de.hpi.fgis.dude.postprocessor
Class StatisticComponent

java.lang.Object
  de.hpi.fgis.dude.postprocessor.StatisticComponent

All Implemented Interfaces:

AutoJsonable

Direct Known Subclasses:

ExtendedStatisticComponent

public class StatisticComponent
extends Object
implements AutoJsonable

StatisticComponent provides functionality for gathering statistics concerning the recall, precision and f-measure. Therefore a collection real duplicates has to be added.

Author:

Matthias Pohl

Field Summary

protected long	actualComparisonCount
protected Algorithm	algorithm
protected boolean	checkMemory
protected Date	endDate
protected long	endTime
protected long	falseNegativesByComparison
protected long	falsePositives
protected long	falsePositivesByComparison
protected GoldStandard	goldStandard
protected long	memoryCheckFrequency
protected static int	NO_STATISTIC_VALUE
protected long	pairCount
protected Date	startDate
protected long	startTime
protected MemoryCheckerTask	task
protected Timer	timer
protected long	trueNegativesByComparison
protected long	truePositives
protected long	truePositivesByComparison

Constructor Summary

protected	StatisticComponent() Internal constructor for Jsonable deserialization.
	StatisticComponent(Algorithm algorithm) Initializes a StatisticComponent with no gold standard.
	StatisticComponent(GoldStandard goldStandard, Algorithm algorithm) Initializes a StatisticComponent using the passed DuDeObjectPairs as real duplicates.

Method Summary

void	addDuplicate(DuDeObjectPair pair) Adds a DuDeObjectPair to the knowledge base that is labeled as a detected duplicate.
void	addDuplicate(DuDeObjectPair pair, boolean actualComparison) Adds a DuDeObjectPair to the knowledge base that is labeled as a detected duplicate.
void	addDuplicate(Iterable<DuDeObjectPair> pairs) Adds several DuDeObjectPairs to the knowledge base that are labeled as detected duplicates.
void	addNonDuplicate(DuDeObjectPair pair) Adds a DuDeObjectPair to the knowledge base that is labeled as a detected non-duplicate.
void	addNonDuplicate(DuDeObjectPair pair, boolean actualComparison) Adds a DuDeObjectPair to the knowledge base that is labeled as a detected non-duplicate.
void	addNonDuplicate(Iterable<DuDeObjectPair> pairs) Adds several DuDeObjectPairs to the knowledge base that are labeled as a detected non-duplicates.
void	addPair(DuDeObjectPair pair, boolean positive) Adds a DuDeObjectPair to the knowledge base.
void	addPair(Iterable<DuDeObjectPair> pairs, boolean positive) Adds several DuDeObjectPairs to the knowledge base.
protected void	checkMemoryUsage() Starts Memoryusage task
String	getAverageMemoryUsed() Gets the registered average amount of memory used during the experiment.
long	getComparisonCount() Returns the number of pairs that were already compared.
Date	getEndDate() Gets the date of the specified end time of an algorithm.
long	getFalseNegatives() Returns the false negatives count.
long	getFalseNegativesByComparison() Returns the false negatives count that are explicitly classified by the comparator.
long	getFalsePositives() Returns the false positives count.
long	getFalsePositivesByComparison() Returns the false positives count that are explicitly classified by the comparator..
double	getFMeasure() Returns the f-measure based on the current knowledge base.
double	getFMeasureByComparison() Returns the f-measure based on the current knowledge base and the actual comparisons.
String	getMaximumMemoryUsed() Gets the registered maximum amount of memory during the experiment.
long	getMemoryCheckFrequency() Gets the frequency of memory checks.
String	getMinimumMemoryUsed() Gets the registered minimum amount of memory during the experiment.
long	getNumberOfCandidateComparisons() Returns the maximum number of pairs that would be generated by the naive approach.
long	getNumberOfRealDuplicates() Returns the size of the gold standard.
long	getObjectCount() Returns the number of records that were processed by the algorithm.
long	getPairCount() Returns the number of pairs that were already considered.
double	getPrecision() Returns the precision based on the current knowledge base.
double	getPrecisionByComparison() Returns the precision based on the current knowledge base and the actual comparisons.
double	getRecall() Returns the recall based on the current knowledge base.
double	getRecallByComparison() Returns the recall based on the current knowledge base and the actual comparisons.
double	getReductionRatio() Returns the reduction ratio based on the current knowledge base.
double	getReductionRatioByComparison() Returns the reduction ratio based on the current knowledge base and the actual comparisons.
long	getRuntime() Gets the time difference between beginning time and finishing time.
Date	getStartDate() Gets the date of the specified start time of an algorithm.
long	getTrueNegatives() Returns the true negatives count.
long	getTrueNegativesByComparison() Returns the true negatives count that are explicitly classified by the comparator.
long	getTruePositives() Returns the true positives count.
long	getTruePositivesByComparison() Returns the true positives count that are explicitly classified by the comparator..
boolean	goldStandardSet() Checks whether a gold standard was passed.
boolean	hasGMD() Checks whether this StatisticComponent calculates the Generalized Merge Distance.
boolean	isCheckMemory() Gets the boolean flag that indicates the activation status of memory checking.
boolean	isDuplicate(DuDeObjectPair pair) Returns true if the DuDeObjectPair exists in the set of real duplicate pairs.
boolean	isNonDuplicate(DuDeObjectPair pair) Checks whether a specific pair exists in the set of real duplicate pairs.
void	reset() Sets the attributes for TruePositives, FalsePositives, TruePositivesByComparison, FalsePositivesByComparison, TrueNegativesByComparison, FalseNegativesByComparison, PairCount and ComparisonCount to 0.
void	setBeginningTime() Deprecated. Replaced by setStartTime()
void	setCheckMemory(boolean checkMemory) Sets the boolean flag that indicates the activation status of memory checking.
void	setEndTime() Sets current time as finishing time for the runtime.
void	setFinishingTime() Deprecated. Replaced by setEndTime()
void	setMemoryCheckFrequency(long memoryCheckFrequency) Gets the frequency of memory checks.
void	setStartTime() Sets the current time as starting time for the runtime and initiates memory monitoring.

