Examples with Annotation gate.Annotation used on opensource projects

Example 1 with Annotation

use of gate.Annotation in project gate-core by GateNLP.

the class CorefEditor method annotationRemoved.

@Override
public void annotationRemoved(AnnotationSetEvent ase) {
    Annotation delAnnot = ase.getAnnotation();
    Integer id = delAnnot.getId();
    Object matchesMapObject = document.getFeatures().get(ANNIEConstants.DOCUMENT_COREF_FEATURE_NAME);
    if (matchesMapObject == null)
        return;
    if (!(matchesMapObject instanceof Map)) {
        // and return
        return;
    }
    @SuppressWarnings("unchecked") Map<String, List<List<Integer>>> matchesMap = (Map<String, List<List<Integer>>>) matchesMapObject;
    Set<String> keySet = matchesMap.keySet();
    if (keySet == null)
        return;
    Iterator<String> iter = keySet.iterator();
    boolean found = false;
    while (iter.hasNext()) {
        String currSet = iter.next();
        List<List<Integer>> matches = matchesMap.get(currSet);
        if (matches == null || matches.size() == 0)
            continue;
        else {
            for (int i = 0; i < matches.size(); i++) {
                List<Integer> ids = matches.get(i);
                if (ids.contains(id)) {
                    // found
                    // so remove this
                    found = true;
                    ids.remove(id);
                    matches.set(i, ids);
                    break;
                }
            }
            if (found) {
                matchesMap.put(currSet, matches);
                explicitCall = true;
                document.getFeatures().put(ANNIEConstants.DOCUMENT_COREF_FEATURE_NAME, matchesMap);
                explicitCall = false;
                break;
            }
        }
    }
    if (found)
        featureMapUpdated();
}

Example 2 with Annotation

use of gate.Annotation in project gate-core by GateNLP.

the class CorpusQualityAssurance method readSetsTypesFeatures.

/**
 * Update set lists.
 * @param documentStart first document to read in the corpus,
 * the first document of the corpus is 0.
 */
protected void readSetsTypesFeatures(final int documentStart) {
    if (!isShowing()) {
        return;
    }
    corpusChanged = false;
    SwingUtilities.invokeLater(new Runnable() {

        @Override
        public void run() {
            progressBar.setMaximum(corpus.size() - 1);
            progressBar.setString("Read sets, types, features");
            reloadCacheAction.setEnabled(false);
        }
    });
    CorpusQualityAssurance.this.setCursor(Cursor.getPredefinedCursor(Cursor.WAIT_CURSOR));
    Runnable runnable = new Runnable() {

        @Override
        public void run() {
            if (docsSetsTypesFeatures.size() != corpus.getDocumentNames().size() || !docsSetsTypesFeatures.keySet().containsAll(corpus.getDocumentNames())) {
                if (documentStart == 0) {
                    docsSetsTypesFeatures.clear();
                }
                TreeMap<String, TreeMap<String, TreeSet<String>>> setsTypesFeatures;
                TreeMap<String, TreeSet<String>> typesFeatures;
                TreeSet<String> features;
                for (int i = documentStart; i < corpus.size(); i++) {
                    // fill in the lists of document, set, type and feature names
                    boolean documentWasLoaded = corpus.isDocumentLoaded(i);
                    Document document = corpus.get(i);
                    if (document != null && document.getAnnotationSetNames() != null) {
                        setsTypesFeatures = new TreeMap<String, TreeMap<String, TreeSet<String>>>(collator);
                        HashSet<String> setNames = new HashSet<String>(document.getAnnotationSetNames());
                        setNames.add("");
                        for (String set : setNames) {
                            typesFeatures = new TreeMap<String, TreeSet<String>>(collator);
                            AnnotationSet annotations = document.getAnnotations(set);
                            for (String type : annotations.getAllTypes()) {
                                features = new TreeSet<String>(collator);
                                for (Annotation annotation : annotations.get(type)) {
                                    for (Object featureKey : annotation.getFeatures().keySet()) {
                                        features.add((String) featureKey);
                                    }
                                }
                                typesFeatures.put(type, features);
                            }
                            setsTypesFeatures.put(set, typesFeatures);
                        }
                        docsSetsTypesFeatures.put(document.getName(), setsTypesFeatures);
                    }
                    if (!documentWasLoaded) {
                        corpus.unloadDocument(document);
                        Factory.deleteResource(document);
                    }
                    final int progressValue = i + 1;
                    SwingUtilities.invokeLater(new Runnable() {

