Examples with Lemma de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma used on opensource projects

Example 11 with Lemma

use of de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma in project webanno by webanno.

the class ConstraintsVerifierTest method test.

@Test
public void test() throws Exception {
    ConstraintsGrammar parser = new ConstraintsGrammar(new FileInputStream("src/test/resources/rules/6.rules"));
    Parse p = parser.Parse();
    ParsedConstraints constraints = p.accept(new ParserVisitor());
    // Get imports
    Map<String, String> imports = new LinkedHashMap<>();
    imports = constraints.getImports();
    // Get rules
    // List<Rule> rules = new ArrayList<>();
    JCas jcas = JCasFactory.createJCas();
    jcas.setDocumentText("Just some text.");
    Lemma lemma1 = new Lemma(jcas, 0, 1);
    lemma1.setValue("good");
    lemma1.addToIndexes();
    Lemma lemma2 = new Lemma(jcas, 1, 2);
    lemma2.setValue("bad");
    lemma2.addToIndexes();
    Verifiable cVerifier = new ConstraintsVerifier();
    for (Lemma lemma : select(jcas, Lemma.class)) {
        if (lemma == lemma1) {
            assertEquals(true, cVerifier.verify(lemma, constraints));
        }
        if (lemma == lemma2) {
            assertEquals(false, cVerifier.verify(lemma, constraints));
        }
    }
}

Example 12 with Lemma

use of de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma in project webanno by webanno.

the class WebAnnoTsv3WriterTestBase method testTokenAttachedAnnotationsWithValues.

@Test
public void testTokenAttachedAnnotationsWithValues() throws Exception {
    JCas jcas = makeJCasOneSentence();
    List<Token> tokens = new ArrayList<>(select(jcas, Token.class));
    Token t1 = tokens.get(0);
    Lemma l1 = new Lemma(jcas, t1.getBegin(), t1.getEnd());
    l1.setValue("lemma1");
    l1.addToIndexes();
    t1.setLemma(l1);
    MorphologicalFeatures m1 = new MorphologicalFeatures(jcas, t1.getBegin(), t1.getEnd());
    m1.setValue("morph");
    m1.setTense("tense1");
    m1.addToIndexes();
    t1.setMorph(m1);
    POS p1 = new POS(jcas, t1.getBegin(), t1.getEnd());
    p1.setPosValue("pos1");
    p1.addToIndexes();
    t1.setPos(p1);
    Stem s1 = new Stem(jcas, t1.getBegin(), t1.getEnd());
    s1.setValue("stem1");
    s1.addToIndexes();
    t1.setStem(s1);
    writeAndAssertEquals(jcas, WebannoTsv3Writer.PARAM_SPAN_LAYERS, asList(MorphologicalFeatures.class, POS.class, Lemma.class, Stem.class));
}

Example 13 with Lemma

use of de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma in project dkpro-tc by dkpro.

the class TypeTokenPairFeatureExtractorTest method setUp.

@Before
public void setUp() throws ResourceInitializationException, AnalysisEngineProcessException {
    AnalysisEngineDescription desc = createEngineDescription(BreakIteratorSegmenter.class);
    AnalysisEngine engine = createEngine(desc);
    jcas1 = engine.newJCas();
    jcas1.setDocumentLanguage("en");
    jcas1.setDocumentText("This is text");
    engine.process(jcas1);
    Lemma lemma1 = new Lemma(jcas1, 0, 4);
    lemma1.setValue("text");
    lemma1.addToIndexes();
    Lemma lemma2 = new Lemma(jcas1, 5, 7);
    lemma2.setValue("is");
    lemma2.addToIndexes();
    Lemma lemma3 = new Lemma(jcas1, 8, 10);
    lemma3.setValue("text");
    lemma3.addToIndexes();
    jcas2 = engine.newJCas();
    jcas2.setDocumentLanguage("en");
    jcas2.setDocumentText("Text is text");
    engine.process(jcas2);
    Lemma lemma4 = new Lemma(jcas2, 0, 4);
    lemma4.setValue("text");
    lemma4.addToIndexes();
    Lemma lemma5 = new Lemma(jcas2, 8, 10);
    lemma5.setValue("text");
    lemma5.addToIndexes();
}

Example 14 with Lemma

use of de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma in project webanno by webanno.

the class MergeCas method reMergeCas.

