Examples with AnnotationSet gate.AnnotationSet used on opensource projects

Example 11 with AnnotationSet

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

the class TestClassificationMeasures method test.

public void test() {
    String type = "sent";
    String feature = "Op";
    Document doc1 = null;
    Document doc2 = null;
    Document doc3 = null;
    Document doc4 = null;
    try {
        Gate.init();
        doc1 = Factory.newDocument(new URL(TestDocument.getTestServerName() + "tests/iaa/beijing-opera.xml"));
        doc2 = Factory.newDocument(new URL(TestDocument.getTestServerName() + "tests/iaa/beijing-opera.xml"));
        doc3 = Factory.newDocument(new URL(TestDocument.getTestServerName() + "tests/iaa/in-outlook-09-aug-2001.xml"));
        doc4 = Factory.newDocument(new URL(TestDocument.getTestServerName() + "tests/iaa/in-outlook-09-aug-2001.xml"));
    } catch (Exception e) {
        e.printStackTrace();
    }
    if (doc1 != null && doc2 != null && doc3 != null && doc4 != null) {
        AnnotationSet as1 = doc1.getAnnotations("ann1");
        AnnotationSet as2 = doc2.getAnnotations("ann2");
        ClassificationMeasures myClassificationMeasures1 = new ClassificationMeasures();
        myClassificationMeasures1.calculateConfusionMatrix(as1, as2, type, feature, true);
        assertEquals(myClassificationMeasures1.getObservedAgreement(), 0.7777778f);
        assertEquals(myClassificationMeasures1.getKappaCohen(), 0.6086957f);
        assertEquals(myClassificationMeasures1.getKappaPi(), 0.59550565f);
        AnnotationSet as3 = doc3.getAnnotations("ann1");
        AnnotationSet as4 = doc4.getAnnotations("ann2");
        ClassificationMeasures myClassificationMeasures2 = new ClassificationMeasures();
        myClassificationMeasures2.calculateConfusionMatrix(as3, as4, type, feature, true);
        assertEquals(myClassificationMeasures2.getObservedAgreement(), 0.96875f);
        assertEquals(myClassificationMeasures2.getKappaCohen(), 0.3263158f);
        assertEquals(myClassificationMeasures2.getKappaPi(), 0.3227513f);
        ArrayList<ClassificationMeasures> tablesList = new ArrayList<ClassificationMeasures>();
        tablesList.add(myClassificationMeasures1);
        tablesList.add(myClassificationMeasures2);
        ClassificationMeasures myNewClassificationMeasures = new ClassificationMeasures(tablesList);
        assertEquals(myNewClassificationMeasures.getObservedAgreement(), 0.94520545f);
        assertEquals(myNewClassificationMeasures.getKappaCohen(), 0.7784521f);
        assertEquals(myNewClassificationMeasures.getKappaPi(), 0.7778622f);
    } else {
        System.out.println("Failed to create docs from URLs.");
    }
}

Example 12 with AnnotationSet

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

the class DocumentStaxUtils method readGateXmlDocument.