                        @Override
                        public void run() {
                            progressBar.setValue(progressValue);
                            if ((progressValue + 1) % 5 == 0) {
                                // update the set list every 5 documents read
                                updateSetList();
                            }
                        }
                    });
                    if (Thread.interrupted()) {
                        return;
                    }
                }
            }
            updateSetList();
            SwingUtilities.invokeLater(new Runnable() {

                @Override
                public void run() {
                    progressBar.setValue(progressBar.getMinimum());
                    progressBar.setString("");
                    CorpusQualityAssurance.this.setCursor(Cursor.getPredefinedCursor(Cursor.DEFAULT_CURSOR));
                    reloadCacheAction.setEnabled(true);
                }
            });
        }
    };
    readSetsTypesFeaturesThread = new Thread(runnable, "readSetsTypesFeatures");
    readSetsTypesFeaturesThread.setPriority(Thread.MIN_PRIORITY);
    readSetsTypesFeaturesThread.start();
}

Example 3 with Annotation

use of gate.Annotation in project gate-core by GateNLP.

the class CorpusQualityAssurance method compareAnnotation.

protected void compareAnnotation() {
    int progressValuePrevious = -1;
    if (readSetsTypesFeaturesThread != null && readSetsTypesFeaturesThread.isAlive()) {
        // stop the thread that reads the sets, types and features
        progressValuePrevious = progressBar.getValue();
        readSetsTypesFeaturesThread.interrupt();
    }
    SwingUtilities.invokeLater(new Runnable() {

