eus.ixa.ixa.pipe.ml.sequence

Class SequenceLabelerEvaluator

java.lang.Object

opennlp.tools.util.eval.Evaluator<SequenceLabelSample>

eus.ixa.ixa.pipe.ml.sequence.SequenceLabelerEvaluator

public class SequenceLabelerEvaluator
extends opennlp.tools.util.eval.Evaluator<SequenceLabelSample>

The SequenceLabelerEvaluator measures the F1 performance of the given SequenceLabeler with the provided reference SequenceLabelSamples. Based on Apache OpenNLP evaluators.

See Also:

Evaluator, SequenceLabeler, SequenceLabelSample

Constructor Summary

Constructors

Constructor and Description
SequenceLabelerEvaluator(opennlp.tools.util.ObjectStream<SequenceLabelSample> trainSamples, String aCorpusFormat, SequenceLabeler seqLabeler, SequenceLabelerEvaluationMonitor... listeners)
SequenceLabelerEvaluator(String aCorpusFormat, SequenceLabeler seqLabeler, SequenceLabelerEvaluationMonitor... listeners)

Method Summary

Modifier and Type	Method and Description
opennlp.tools.util.eval.FMeasure	getFMeasure()
double	getKnownAccuracy()
double	getSentenceAccuracy()
double	getUnknownWordAccuracy()
double	getWordAccuracy() Retrieves the word accuracy.
long	getWordCount() Retrieves the total number of words considered in the evaluation.
protected SequenceLabelSample	processSample(SequenceLabelSample reference) Evaluates the given reference SequenceLabelSample object.
void	updateAccuracyScores(String[] referenceTokens, Object[] references, Object[] predictions)

Methods inherited from class opennlp.tools.util.eval.Evaluator

evaluate, evaluateSample

Methods inherited from class java.lang.Object

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

Constructor Detail

SequenceLabelerEvaluator

public SequenceLabelerEvaluator(String aCorpusFormat,
                                SequenceLabeler seqLabeler,
                                SequenceLabelerEvaluationMonitor... listeners)

SequenceLabelerEvaluator

public SequenceLabelerEvaluator(opennlp.tools.util.ObjectStream<SequenceLabelSample> trainSamples,
                                String aCorpusFormat,
                                SequenceLabeler seqLabeler,
                                SequenceLabelerEvaluationMonitor... listeners)

Method Detail

processSample

protected SequenceLabelSample processSample(SequenceLabelSample reference)

Evaluates the given reference SequenceLabelSample object. This is done by finding the sequences with the SequenceLabeler in the sentence from the reference SequenceLabelSample. The found sequences are then used to calculate and update the scores.

Specified by:

processSample in class opennlp.tools.util.eval.Evaluator<SequenceLabelSample>

Parameters:

reference - the reference SequenceLabelSample.

Returns:

the predicted SequenceLabelSample.

getFMeasure

public opennlp.tools.util.eval.FMeasure getFMeasure()

getWordAccuracy

public double getWordAccuracy()

Retrieves the word accuracy. This is defined as: word accuracy = correctly detected tags / total words

Returns:

the word accuracy

getSentenceAccuracy

public double getSentenceAccuracy()

getUnknownWordAccuracy

public double getUnknownWordAccuracy()

getKnownAccuracy

public double getKnownAccuracy()

getWordCount

public long getWordCount()

Retrieves the total number of words considered in the evaluation.

Returns:

the word count

updateAccuracyScores

public void updateAccuracyScores(String[] referenceTokens,
                                 Object[] references,
                                 Object[] predictions)