/**
 * Using {@code DiffResult}, determine the annotations to be deleted from the randomly generated
 * MergeCase. The initial Merge CAs is stored under a name {@code CurationPanel#CURATION_USER}.
 * <p>
 * Any similar annotations stacked in a {@code CasDiff2.Position} will be assumed a difference
 * <p>
 * Any two annotation with different value will be assumed a difference
 * <p>
 * Any non stacked empty/null annotations are assumed agreement
 * <p>
 * Any non stacked annotations with similar values for each of the features are assumed
 * agreement
 * <p>
 * Any two link mode / slotable annotations which agree on the base features are assumed
 * agreement
 *
 * @param aDiff
 *            the {@code CasDiff2.DiffResult}
 * @param aJCases
 *            a map of{@code JCas}s for each users and the random merge
 * @return the actual merge {@code JCas}
 */
public static JCas reMergeCas(DiffResult aDiff, Map<String, JCas> aJCases) {
    Set<FeatureStructure> slotFeaturesToReset = new HashSet<>();
    Set<FeatureStructure> annotationsToDelete = new HashSet<>();
    Set<String> users = aJCases.keySet();
    for (Position position : aDiff.getPositions()) {
        Map<String, List<FeatureStructure>> annosPerUser = new HashMap<>();
        ConfigurationSet cfgs = aDiff.getConfigurtionSet(position);
        if (cfgs.getConfigurations(WebAnnoConst.CURATION_USER).isEmpty()) {
            // annotations
            continue;
        }
        AnnotationFS mergeAnno = (AnnotationFS) cfgs.getConfigurations(WebAnnoConst.CURATION_USER).get(0).getFs(WebAnnoConst.CURATION_USER, aJCases);
        // Get Annotations per user in this position
        getAllAnnosOnPosition(aJCases, annosPerUser, users, mergeAnno);
        for (FeatureStructure mergeFs : annosPerUser.get(WebAnnoConst.CURATION_USER)) {
            // incomplete annotations
            if (aJCases.size() != annosPerUser.size()) {
                annotationsToDelete.add(mergeFs);
            } else // agreed and not stacked
            if (isAgree(mergeFs, annosPerUser)) {
                Type t = mergeFs.getType();
                Feature sourceFeat = t.getFeatureByBaseName(WebAnnoConst.FEAT_REL_SOURCE);
                Feature targetFeat = t.getFeatureByBaseName(WebAnnoConst.FEAT_REL_TARGET);
                // Is this a relation?
                if (sourceFeat != null && targetFeat != null) {
                    AnnotationFS source = (AnnotationFS) mergeFs.getFeatureValue(sourceFeat);
                    AnnotationFS target = (AnnotationFS) mergeFs.getFeatureValue(targetFeat);
                    // all span anno on this source positions
                    Map<String, List<FeatureStructure>> sourceAnnosPerUser = new HashMap<>();
                    // all span anno on this target positions
                    Map<String, List<FeatureStructure>> targetAnnosPerUser = new HashMap<>();
                    getAllAnnosOnPosition(aJCases, sourceAnnosPerUser, users, source);
                    getAllAnnosOnPosition(aJCases, targetAnnosPerUser, users, target);
                    if (isAgree(source, sourceAnnosPerUser) && isAgree(target, targetAnnosPerUser)) {
                        slotFeaturesToReset.add(mergeFs);
                    } else {
                        annotationsToDelete.add(mergeFs);
                    }
                } else {
                    slotFeaturesToReset.add(mergeFs);
                }
            } else // disagree or stacked annotations
            {
                annotationsToDelete.add(mergeFs);
            }
        // remove dangling rels
        // setDanglingRelToDel(aJCases.get(CurationPanel.CURATION_USER),
        // mergeFs, annotationsToDelete);
        }
    }
    // remove annotations that do not agree or are a stacked ones
    for (FeatureStructure fs : annotationsToDelete) {
        if (!slotFeaturesToReset.contains(fs)) {
            JCas mergeCas = aJCases.get(WebAnnoConst.CURATION_USER);
            // Check if this difference is on POS, STEM and LEMMA (so remove from the token too)
            Type type = fs.getType();
            int fsBegin = ((AnnotationFS) fs).getBegin();
            int fsEnd = ((AnnotationFS) fs).getEnd();
            if (type.getName().equals(POS.class.getName())) {