        @Override
        public void run() {
            progressBar.setMaximum(corpus.size() - 1);
            progressBar.setString("Compare annotations");
            setList.setEnabled(false);
            setCheck.setEnabled(false);
            typeList.setEnabled(false);
            typeCheck.setEnabled(false);
            featureList.setEnabled(false);
            featureCheck.setEnabled(false);
            optionsButton.setEnabled(false);
            measureTabbedPane.setEnabled(false);
            measureList.setEnabled(false);
            exportToHtmlAction.setEnabled(false);
            reloadCacheAction.setEnabled(false);
        }
    });
    boolean useBdm = false;
    if (measuresType == FSCORE_MEASURES) {
        differsByDocThenType.clear();
        documentNames.clear();
        for (Object measure : measureList.getSelectedValues()) {
            if (((String) measure).contains("BDM")) {
                useBdm = true;
                break;
            }
        }
    }
    List<ClassificationMeasures> classificationMeasuresList = new ArrayList<ClassificationMeasures>();
    List<OntologyMeasures> documentOntologyMeasuresList = new ArrayList<OntologyMeasures>();
    List<OntologyMeasures> annotationOntologyMeasuresList = new ArrayList<OntologyMeasures>();
    // for each document
    for (int row = 0; row < corpus.size(); row++) {
        boolean documentWasLoaded = corpus.isDocumentLoaded(row);
        Document document = corpus.get(row);
        documentNames.add(document.getName());
        Set<Annotation> keys = new HashSet<Annotation>();
        Set<Annotation> responses = new HashSet<Annotation>();
        // get annotations from selected annotation sets
        if (keySetName.equals("[Default set]")) {
            keys = document.getAnnotations();
        } else if (document.getAnnotationSetNames() != null && document.getAnnotationSetNames().contains(keySetName)) {
            keys = document.getAnnotations(keySetName);
        }
        if (responseSetName.equals("[Default set]")) {
            responses = document.getAnnotations();
        } else if (document.getAnnotationSetNames() != null && document.getAnnotationSetNames().contains(responseSetName)) {
            responses = document.getAnnotations(responseSetName);
        }
        if (!documentWasLoaded) {
            // in case of datastore
            corpus.unloadDocument(document);
            Factory.deleteResource(document);
        }
        // add data to the fscore document table
        if (measuresType == FSCORE_MEASURES) {
            types.clear();
            for (Object type : typeList.getSelectedValues()) {
                types.add((String) type);
            }
            if (typeList.isSelectionEmpty()) {
                for (int i = 0; i < typeList.getModel().getSize(); i++) {
                    types.add((String) typeList.getModel().getElementAt(i));
                }
            }
            Set<String> featureSet = new HashSet<String>();
            for (Object feature : featureList.getSelectedValues()) {
                featureSet.add((String) feature);
            }
            HashMap<String, AnnotationDiffer> differsByType = new HashMap<String, AnnotationDiffer>();
            AnnotationDiffer differ;
            Set<Annotation> keysIter = new HashSet<Annotation>();
            Set<Annotation> responsesIter = new HashSet<Annotation>();
            for (String type : types) {
                if (!keys.isEmpty() && !types.isEmpty()) {
                    keysIter = ((AnnotationSet) keys).get(type);
                }
                if (!responses.isEmpty() && !types.isEmpty()) {
                    responsesIter = ((AnnotationSet) responses).get(type);
                }
                differ = new AnnotationDiffer();
                differ.setSignificantFeaturesSet(featureSet);
                // compare
                differ.calculateDiff(keysIter, responsesIter);
                differsByType.put(type, differ);
            }
            differsByDocThenType.add(differsByType);
            differ = new AnnotationDiffer(differsByType.values());
            List<String> measuresRow;
            if (useBdm) {
                OntologyMeasures ontologyMeasures = new OntologyMeasures();
                ontologyMeasures.setBdmFile(bdmFileUrl);
                ontologyMeasures.calculateBdm(differsByType.values());
                documentOntologyMeasuresList.add(ontologyMeasures);
                measuresRow = ontologyMeasures.getMeasuresRow(measureList.getSelectedValues(), documentNames.get(documentNames.size() - 1));
            } else {
                measuresRow = differ.getMeasuresRow(measureList.getSelectedValues(), documentNames.get(documentNames.size() - 1));
            }
            documentTableModel.addRow(measuresRow.toArray());
        // add data to the classification document table
        } else if (measuresType == CLASSIFICATION_MEASURES && !keys.isEmpty() && !responses.isEmpty()) {
            ClassificationMeasures classificationMeasures = new ClassificationMeasures();
            classificationMeasures.calculateConfusionMatrix((AnnotationSet) keys, (AnnotationSet) responses, (String) typeList.getSelectedValue(), (String) featureList.getSelectedValue(), verboseOptionCheckBox.isSelected());
            classificationMeasuresList.add(classificationMeasures);
            List<String> measuresRow = classificationMeasures.getMeasuresRow(measure2List.getSelectedValues(), documentNames.get(documentNames.size() - 1));
            document2TableModel.addRow(measuresRow.toArray());
            List<List<String>> matrix = classificationMeasures.getConfusionMatrix(documentNames.get(documentNames.size() - 1));
            for (List<String> matrixRow : matrix) {
                while (confusionTableModel.getColumnCount() < matrix.size()) {
                    confusionTableModel.addColumn(" ");
                }
                confusionTableModel.addRow(matrixRow.toArray());
            }
        }
        final int progressValue = row + 1;
        SwingUtilities.invokeLater(new Runnable() {

            @Override
            public void run() {
                progressBar.setValue(progressValue);
            }
        });
    }
    // add data to the fscore annotation table
    if (measuresType == FSCORE_MEASURES) {
        for (String type : types) {
            ArrayList<AnnotationDiffer> differs = new ArrayList<AnnotationDiffer>();
            for (HashMap<String, AnnotationDiffer> differsByType : differsByDocThenType) {
                differs.add(differsByType.get(type));
            }
            List<String> measuresRow;
            if (useBdm) {
                OntologyMeasures ontologyMeasures = new OntologyMeasures();
                ontologyMeasures.setBdmFile(bdmFileUrl);
                ontologyMeasures.calculateBdm(differs);
                annotationOntologyMeasuresList.add(ontologyMeasures);
                measuresRow = ontologyMeasures.getMeasuresRow(measureList.getSelectedValues(), type);
            } else {
                AnnotationDiffer differ = new AnnotationDiffer(differs);
                measuresRow = differ.getMeasuresRow(measureList.getSelectedValues(), type);
            }
            annotationTableModel.addRow(measuresRow.toArray());
        }
    }
    // add summary rows to the fscore tables
    if (measuresType == FSCORE_MEASURES) {
        if (useBdm) {
            OntologyMeasures ontologyMeasures = new OntologyMeasures(documentOntologyMeasuresList);
            printSummary(ontologyMeasures, documentTableModel, 5, documentTableModel.getRowCount(), measureList.getSelectedValues());
            ontologyMeasures = new OntologyMeasures(annotationOntologyMeasuresList);
            printSummary(ontologyMeasures, annotationTableModel, 5, annotationTableModel.getRowCount(), measureList.getSelectedValues());
        } else {
            List<AnnotationDiffer> differs = new ArrayList<AnnotationDiffer>();
            for (Map<String, AnnotationDiffer> differsByType : differsByDocThenType) {
                differs.addAll(differsByType.values());
            }
            AnnotationDiffer differ = new AnnotationDiffer(differs);
            printSummary(differ, documentTableModel, 5, documentTableModel.getRowCount(), measureList.getSelectedValues());
            printSummary(differ, annotationTableModel, 5, annotationTableModel.getRowCount(), measureList.getSelectedValues());
        }
    // add summary rows to the classification tables
    } else if (measuresType == CLASSIFICATION_MEASURES) {
        ClassificationMeasures classificationMeasures = new ClassificationMeasures(classificationMeasuresList);
        printSummary(classificationMeasures, document2TableModel, 3, document2TableModel.getRowCount(), measure2List.getSelectedValues());
        List<List<String>> matrix = classificationMeasures.getConfusionMatrix("Whole corpus");
        int insertionRow = 0;
        for (List<String> row : matrix) {
            while (confusionTableModel.getColumnCount() < matrix.size()) {
                confusionTableModel.addColumn(" ");
            }
            confusionTableModel.insertRow(insertionRow++, row.toArray());
        }
    }
    SwingUtilities.invokeLater(new Runnable() {

