Examples with Document gate.Document used on opensource projects

Example 16 with Document

use of gate.Document 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 17 with Document

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

the class SerialCorpusImpl method unloadDocument.

/**
 * Unloads a document from memory.
 *
 * @param index the index of the document to be unloaded.
 * @param sync should the document be sync'ed (i.e. saved) before
 *          unloading.
 */
public void unloadDocument(int index, boolean sync) {
    // if a persistent doc is not loaded, there's nothing we need to do
    if ((!isDocumentLoaded(index)) && isPersistentDocument(index))
        return;
    // any more
    if (sync) {
        Document doc = documents.get(index);
        try {
            // first
            if (doc.getLRPersistenceId() == null) {
                doc = (Document) this.getDataStore().adopt(doc);
                this.getDataStore().sync(doc);
                this.setDocumentPersistentID(index, doc.getLRPersistenceId());
            } else
                // if it is adopted, just sync it
                this.getDataStore().sync(doc);
        } catch (PersistenceException ex) {
            throw new GateRuntimeException("Error unloading document from corpus" + "because document sync failed: " + ex.getMessage(), ex);
        }
    }
    // 3. remove the document from the memory
    // do this, only if the saving has succeeded
    documents.set(index, null);
}

Example 18 with Document

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

the class SerialCorpusImpl method setTransientSource.

public void setTransientSource(Object source) {
    if (!(source instanceof Corpus))
        return;
    // are restored.
    if (this.dataStore != null && this.lrPersistentId != null)
        return;
    Corpus tCorpus = (Corpus) source;
    // copy the corpus name and features from the one in memory
    this.setName(tCorpus.getName());
    this.setFeatures(tCorpus.getFeatures());
    docDataList = new ArrayList<DocumentData>();
    // now cache the names of all docs for future use
    List<String> docNames = tCorpus.getDocumentNames();
    for (int i = 0; i < docNames.size(); i++) {
        Document aDoc = tCorpus.get(i);
        docDataList.add(new DocumentData(docNames.get(i), null, aDoc.getClass().getName()));
    }
    // copy all the documents from the transient corpus
    documents = new ArrayList<Document>();
    documents.addAll(tCorpus);
    this.addedDocs = new Vector<Document>();
    this.removedDocIDs = new Vector<String>();
    this.changedDocs = new Vector<Document>();
    // make sure we fire events when docs are added/removed/etc
    Gate.getCreoleRegister().addCreoleListener(this);
}

Example 19 with Document

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

the class SerialCorpusImpl method add.

@Override
public void add(int index, Document o) {
    if (o == null)
        return;
    Document doc = o;
    DocumentData docData = new DocumentData(doc.getName(), doc.getLRPersistenceId(), doc.getClass().getName());
    docDataList.add(index, docData);
    documents.add(index, doc);
    documentAdded(doc);
    fireDocumentAdded(new CorpusEvent(SerialCorpusImpl.this, doc, index, doc.getLRPersistenceId(), CorpusEvent.DOCUMENT_ADDED));
}

Example 20 with Document

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

the class SerialCorpusImpl method setIndexDefinition.

// readObject
@Override
public void setIndexDefinition(IndexDefinition definition) {
    if (definition != null) {
        this.getFeatures().put(GateConstants.CORPUS_INDEX_DEFINITION_FEATURE_KEY, definition);
        String className = definition.getIrEngineClassName();
        try {
            // Class aClass = Class.forName(className);
            Class<?> aClass = Class.forName(className, true, Gate.getClassLoader());
            IREngine engine = (IREngine) aClass.newInstance();
            this.indexManager = engine.getIndexmanager();
            this.indexManager.setIndexDefinition(definition);
            this.indexManager.setCorpus(this);
        } catch (Exception e) {
            e.printStackTrace(Err.getPrintWriter());
        }
        // switch (definition.getIndexType()) {
        // case GateConstants.IR_LUCENE_INVFILE:
        // this.indexManager = new LuceneIndexManager();
        // this.indexManager.setIndexDefinition(definition);
        // this.indexManager.setCorpus(this);
        // break;
        // }
        this.addedDocs = new Vector<Document>();
        this.removedDocIDs = new Vector<String>();
        this.changedDocs = new Vector<Document>();
    }
}