON-NLS-1$
}
if (metric.isGSFactorSupported()) {
new Property(
m,
Resources.getMessage("PropertiesView.151"),
new String[] {
SWTUtil.getPrettyString(metric.getGeneralizationSuppressionFactor())
}); //$NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.152"),
new String[] {SWTUtil.getPrettyString(metric.getGeneralizationFactor())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.153"),
new String[] {SWTUtil.getPrettyString(metric.getSuppressionFactor())}); // $NON-NLS-1$
}
new Property(
m,
Resources.getMessage("PropertiesView.155"),
new String[] {SWTUtil.getPrettyString(metric.isMonotonic())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.156"),
new String[] {SWTUtil.getPrettyString(metric.isWeighted())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.157"),
new String[] {SWTUtil.getPrettyString(metric.isPrecomputed())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.158"),
new String[] {
SWTUtil.getPrettyString(metric.isAbleToHandleMicroaggregation())
}); //$NON-NLS-1$
// Financial configuration
if (metric instanceof MetricSDNMPublisherPayout) {
// Obtain for output data
ARXFinancialConfiguration financial =
((MetricSDNMPublisherPayout) metric).getFinancialConfiguration();
// Obtain for input only. This is a bit ugly.
if (financial == null) {
financial = ARXFinancialConfiguration.create();
financial
.setAdversaryCost(config.getAdversaryCost())
.setAdversaryGain(config.getAdversaryGain())
.setPublisherBenefit(config.getPublisherBenefit())
.setPublisherLoss(config.getPublisherLoss());
}
// Render
new Property(
m,
Resources.getMessage("PropertiesView.135"),
new String[] {SWTUtil.getPrettyString(financial.getPublisherBenefit())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.136"),
new String[] {SWTUtil.getPrettyString(financial.getPublisherLoss())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.137"),
new String[] {SWTUtil.getPrettyString(financial.getAdversaryGain())}); // $NON-NLS-1$
new Property(
m,
Resources.getMessage("PropertiesView.138"),
new String[] {SWTUtil.getPrettyString(financial.getAdversaryCost())}); // $NON-NLS-1$
if (((MetricSDNMPublisherPayout) metric).isProsecutorAttackerModel()) {
new Property(
m,
Resources.getMessage("PropertiesView.139"),
new String[] {Resources.getMessage("PropertiesView.160")}); // $NON-NLS-1$ //$NON-NLS-2$
}
if (((MetricSDNMPublisherPayout) metric).isJournalistAttackerModel()) {
new Property(
m,
Resources.getMessage("PropertiesView.139"),
new String[] {Resources.getMessage("PropertiesView.161")}); // $NON-NLS-1$ //$NON-NLS-2$
}
}
// Attributes
final Property attributes =
new Property(
Resources.getMessage("PropertiesView.12"),
new String[] {String.valueOf(data.getNumColumns())}); // $NON-NLS-1$
// Print identifying attributes
final Property identifying =
new Property(
attributes,
Resources.getMessage("PropertiesView.13"),
new String[] {
String.valueOf(definition.getIdentifyingAttributes().size())
}); //$NON-NLS-1$
int index = 0;
for (int i = 0; i < data.getNumColumns(); i++) {
final String s = data.getAttributeName(i);
if (definition.getIdentifyingAttributes().contains(s)) {
final String[] values =
new String[] {
"", "", "", "", ""
}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$
values[0] = s;
values[1] = definition.getDataType(s).toString();
new Property(
identifying,
Resources.getMessage("PropertiesView.19") + (index++),
values); //$NON-NLS-1$
}
}
// Print quasi-identifying attributes
final Property quasiIdentifying =
new Property(
attributes,
Resources.getMessage("PropertiesView.20"),
new String[] {
String.valueOf(definition.getQuasiIdentifyingAttributes().size())
}); //$NON-NLS-1$
index = 0;
for (int i = 0; i < data.getNumColumns(); i++) {
final String s = data.getAttributeName(i);
if (definition.getQuasiIdentifyingAttributes().contains(s)) {
final String[] values =
new String[] {
"", "", "", "", "", "", "", ""
}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$ //$NON-NLS-6$
// //$NON-NLS-7$ //$NON-NLS-8$
values[0] = s;
if (definition.getHierarchy(s) != null) {
DataType<?> type = definition.getDataType(s);
values[1] = type.getDescription().getLabel();
if (type.getDescription().hasFormat()
&& ((DataTypeWithFormat) type).getFormat() != null) {
values[2] = ((DataTypeWithFormat) type).getFormat();
}
// Determine height of hierarchy
int height = 0;
String[][] hierarchy = definition.getHierarchy(s);
if (hierarchy != null && hierarchy.length != 0 && hierarchy[0] != null) {
height = hierarchy[0].length;
}
values[3] = String.valueOf(height);
values[4] = String.valueOf(definition.getMinimumGeneralization(s));
values[5] = String.valueOf(definition.getMaximumGeneralization(s));