/**
 * Reads GATE XML format data from the given XMLStreamReader and puts
 * the content and annotation sets into the given Document, replacing
 * its current content. The reader must be positioned on the opening
 * GateDocument tag (i.e. the last event was a START_ELEMENT for which
 * getLocalName returns "GateDocument"), and when the method returns
 * the reader will be left positioned on the corresponding closing
 * tag.
 *
 * @param xsr the source of the XML to parse
 * @param doc the document to update
 * @param statusListener optional status listener to receive status
 *          messages
 * @throws XMLStreamException
 */
public static void readGateXmlDocument(XMLStreamReader xsr, Document doc, StatusListener statusListener) throws XMLStreamException {
    DocumentContent savedContent = null;
    // check the precondition
    xsr.require(XMLStreamConstants.START_ELEMENT, null, "GateDocument");
    // process the document features
    xsr.nextTag();
    xsr.require(XMLStreamConstants.START_ELEMENT, null, "GateDocumentFeatures");
    if (statusListener != null) {
        statusListener.statusChanged("Reading document features");
    }
    FeatureMap documentFeatures = readFeatureMap(xsr);
    // read document text, building the map of node IDs to offsets
    xsr.nextTag();
    xsr.require(XMLStreamConstants.START_ELEMENT, null, "TextWithNodes");
    Map<Integer, Long> nodeIdToOffsetMap = new HashMap<Integer, Long>();
    if (statusListener != null) {
        statusListener.statusChanged("Reading document content");
    }
    String documentText = readTextWithNodes(xsr, nodeIdToOffsetMap);
    // save the content, in case anything goes wrong later
    savedContent = doc.getContent();
    // set the document content to the text with nodes text.
    doc.setContent(new DocumentContentImpl(documentText));
    try {
        int numAnnots = 0;
        // process annotation sets, using the node map built above
        Integer maxAnnotId = null;
        // initially, we don't know whether annotation IDs are required or
        // not
        Boolean requireAnnotationIds = null;
        int eventType = xsr.nextTag();
        while (eventType == XMLStreamConstants.START_ELEMENT && xsr.getLocalName().equals("AnnotationSet")) {
            xsr.require(XMLStreamConstants.START_ELEMENT, null, "AnnotationSet");
            String annotationSetName = xsr.getAttributeValue(null, "Name");
            AnnotationSet annotationSet = null;
            if (annotationSetName == null) {
                if (statusListener != null) {
                    statusListener.statusChanged("Reading default annotation set");
                }
                annotationSet = doc.getAnnotations();
            } else {
                if (statusListener != null) {
                    statusListener.statusChanged("Reading \"" + annotationSetName + "\" annotation set");
                }
                annotationSet = doc.getAnnotations(annotationSetName);
            }
            annotationSet.clear();
            SortedSet<Integer> annotIdsInSet = new TreeSet<Integer>();
            requireAnnotationIds = readAnnotationSet(xsr, annotationSet, nodeIdToOffsetMap, annotIdsInSet, requireAnnotationIds);
            if (annotIdsInSet.size() > 0 && (maxAnnotId == null || annotIdsInSet.last().intValue() > maxAnnotId.intValue())) {
                maxAnnotId = annotIdsInSet.last();
            }
            numAnnots += annotIdsInSet.size();
            // readAnnotationSet leaves reader positioned on the
            // </AnnotationSet> tag, so nextTag takes us to either the next
            // <AnnotationSet>, a <RelationSet>, or </GateDocument>
            eventType = xsr.nextTag();
        }
        while (eventType == XMLStreamConstants.START_ELEMENT && xsr.getLocalName().equals("RelationSet")) {
            xsr.require(XMLStreamConstants.START_ELEMENT, null, "RelationSet");
            String relationSetName = xsr.getAttributeValue(null, "Name");
            RelationSet relations = null;
            if (relationSetName == null) {
                if (statusListener != null) {
                    statusListener.statusChanged("Reading relation set for default annotation set");
                }
                relations = doc.getAnnotations().getRelations();
            } else {
                if (statusListener != null) {
                    statusListener.statusChanged("Reading relation set for \"" + relationSetName + "\" annotation set");
                }
                relations = doc.getAnnotations(relationSetName).getRelations();
            }
            SortedSet<Integer> relIdsInSet = new TreeSet<Integer>();
            readRelationSet(xsr, relations, relIdsInSet);
            if (relIdsInSet.size() > 0 && (maxAnnotId == null || relIdsInSet.last().intValue() > maxAnnotId.intValue())) {
                maxAnnotId = relIdsInSet.last();
            }
            numAnnots += relIdsInSet.size();
            // readAnnotationSet leaves reader positioned on the
            // </RelationSet> tag, so nextTag takes us to either the next
            // <RelationSet> or to the </GateDocument>
            eventType = xsr.nextTag();
        }
        // check we are on the end document tag
        xsr.require(XMLStreamConstants.END_ELEMENT, null, "GateDocument");
        doc.setFeatures(documentFeatures);
        // set the ID generator, if doc is a DocumentImpl
        if (doc instanceof DocumentImpl && maxAnnotId != null) {
            ((DocumentImpl) doc).setNextAnnotationId(maxAnnotId.intValue() + 1);
        }
        if (statusListener != null) {
            statusListener.statusChanged("Finished.  " + numAnnots + " annotation(s) processed");
        }
    }// in case of exception, reset document content to the unparsed XML
     catch (XMLStreamException xse) {
        doc.setContent(savedContent);
        throw xse;
    } catch (RuntimeException re) {
        doc.setContent(savedContent);
        throw re;
    }
}

Example 13 with AnnotationSet

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

the class AnnotationSetImpl method get.