Methods inherited from class java.lang.Object

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

Field Detail

NO_STATISTIC_VALUE

protected static final int NO_STATISTIC_VALUE

See Also:

Constant Field Values

goldStandard

protected GoldStandard goldStandard

truePositives

protected long truePositives

falsePositives

protected long falsePositives

truePositivesByComparison

protected long truePositivesByComparison

falsePositivesByComparison

protected long falsePositivesByComparison

trueNegativesByComparison

protected long trueNegativesByComparison

falseNegativesByComparison

protected long falseNegativesByComparison

pairCount

protected long pairCount

actualComparisonCount

protected long actualComparisonCount

startTime

protected long startTime

startDate

protected Date startDate

endTime

protected long endTime

endDate

protected Date endDate

algorithm

protected Algorithm algorithm

checkMemory

protected boolean checkMemory

task

protected transient MemoryCheckerTask task

timer

protected transient Timer timer

memoryCheckFrequency

protected long memoryCheckFrequency

Constructor Detail

StatisticComponent

public StatisticComponent(Algorithm algorithm)

Initializes a StatisticComponent with no gold standard.

Parameters:

algorithm - The used algorithm.

StatisticComponent

public StatisticComponent(GoldStandard goldStandard,
                          Algorithm algorithm)

Initializes a StatisticComponent using the passed DuDeObjectPairs as real duplicates.

Parameters:

goldStandard - The gold standard which these statistics are based on.

algorithm - Used algorithm.

StatisticComponent

protected StatisticComponent()

Internal constructor for Jsonable deserialization.

Method Detail

hasGMD

public boolean hasGMD()

Checks whether this StatisticComponent calculates the Generalized Merge Distance. Important for the output components.

Returns:

false since this is the original component without GMD computation

goldStandardSet

public boolean goldStandardSet()

Checks whether a gold standard was passed. If no gold standard was set, some of the statistics cannot be calculated.

Returns:

true, if a gold standard was set; otherwise false.

getTruePositives

public long getTruePositives()

Returns the true positives count.

Returns:

The true positives count.

getFalsePositives

public long getFalsePositives()

Returns the false positives count.

Returns:

The false positives count.

getTrueNegatives

public long getTrueNegatives()

Returns the true negatives count.

Returns:

The true negatives count.

getFalseNegatives

public long getFalseNegatives()

Returns the false negatives count.

Returns:

The false negatives count.

getTruePositivesByComparison

public long getTruePositivesByComparison()

Returns the true positives count that are explicitly classified by the comparator..

