Examples of org.apache.stanbol.enhancer.topic.ClassificationReport

Package org.apache.stanbol.enhancer.topic

Examples of org.apache.stanbol.enhancer.topic.ClassificationReport

org.apache.stanbol.enhancer.topic.ClassificationReport
kipedia.org/wiki/Precision_and_recall

Precision, Recall are F1-score and preferred over a simple rate of good classification so as to account for potentially imbalanced evaluation set (e.g. when the number of negative examples is much larger than the number of positive examples).

        assertEquals(0, classifier.updateModel(true));
    }


    @Test
    public void testUpdatePerformanceEstimates() throws Exception {
        ClassificationReport performanceEstimates;
        // no registered topic
        try {
            classifier.getPerformanceEstimates("urn:t/001");
            fail("Should have raised ClassifierException");
        } catch (ClassifierException e) {

View Full Code Here


    @Test
    public void testCrossValidation() throws Exception {
        // seed a pseudo random number generator for reproducible tests
        Random rng = new Random(0);
        ClassificationReport performanceEstimates;


        // build an artificial data set used for training models and evaluation
        int numberOfTopics = 10;
        int numberOfDocuments = 100;
        int vocabSizeMin = 20;

View Full Code Here

                metadata.add(new TripleImpl(enhancement,
                        org.apache.stanbol.enhancer.servicesapi.rdf.Properties.ENHANCER_CONFIDENCE, lf
                                .createTypedLiteral(Double.valueOf(topic.score))));


                // add performance estimates of the classifier if available
                ClassificationReport perf = getPerformanceEstimates(topic.conceptUri);
                if (perf.uptodate) {
                    metadata.add(new TripleImpl(enhancement, precision, lf.createTypedLiteral(Double
                            .valueOf(perf.precision))));
                    metadata.add(new TripleImpl(enhancement, recall, lf.createTypedLiteral(Double
                            .valueOf(perf.recall))));

View Full Code Here

            Float recall = computeMeanValue(metadata, recallField);
            int positiveSupport = computeSumValue(metadata, positiveSupportField);
            int negativeSupport = computeSumValue(metadata, negativeSupportField);
            Date evaluationDate = (Date) metadata.getFirstValue(modelEvaluationDateField);
            boolean uptodate = evaluationDate != null;
            ClassificationReport report = new ClassificationReport(precision, recall, positiveSupport,
                    negativeSupport, uptodate, evaluationDate);
            if (metadata.getFieldValues(falsePositivesField) == null) {
                metadata.setField(falsePositivesField, new ArrayList<Object>());
            }
            for (Object falsePositiveId : metadata.getFieldValues(falsePositivesField)) {

View Full Code Here

        assertEquals(0, classifier.updateModel(true));
    }


    @Test
    public void testUpdatePerformanceEstimates() throws Exception {
        ClassificationReport performanceEstimates;
        // no registered topic
        try {
            classifier.getPerformanceEstimates("urn:t/001");
            fail("Should have raised ClassifierException");
        } catch (ClassifierException e) {

View Full Code Here


    @Test
    public void testCrossValidation() throws Exception {
        // seed a pseudo random number generator for reproducible tests
        Random rng = new Random(0);
        ClassificationReport performanceEstimates;


        // build an artificial data set used for training models and evaluation
        int numberOfTopics = 10;
        int numberOfDocuments = 100;
        int vocabSizeMin = 20;

View Full Code Here

            Float recall = computeMeanValue(metadata, recallField);
            int positiveSupport = computeSumValue(metadata, positiveSupportField);
            int negativeSupport = computeSumValue(metadata, negativeSupportField);
            Date evaluationDate = (Date) metadata.getFirstValue(modelEvaluationDateField);
            boolean uptodate = evaluationDate != null;
            ClassificationReport report = new ClassificationReport(precision, recall, positiveSupport,
                    negativeSupport, uptodate, evaluationDate);
            if (metadata.getFieldValues(falsePositivesField) == null) {
                metadata.setField(falsePositivesField, new ArrayList<Object>());
            }
            for (Object falsePositiveId : metadata.getFieldValues(falsePositivesField)) {

View Full Code Here

TOP

Related Classes of org.apache.stanbol.enhancer.topic.ClassificationReport

org.apache.stanbol.enhancer.engine.topic.TopicClassificationEngine

org.apache.stanbol.enhancer.engine.topic.TopicEngineTest

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.