}
if (definition.getMicroAggregationFunction(s) != null) {
values[7] = definition.getMicroAggregationFunction(s).getLabel();
}
values[6] = SWTUtil.getPrettyString(config.getAttributeWeight(s));
new Property(
quasiIdentifying,
Resources.getMessage("PropertiesView.26") + (index++),
values); //$NON-NLS-1$
}
}
// Print sensitive attributes
final Property sensitive =
new Property(
attributes,
Resources.getMessage("PropertiesView.27"),
new String[] {
String.valueOf(definition.getSensitiveAttributes().size())
}); //$NON-NLS-1$
index = 0;
for (int i = 0; i < data.getNumColumns(); i++) {
final String s = data.getAttributeName(i);
if (definition.getSensitiveAttributes().contains(s)) {
final String[] values =
new String[] {
"", "", "", "", ""
}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$
values[0] = s;
if (config.getHierarchy(s) != null && config.getHierarchy(s).getHierarchy() != null) {
int height = 0;
if (config.getHierarchy(s).getHierarchy().length > 0) {
height = config.getHierarchy(s).getHierarchy()[0].length;
}
values[1] = definition.getDataType(s).toString();
values[2] = String.valueOf(height);
}
new Property(
sensitive,
Resources.getMessage("PropertiesView.33") + (index++),
values); //$NON-NLS-1$
}
}
// Print insensitive attributes
final Property insensitive =
new Property(
attributes,
Resources.getMessage("PropertiesView.34"),
new String[] {
String.valueOf(definition.getInsensitiveAttributes().size())
}); //$NON-NLS-1$
index = 0;
for (int i = 0; i < data.getNumColumns(); i++) {
final String s = data.getAttributeName(i);
if (definition.getInsensitiveAttributes().contains(s)) {
final String[] values =
new String[] {
"", "", "", "", ""
}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$
values[0] = s;
values[1] = definition.getDataType(s).toString();
new Property(
insensitive,
Resources.getMessage("PropertiesView.40") + (index++),
values); //$NON-NLS-1$
}
}
// Refresh and initialize
refresh();
// Redraw
root.setRedraw(true);
}
/**
* Returns the test cases.
*
* @return
*/
@Parameters(name = "{index}:[{0}]")
public static Collection<Object[]> cases() {
return Arrays.asList(
new Object[][] {
/* 0 */ {
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(5)),
"./data/adult.csv",
255559.85455731067,
new int[] {1, 0, 1, 1, 3, 2, 2, 0, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(100)),
"./data/adult.csv",
379417.3460570988,
new int[] {1, 1, 1, 1, 3, 2, 2, 1, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(5)),
"./data/adult.csv",
407289.5388925293,
new int[] {1, 2, 1, 1, 3, 2, 2, 1, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(100)),
"./data/adult.csv",
453196.8932458743,
new int[] {0, 4, 1, 1, 3, 2, 2, 1, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(5)),
"./data/adult.csv",
255559.85455731067,
new int[] {1, 0, 1, 1, 3, 2, 2, 0, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new KAnonymity(100)),
"./data/adult.csv",
379417.3460570988,
new int[] {1, 1, 1, 1, 3, 2, 2, 1, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(5)),
"./data/cup.csv",
1764006.4033760305,
new int[] {2, 4, 0, 1, 0, 4, 4, 4},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(100)),
"./data/cup.csv",
1994002.8308631124,
new int[] {3, 4, 1, 1, 0, 4, 4, 4},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(5)),
"./data/cup.csv",
2445878.424834677,
new int[] {4, 4, 1, 1, 1, 4, 4, 4},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(100)),
"./data/cup.csv",
2517471.5816586106,
new int[] {5, 4, 1, 0, 1, 4, 4, 4},
false)
},
/* 10 */ {
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(5)),
"./data/cup.csv",
1764006.4033760305,
new int[] {2, 4, 0, 1, 0, 4, 4, 4},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(new KAnonymity(100)),
"./data/cup.csv",
2001343.4737485605,
new int[] {3, 4, 1, 1, 0, 1, 2, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(5)),
"./data/fars.csv",
4469271.0,
new int[] {0, 2, 2, 2, 1, 2, 1, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(100)),
"./data/fars.csv",
5.6052481E7,
new int[] {0, 2, 3, 3, 1, 2, 2, 2},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(5)),
"./data/fars.csv",
1.42377891E8,
new int[] {1, 2, 3, 3, 1, 2, 1, 2},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(100)),
"./data/fars.csv",
4.36925397E8,
new int[] {5, 2, 3, 3, 1, 2, 0, 2},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(5)),
"./data/fars.csv",
4469271.0,
new int[] {0, 2, 2, 2, 1, 2, 1, 0},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new KAnonymity(100)),
"./data/fars.csv",
5.6052481E7,
new int[] {0, 2, 3, 3, 1, 2, 2, 2},
true)
},
});
}
/**
* Returns the test cases.