        @Override
        public void run() {
            progressBar.setValue(progressBar.getMinimum());
            progressBar.setString("");
            setList.setEnabled(true);
            setCheck.setEnabled(true);
            typeList.setEnabled(true);
            typeCheck.setEnabled(true);
            featureList.setEnabled(true);
            featureCheck.setEnabled(true);
            optionsButton.setEnabled(true);
            measureTabbedPane.setEnabled(true);
            measureList.setEnabled(true);
            exportToHtmlAction.setEnabled(true);
            reloadCacheAction.setEnabled(true);
        }
    });
    if (progressValuePrevious > -1) {
        // restart the thread where it was interrupted
        readSetsTypesFeatures(progressValuePrevious);
    }
}

Example 4 with Annotation

use of gate.Annotation in project gate-core by GateNLP.

the class AnnotationMerging method mergeAnnotation.

/**
 * Merge all annotationset from an array. If one annotation is in at least
 * numK annotation sets, then put it into the merging annotation set.
 */
public static void mergeAnnotation(AnnotationSet[] annsArr, String nameFeat, HashMap<Annotation, String> mergeAnns, int numMinK, boolean isTheSameInstances) {
    int numA = annsArr.length;
    // First copy the annotatioin sets into a temp array
    @SuppressWarnings({ "unchecked", "rawtypes" }) Set<Annotation>[] annsArrTemp = new Set[numA];
    for (int i = 0; i < numA; ++i) {
        if (annsArr[i] != null) {
            annsArrTemp[i] = new HashSet<Annotation>();
            for (Annotation ann : annsArr[i]) annsArrTemp[i].add(ann);
        }
    }
    HashSet<String> featSet = new HashSet<String>();
    if (nameFeat != null)
        featSet.add(nameFeat);
    if (numMinK < 1)
        numMinK = 1;
    for (int iA = 0; iA < numA - numMinK + 1; ++iA) {
        if (annsArrTemp[iA] != null) {
            for (Annotation ann : annsArrTemp[iA]) {
                int numContained = 1;
                StringBuffer featAdd = new StringBuffer();
                featAdd.append(iA);
                StringBuffer featDisa = new StringBuffer();
                if (iA > 0) {
                    featDisa.append("0");
                    for (int i = 1; i < iA; ++i) featDisa.append("-" + i);
                }
                int numDisagreed = iA;
                for (int i = iA + 1; i < numA; ++i) {
                    boolean isContained = false;
                    if (annsArrTemp[i] != null) {
                        Annotation annT = null;
                        for (Annotation ann0 : annsArrTemp[i]) {
                            if (ann0.isCompatible(ann, featSet)) {
                                ++numContained;
                                featAdd.append("-" + i);
                                annT = ann0;
                                isContained = true;
                                break;
                            }
                        }
                        if (isContained)
                            annsArrTemp[i].remove(annT);
                    }
                    if (!isContained) {
                        if (numDisagreed == 0)
                            featDisa.append(i);
                        else
                            featDisa.append("-" + i);
                        ++numDisagreed;
                    }
                }
                if (numContained >= numMinK) {
                    mergeAnns.put(ann, featAdd.toString());
                } else if (isTheSameInstances && nameFeat != null) {
                    ann.getFeatures().remove(nameFeat);
                    mergeAnns.put(ann, featAdd.toString());
                }
            }
        }
    }
    // Remove the annotation in the same place
    removeDuplicate(mergeAnns);
    return;
}

Example 5 with Annotation

use of gate.Annotation in project gate-core by GateNLP.

the class AnnotationMerging method mergeAnnotationMajority.