Returns:

The true positives count that are explicitly classified by the comparator..

getFalsePositivesByComparison

public long getFalsePositivesByComparison()

Returns the false positives count that are explicitly classified by the comparator..

Returns:

The false positives count that are explicitly classified by the comparator..

getTrueNegativesByComparison

public long getTrueNegativesByComparison()

Returns the true negatives count that are explicitly classified by the comparator.

Returns:

The true negatives count that are explicitly classified by the comparator.

getFalseNegativesByComparison

public long getFalseNegativesByComparison()

Returns the false negatives count that are explicitly classified by the comparator.

Returns:

The false negatives count that are explicitly classified by the comparator.

getPairCount

public long getPairCount()

Returns the number of pairs that were already considered. These are all pairs, including those that should not be counted as comparison (e.g. those created by building the transitive closure).

Returns:

The number of pairs that were already considered.

getComparisonCount

public long getComparisonCount()

Returns the number of pairs that were already compared. These are only those pairs that should be counted as comparison (e.g. NOT those created by building the transitive closure).

Returns:

The number of pairs that were already compared.

addPair

public void addPair(DuDeObjectPair pair,
                    boolean positive)

Adds a DuDeObjectPair to the knowledge base. The pair is counted as comparison.

Parameters:

pair - The pair that shall be considered in the statistics.

positive - true, if the passed pair was detected as an duplicate; otherwise false.

addPair

public void addPair(Iterable<DuDeObjectPair> pairs,
                    boolean positive)

Adds several DuDeObjectPairs to the knowledge base. The pairs are counted as comparisons.

Parameters:

pairs - The pairs that shall be considered in the statistics.

positive - true, if the passed pair was detected as an duplicate; otherwise false.

addDuplicate

public void addDuplicate(DuDeObjectPair pair)

Adds a DuDeObjectPair to the knowledge base that is labeled as a detected duplicate. The pair is counted as comparison.

Parameters:

pair - A detected duplicate.

addDuplicate

public void addDuplicate(DuDeObjectPair pair,
                         boolean actualComparison)

Adds a DuDeObjectPair to the knowledge base that is labeled as a detected duplicate.

Parameters:

pair - A detected duplicate.

actualComparison - true, if the pair should be counted as comparison; otherwise false.

addDuplicate

public void addDuplicate(Iterable<DuDeObjectPair> pairs)

Adds several DuDeObjectPairs to the knowledge base that are labeled as detected duplicates. The pairs are counted as comparisons.

Parameters:

pairs - The pairs that shall be considered as detected duplicates in the statistics.

addNonDuplicate

public void addNonDuplicate(DuDeObjectPair pair)

Adds a DuDeObjectPair to the knowledge base that is labeled as a detected non-duplicate. The pair is counted as comparison.

Parameters:

pair - The pair that shall be considered as a detected non-duplicate in the statistics.

addNonDuplicate

public void addNonDuplicate(DuDeObjectPair pair,
                            boolean actualComparison)

Adds a DuDeObjectPair to the knowledge base that is labeled as a detected non-duplicate.

Parameters:

pair - The pair that shall be considered as a detected non-duplicate in the statistics.

actualComparison - true, if the pair is an actual comparison; otherwise false.

addNonDuplicate

public void addNonDuplicate(Iterable<DuDeObjectPair> pairs)

Adds several DuDeObjectPairs to the knowledge base that are labeled as a detected non-duplicates. The pairs are counted as comparisons.

Parameters:

pairs - The pairs that shall be considered as detected non-duplicates in the statistics.

setBeginningTime

@Deprecated
public void setBeginningTime()

Deprecated. Replaced by setStartTime()

Sets the current time as starting time for the runtime and initiates memory monitoring.

setStartTime

public void setStartTime()

Sets the current time as starting time for the runtime and initiates memory monitoring.

getStartDate

public Date getStartDate()

Gets the date of the specified start time of an algorithm.

Returns:

Date object

setFinishingTime

@Deprecated
public void setFinishingTime()

Deprecated. Replaced by setEndTime()

Sets current time as finishing time for the runtime. Memory monitoring ends if it is activated.

setEndTime

public void setEndTime()

Sets current time as finishing time for the runtime. Memory monitoring ends if it is activated.

getEndDate

public Date getEndDate()

Gets the date of the specified end time of an algorithm.