*
* @return
*/
@Parameters(name = "{index}:[{0}]")
public static Collection<Object[]> cases() {
return Arrays.asList(
new Object[][] {
/* 0 */
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 5, EntropyEstimator.GRASSBERGER)),
"occupation",
"./data/adult.csv",
216092.124036387,
new int[] {1, 0, 1, 0, 3, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 100, EntropyEstimator.SHANNON)),
"occupation",
"./data/adult.csv",
0.0d,
null,
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 5, EntropyEstimator.GRASSBERGER)),
"occupation",
"./data/adult.csv",
324620.5269918692,
new int[] {1, 1, 1, 1, 3, 2, 2, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.05d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 3, EntropyEstimator.GRASSBERGER)),
"occupation",
"./data/adult.csv",
180347.4325366015,
new int[] {0, 0, 1, 1, 2, 2, 2, 0},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 5, EntropyEstimator.SHANNON)),
"occupation",
"./data/adult.csv",
228878.2039109517,
new int[] {1, 0, 1, 1, 2, 2, 2, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.1d, Metric.createPrecomputedEntropyMetric(0.1d, false))
.addPrivacyModel(
new EntropyLDiversity("occupation", 100, EntropyEstimator.GRASSBERGER)),
"occupation",
"./data/adult.csv",
0.0d,
null,
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(
new EntropyLDiversity("RAMNTALL", 5, EntropyEstimator.GRASSBERGER)),
"RAMNTALL",
"./data/cup.csv",
1833435.0,
new int[] {4, 0, 1, 0, 1, 3, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.03d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(
new EntropyLDiversity("RAMNTALL", 100, EntropyEstimator.GRASSBERGER)),
"RAMNTALL",
"./data/cup.csv",
4.5168281E7,
new int[] {4, 4, 0, 0, 1, 3, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new EntropyLDiversity("RAMNTALL", 5)),
"RAMNTALL",
"./data/cup.csv",
3.01506905E8,
new int[] {4, 4, 1, 1, 1, 4, 4},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(new EntropyLDiversity("RAMNTALL", 3)),
"RAMNTALL",
"./data/cup.csv",
9.2264547E7,
new int[] {4, 4, 1, 0, 1, 4, 4},
false)
},
/* 10 */ {
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(
new EntropyLDiversity("RAMNTALL", 5, EntropyEstimator.SHANNON)),
"RAMNTALL",
"./data/cup.csv",
2823649.0,
new int[] {4, 0, 0, 1, 1, 3, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.1d, Metric.createDiscernabilityMetric(true))
.addPrivacyModel(
new EntropyLDiversity("RAMNTALL", 100, EntropyEstimator.GRASSBERGER)),
"RAMNTALL",
"./data/cup.csv",
3.4459973E7,
new int[] {5, 0, 0, 2, 1, 2, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new EntropyLDiversity("EDUC", 5, EntropyEstimator.GRASSBERGER)),
"EDUC",
"./data/ihis.csv",
7735322.29514608,
new int[] {0, 0, 0, 1, 3, 0, 0, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new EntropyLDiversity("EDUC", 2, EntropyEstimator.GRASSBERGER)),
"EDUC",
"./data/ihis.csv",
5428093.534997522,
new int[] {0, 0, 0, 0, 2, 0, 0, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new EntropyLDiversity("EDUC", 5, EntropyEstimator.SHANNON)),
"EDUC",
"./data/ihis.csv",
1.2258628558792587E7,
new int[] {0, 0, 0, 3, 3, 2, 0, 1},
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.0d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new EntropyLDiversity("EDUC", 100, EntropyEstimator.GRASSBERGER)),
"EDUC",
"./data/ihis.csv",
0.0d,
null,
false)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.04d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(
new EntropyLDiversity("EDUC", 5, EntropyEstimator.GRASSBERGER)),
"EDUC",
"./data/ihis.csv",
7735322.29514608,
new int[] {0, 0, 0, 1, 3, 0, 0, 1},
true)
},
{
new ARXAnonymizationTestCase(
ARXConfiguration.create(0.02d, Metric.createPrecomputedEntropyMetric(0.1d, true))
.addPrivacyModel(new EntropyLDiversity("EDUC", 3, EntropyEstimator.SHANNON)),
"EDUC",
"./data/ihis.csv",
7578152.206004559,
new int[] {0, 0, 0, 2, 2, 0, 0, 1},
true)
},
});
}
/**
* Entry point.