                Token t = JCasUtil.selectCovered(mergeCas, Token.class, fsBegin, fsEnd).get(0);
                t.setPos(null);
            }
            if (type.getName().equals(Stem.class.getName())) {
                Token t = JCasUtil.selectCovered(mergeCas, Token.class, fsBegin, fsEnd).get(0);
                t.setStem(null);
            }
            if (type.getName().equals(Lemma.class.getName())) {
                Token t = JCasUtil.selectCovered(mergeCas, Token.class, fsBegin, fsEnd).get(0);
                t.setLemma(null);
            }
            if (type.getName().equals(MorphologicalFeatures.class.getName())) {
                Token t = JCasUtil.selectCovered(mergeCas, Token.class, fsBegin, fsEnd).get(0);
                t.setMorph(null);
            }
            mergeCas.removeFsFromIndexes(fs);
        }
    }
    // if slot bearing annotation, clean
    for (FeatureStructure baseFs : slotFeaturesToReset) {
        for (Feature roleFeature : baseFs.getType().getFeatures()) {
            if (isLinkMode(baseFs, roleFeature)) {
                // FeatureStructure roleFs = baseFs.getFeatureValue(f);
                ArrayFS roleFss = (ArrayFS) WebAnnoCasUtil.getFeatureFS(baseFs, roleFeature.getShortName());
                if (roleFss == null) {
                    continue;
                }
                Map<String, ArrayFS> roleAnnosPerUser = new HashMap<>();
                setAllRoleAnnosOnPosition(aJCases, roleAnnosPerUser, users, baseFs, roleFeature);
                List<FeatureStructure> linkFSes = new LinkedList<>(Arrays.asList(roleFss.toArray()));
                for (FeatureStructure roleFs : roleFss.toArray()) {
                    if (isRoleAgree(roleFs, roleAnnosPerUser)) {
                        for (Feature targetFeature : roleFs.getType().getFeatures()) {
                            if (isBasicFeature(targetFeature)) {
                                continue;
                            }
                            if (!targetFeature.getShortName().equals("target")) {
                                continue;
                            }
                            AnnotationFS targetFs = (AnnotationFS) roleFs.getFeatureValue(targetFeature);
                            if (targetFs == null) {
                                continue;
                            }
                            Map<String, List<FeatureStructure>> targetAnnosPerUser = new HashMap<>();
                            getAllAnnosOnPosition(aJCases, targetAnnosPerUser, users, targetFs);
                            // do not agree on targets
                            if (!isAgree(targetFs, targetAnnosPerUser)) {
                                linkFSes.remove(roleFs);
                            }
                        }
                    } else // do not agree on some role features
                    {
                        linkFSes.remove(roleFs);
                    }
                }
                ArrayFS array = baseFs.getCAS().createArrayFS(linkFSes.size());
                array.copyFromArray(linkFSes.toArray(new FeatureStructure[linkFSes.size()]), 0, 0, linkFSes.size());
                baseFs.setFeatureValue(roleFeature, array);
            }
        }
    }
    return aJCases.get(WebAnnoConst.CURATION_USER);
}

Example 15 with Lemma

use of de.tudarmstadt.ukp.dkpro.core.api.segmentation.type.Lemma in project webanno by webanno.

the class ConstraintsGeneratorTest method testSimpleFeature.

@Test
public void testSimpleFeature() throws Exception {
    ConstraintsGrammar parser = new ConstraintsGrammar(new FileInputStream("src/test/resources/rules/9.rules"));
    Parse p = parser.Parse();
    ParsedConstraints constraints = p.accept(new ParserVisitor());
    JCas jcas = JCasFactory.createJCas();
    jcas.setDocumentText("is");
    Lemma lemma = new Lemma(jcas, 0, 2);
    lemma.setValue("be");
    lemma.addToIndexes();
    Evaluator constraintsEvaluator = new ValuesGenerator();
    List<PossibleValue> possibleValues = constraintsEvaluator.generatePossibleValues(lemma, "value", constraints);
    List<PossibleValue> expectedOutput = new LinkedList<>();
    expectedOutput.add(new PossibleValue("be", true));
    assertEquals(expectedOutput, possibleValues);
}