Returns:

Date object

getRuntime

public long getRuntime()

Gets the time difference between beginning time and finishing time.

Returns:

Runtime in milliseconds

getPrecision

public double getPrecision()

Returns the precision based on the current knowledge base.

Returns:

The precision value.

getRecall

public double getRecall()

Returns the recall based on the current knowledge base. In the context of blocking algorithms the returned value can be considered as pair completeness ratio.

Returns:

The recall value.

getFMeasure

public double getFMeasure()

Returns the f-measure based on the current knowledge base.

Returns:

The f-measure.

getReductionRatio

public double getReductionRatio()

Returns the reduction ratio based on the current knowledge base.

Returns:

The reduction ratio.

getPrecisionByComparison

public double getPrecisionByComparison()

Returns the precision based on the current knowledge base and the actual comparisons.

Returns:

The precision value.

getRecallByComparison

public double getRecallByComparison()

Returns the recall based on the current knowledge base and the actual comparisons. In the context of blocking algorithms the returned value can be considered as pair completeness ratio.

Returns:

The recall value.

getFMeasureByComparison

public double getFMeasureByComparison()

Returns the f-measure based on the current knowledge base and the actual comparisons.

Returns:

The f-measure.

getReductionRatioByComparison

public double getReductionRatioByComparison()

Returns the reduction ratio based on the current knowledge base and the actual comparisons.

Returns:

The reduction ratio.

getNumberOfRealDuplicates

public long getNumberOfRealDuplicates()

Returns the size of the gold standard.

Returns:

The gold standard's size.

getObjectCount

public long getObjectCount()

Returns the number of records that were processed by the algorithm.

Returns:

The record count of the source data set.

getNumberOfCandidateComparisons

public long getNumberOfCandidateComparisons()

Returns the maximum number of pairs that would be generated by the naive approach.

Returns:

The number of pairs that would be generated by the naive approach.

isDuplicate

public boolean isDuplicate(DuDeObjectPair pair)

Returns true if the DuDeObjectPair exists in the set of real duplicate pairs.

Parameters:

pair - The duplicate pair that is to be checked.

Returns:

true, if the duplicate pair exists in the set of real duplicate pairs.

isNonDuplicate

public boolean isNonDuplicate(DuDeObjectPair pair)

Checks whether a specific pair exists in the set of real duplicate pairs.

Parameters:

pair - The duplicate pair that is to be checked.

Returns:

false, if the duplicate pair exists in the set of real duplicate pairs.

getMemoryCheckFrequency

public long getMemoryCheckFrequency()

Gets the frequency of memory checks.

Returns:

Frequency of memory checks in ms.

setMemoryCheckFrequency

public void setMemoryCheckFrequency(long memoryCheckFrequency)

Gets the frequency of memory checks.

Parameters:

memoryCheckFrequency - Frequency of memory checks in ms. Default value is 5000 ms.

isCheckMemory

public boolean isCheckMemory()

Gets the boolean flag that indicates the activation status of memory checking.

Returns:

Boolean flag is TRUE if memory checking is performed.

setCheckMemory

public void setCheckMemory(boolean checkMemory)

Sets the boolean flag that indicates the activation status of memory checking.

Parameters:

checkMemory - Is set to False if memory checking should not be performed. Default value is True.

checkMemoryUsage

protected void checkMemoryUsage()

Starts Memoryusage task

getMaximumMemoryUsed

public String getMaximumMemoryUsed()

Gets the registered maximum amount of memory during the experiment.

Returns:

Maximum amount of memory in KB.

getMinimumMemoryUsed

public String getMinimumMemoryUsed()

Gets the registered minimum amount of memory during the experiment.

Returns:

Minimum amount of memory in KB.

getAverageMemoryUsed

public String getAverageMemoryUsed()

Gets the registered average amount of memory used during the experiment.

Returns:

Average amount of memory in KB.

reset

public void reset()

Sets the attributes for TruePositives, FalsePositives, TruePositivesByComparison, FalsePositivesByComparison, TrueNegativesByComparison, FalseNegativesByComparison, PairCount and ComparisonCount to 0. The algorithm and the gold standard are not changed. This allows to reuse the statistic component for several experiments. Please note that the start and end time for the next experiment has to be set again by using the corresponding methods.