*
* @param args the arguments
* @throws IOException
*/
public static void main(String[] args) throws IOException {
// Define data
DefaultData data = Data.create();
data.add("zipcode", "age", "disease");
data.add("47677", "29", "gastric ulcer");
data.add("47602", "22", "gastritis");
data.add("47678", "27", "stomach cancer");
data.add("47905", "43", "gastritis");
data.add("47909", "52", "flu");
data.add("47906", "47", "bronchitis");
data.add("47605", "30", "bronchitis");
data.add("47673", "36", "pneumonia");
data.add("47607", "32", "stomach cancer");
// Define hierarchies
DefaultHierarchy age = Hierarchy.create();
age.add("29", "<=40", "*");
age.add("22", "<=40", "*");
age.add("27", "<=40", "*");
age.add("43", ">40", "*");
age.add("52", ">40", "*");
age.add("47", ">40", "*");
age.add("30", "<=40", "*");
age.add("36", "<=40", "*");
age.add("32", "<=40", "*");
// Only excerpts for readability
DefaultHierarchy zipcode = Hierarchy.create();
zipcode.add("47677", "4767*", "476**", "47***", "4****", "*****");
zipcode.add("47602", "4760*", "476**", "47***", "4****", "*****");
zipcode.add("47678", "4767*", "476**", "47***", "4****", "*****");
zipcode.add("47905", "4790*", "479**", "47***", "4****", "*****");
zipcode.add("47909", "4790*", "479**", "47***", "4****", "*****");
zipcode.add("47906", "4790*", "479**", "47***", "4****", "*****");
zipcode.add("47605", "4760*", "476**", "47***", "4****", "*****");
zipcode.add("47673", "4767*", "476**", "47***", "4****", "*****");
zipcode.add("47607", "4760*", "476**", "47***", "4****", "*****");
// Define sensitive value hierarchy
DefaultHierarchy disease = Hierarchy.create();
disease.add(
"flu",
"respiratory infection",
"vascular lung disease",
"respiratory & digestive system disease");
disease.add(
"pneumonia",
"respiratory infection",
"vascular lung disease",
"respiratory & digestive system disease");
disease.add(
"bronchitis",
"respiratory infection",
"vascular lung disease",
"respiratory & digestive system disease");
disease.add(
"pulmonary edema",
"vascular lung disease",
"vascular lung disease",
"respiratory & digestive system disease");
disease.add(
"pulmonary embolism",
"vascular lung disease",
"vascular lung disease",
"respiratory & digestive system disease");
disease.add(
"gastric ulcer",
"stomach disease",
"digestive system disease",
"respiratory & digestive system disease");
disease.add(
"stomach cancer",
"stomach disease",
"digestive system disease",
"respiratory & digestive system disease");
disease.add(
"gastritis",
"stomach disease",
"digestive system disease",
"respiratory & digestive system disease");
disease.add(
"colitis",
"colon disease",
"digestive system disease",
"respiratory & digestive system disease");
disease.add(
"colon cancer",
"colon disease",
"digestive system disease",
"respiratory & digestive system disease");
data.getDefinition().setAttributeType("age", age);
data.getDefinition().setAttributeType("zipcode", zipcode);
data.getDefinition().setAttributeType("disease", AttributeType.SENSITIVE_ATTRIBUTE);
// Create an instance of the anonymizer
ARXAnonymizer anonymizer = new ARXAnonymizer();
ARXConfiguration config = ARXConfiguration.create();
config.addCriterion(new KAnonymity(3));
config.addCriterion(new HierarchicalDistanceTCloseness("disease", 0.6d, disease));
config.setMaxOutliers(0d);
config.setMetric(Metric.createEntropyMetric());
// Now anonymize
ARXResult result = anonymizer.anonymize(data, config);
// Print info
printResult(result, data);
// Process results
System.out.println(" - Transformed data:");
Iterator<String[]> transformed = result.getOutput(false).iterator();
while (transformed.hasNext()) {
System.out.print(" ");
System.out.println(Arrays.toString(transformed.next()));
}
}