/**
 * Merge all annotationset from an array. If one annotation is agreed by
 * the majority of the annotators, then put it into the merging annotation set.
 */
public static void mergeAnnotationMajority(AnnotationSet[] annsArr, String nameFeat, HashMap<Annotation, String> mergeAnns, boolean isTheSameInstances) {
    int numA = annsArr.length;
    if (nameFeat == null) {
        mergeAnnogationMajorityNoFeat(annsArr, mergeAnns, isTheSameInstances);
        return;
    }
    // First copy the annotatioin sets into a temp array
    @SuppressWarnings({ "unchecked", "rawtypes" }) Set<Annotation>[] annsArrTemp = new Set[numA];
    for (int i = 0; i < numA; ++i) {
        if (annsArr[i] != null) {
            annsArrTemp[i] = new HashSet<Annotation>();
            for (Annotation ann : annsArr[i]) annsArrTemp[i].add(ann);
        }
    }
    for (int iA = 0; iA < numA; ++iA) {
        if (annsArrTemp[iA] != null) {
            for (Annotation ann : annsArrTemp[iA]) {
                int numDisagreed = 0;
                // Already the iA annotators don't agree the annotation
                numDisagreed = iA;
                StringBuffer featDisa = new StringBuffer();
                if (iA > 0) {
                    featDisa.append("0");
                    for (int i = 1; i < iA; ++i) featDisa.append("-" + i);
                }
                HashMap<String, String> featOthers = new HashMap<String, String>();
                String featTh = null;
                if (ann.getFeatures().get(nameFeat) != null)
                    featTh = ann.getFeatures().get(nameFeat).toString();
                featOthers.put(featTh, new Integer(iA).toString());
                HashMap<String, Annotation> annAll = new HashMap<String, Annotation>();
                annAll.put(featTh, ann);
                for (int i = iA + 1; i < numA; ++i) {
                    boolean isContained = false;
                    if (annsArrTemp[i] != null) {
                        Annotation annT = null;
                        for (Annotation ann0 : annsArrTemp[i]) {
                            if (ann0.coextensive(ann)) {
                                String featValue = null;
                                if (ann0.getFeatures().get(nameFeat) != null)
                                    featValue = ann0.getFeatures().get(nameFeat).toString();
                                if (!featOthers.containsKey(featValue)) {
                                    featOthers.put(featValue, new Integer(i).toString());
                                    annAll.put(featValue, ann0);
                                } else {
                                    String str = featOthers.get(featValue);
                                    featOthers.put(featValue, str + "-" + i);
                                }
                                annT = ann0;
                                isContained = true;
                                break;
                            }
                        }
                        if (isContained)
                            annsArrTemp[i].remove(annT);
                    }
                    if (!isContained) {
                        if (numDisagreed == 0)
                            featDisa.append(i);
                        else
                            featDisa.append("-" + i);
                        ++numDisagreed;
                    }
                }
                // end of the loop for the following annotation set
                int numAgreed = -1;
                String agreeFeat = null;
                for (String str : featOthers.keySet()) {
                    String str0 = featOthers.get(str);
                    int num = 1;
                    while (str0.contains("-")) {
                        ++num;
                        str0 = str0.substring(str0.indexOf('-') + 1);
                    }
                    if (numAgreed < num) {
                        numAgreed = num;
                        agreeFeat = str;
                    }
                }
                if (numAgreed >= numDisagreed) {
                    mergeAnns.put(annAll.get(agreeFeat), featOthers.get(agreeFeat));
                } else if (isTheSameInstances) {
                    if (ann.getFeatures().get(nameFeat) != null)
                        ann.getFeatures().remove(nameFeat);
                    mergeAnns.put(ann, featDisa.toString());
                }
            }
        // for each ann in the current annotation set
        }
    }
    return;
}