// get(types)
/**
 * Select annotations by type and features
 *
 * This will return an annotation set containing just those annotations of a
 * particular type (i.e. with a particular name) and which have features with
 * specific names and values. (It will also return annotations that have
 * features besides those specified, but it will not return any annotations
 * that do not have all the specified feature-value pairs.)
 *
 * However, if constraints contains a feature whose value is equal to
 * gate.creole.ANNIEConstants.LOOKUP_CLASS_FEATURE_NAME (which is normally
 * "class"), then GATE will attempt to match that feature using an ontology
 * which it will try to retreive from a feature on the both the annotation and
 * in constraints. If these do not return identical ontologies, or if either
 * the annotation or constraints does not contain an ontology, then matching
 * will fail, and the annotation will not be added. In summary, this method
 * will not work normally for features with the name "class".
 *
 * @param type
 *          The name of the annotations to return.
 * @param constraints
 *          A feature map containing all of the feature value pairs that the
 *          annotation must have in order for them to be returned.
 * @return An annotation set containing only those annotations with the given
 *         name and which have the specified set of feature-value pairs.
 */
@Override
public AnnotationSet get(String type, FeatureMap constraints) {
    if (annotsByType == null)
        indexByType();
    AnnotationSet typeSet = get(type);
    if (typeSet == null)
        return null;
    Iterator<Annotation> iter = typeSet.iterator();
    List<Annotation> annotationsToAdd = new ArrayList<Annotation>();
    while (iter.hasNext()) {
        Annotation a = iter.next();
        // (a.getFeatures().entrySet().containsAll(constraints.entrySet()))
        if (a.getFeatures().subsumes(constraints))
            annotationsToAdd.add(a);
    }
    // while
    if (annotationsToAdd.isEmpty())
        return emptyAS();
    return new ImmutableAnnotationSetImpl(doc, annotationsToAdd);
}

Example 14 with AnnotationSet

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

the class AnnotationSetImpl method get.

// get(type, constraints)
/**
 * Select annotations by type and feature names
 */
@Override
public AnnotationSet get(String type, Set<? extends Object> featureNames) {
    if (annotsByType == null)
        indexByType();
    AnnotationSet typeSet = null;
    if (type != null) {
        // if a type is provided, try finding annotations of this type
        typeSet = get(type);
        // if none exist, then return coz nothing left to do
        if (typeSet == null)
            return null;
    }
    List<Annotation> annotationsToAdd = new ArrayList<Annotation>();
    Iterator<Annotation> iter = null;
    if (type != null)
        iter = typeSet.iterator();
    else
        iter = annotsById.values().iterator();
    while (iter.hasNext()) {
        Annotation a = iter.next();
        // key/value pairs from the constraints map
        if (a.getFeatures().keySet().containsAll(featureNames))
            annotationsToAdd.add(a);
    }
    // while
    if (annotationsToAdd.isEmpty())
        return emptyAS();
    return new ImmutableAnnotationSetImpl(doc, annotationsToAdd);
}

Example 15 with AnnotationSet

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

the class AnnotationSetImpl method addToTypeIndex.

// indexByStartOffset()
/**
 * Add an annotation to the type index. Does nothing if the index doesn't
 * exist.
 */
void addToTypeIndex(Annotation a) {
    if (annotsByType == null)
        return;
    String type = a.getType();
    AnnotationSet sameType = annotsByType.get(type);
    if (sameType == null) {
        sameType = new AnnotationSetImpl(doc);
        annotsByType.put(type, sameType);
    }
    sameType.add(a);
}