Examples with DuplicateDataConfig org.apache.rya.indexing.smarturi.duplication.conf.DuplicateDataConfig used on opensource projects

Example 1 with DuplicateDataConfig

use of org.apache.rya.indexing.smarturi.duplication.conf.DuplicateDataConfig in project incubator-rya by apache.

the class DuplicateDataDetectorIT method testCreateEntityNearDuplicateConfigDisabled.

@Test
public void testCreateEntityNearDuplicateConfigDisabled() throws EntityStorageException, TypeStorageException, ConfigurationException, ObjectStorageException {
    // Create the types the Entity uses.
    final TypeStorage typeStorage = new MongoTypeStorage(super.getMongoClient(), RYA_INSTANCE_NAME);
    final Type personType = createPersonType();
    final Type employeeType = createEmployeeType();
    typeStorage.create(personType);
    typeStorage.create(employeeType);
    final Optional<Type> storedPersonType = typeStorage.get(personType.getId());
    final Optional<Type> storedEmployeeType = typeStorage.get(employeeType.getId());
    assertTrue(storedPersonType.isPresent());
    assertTrue(storedEmployeeType.isPresent());
    // Create it.
    final DuplicateDataConfig duplicateDataConfig = new DuplicateDataConfig(// boolean
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // byte
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // date
    new Tolerance(500.0, ToleranceType.DIFFERENCE), // double
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // float
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // integer
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // long
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // short
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // string
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // uri
    new Tolerance(1.0, ToleranceType.DIFFERENCE), new HashMap<String, List<String>>(), false);
    final DuplicateDataDetector duplicateDataDetector = new DuplicateDataDetector(duplicateDataConfig);
    final EntityStorage entityStorage = new MongoEntityStorage(super.getMongoClient(), RYA_INSTANCE_NAME, duplicateDataDetector);
    final Entity bobEntity = createBobEntity();
    entityStorage.create(bobEntity);
    assertTrue(entityStorage.get(bobEntity.getSubject()).isPresent());
    final Builder duplicateBobBuilder = Entity.builder(createBobEntity());
    duplicateBobBuilder.setSubject(createRyaUri("Robert"));
    // Modify a property for each type that is within tolerance
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 41)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(250.76f)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(72.499)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(50001)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 2)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(123456789013L)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 1).minusYears(40))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 1))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(true)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bob.smitch01@gmail.com"))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("124 Fake St. Washington, DC 20024")));
    duplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 556)));
    final Entity duplicateBobEntity = duplicateBobBuilder.build();
    // Data duplication detection is disabled so it will be created.
    try {
        entityStorage.create(duplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        fail();
    }
    assertTrue(entityStorage.get(duplicateBobEntity.getSubject()).isPresent());
    final Builder notDuplicateBobBuilder = Entity.builder(createBobEntity());
    notDuplicateBobBuilder.setSubject(createRyaUri("Not Bob"));
    // Modify a property for each type that is within tolerance
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 50)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(300.0f)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(100.0)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(60000)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 5)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(9L)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 10000000L).minusYears(40))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 10000000L))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(false)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bad.email.address@gmail.com"))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("123456789 Fake St. Washington, DC 20024")));
    notDuplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 1000)));
    final Entity notDuplicateBobEntity = notDuplicateBobBuilder.build();
    // Data duplication detection is disabled so it will be created.
    try {
        entityStorage.create(notDuplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        fail();
    }
    assertTrue(entityStorage.get(notDuplicateBobEntity.getSubject()).isPresent());
}

Example 2 with DuplicateDataConfig

use of org.apache.rya.indexing.smarturi.duplication.conf.DuplicateDataConfig in project incubator-rya by apache.

the class DuplicateDataDetectorIT method testCreateEntityNearDuplicate.

@Test
public void testCreateEntityNearDuplicate() throws EntityStorageException, TypeStorageException, ObjectStorageException {
    // Create the types the Entity uses.
    final TypeStorage typeStorage = new MongoTypeStorage(super.getMongoClient(), RYA_INSTANCE_NAME);
    final Type personType = createPersonType();
    final Type employeeType = createEmployeeType();
    typeStorage.create(personType);
    typeStorage.create(employeeType);
    final Optional<Type> storedPersonType = typeStorage.get(personType.getId());
    final Optional<Type> storedEmployeeType = typeStorage.get(employeeType.getId());
    assertTrue(storedPersonType.isPresent());
    assertTrue(storedEmployeeType.isPresent());
    // Create it.
    final DuplicateDataConfig duplicateDataConfig = new DuplicateDataConfig(// boolean
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // byte
    new Tolerance(0.0, ToleranceType.DIFFERENCE), // date
    new Tolerance(500.0, ToleranceType.DIFFERENCE), // double
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // float
    new Tolerance(0.0001, ToleranceType.PERCENTAGE), // integer
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // long
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // short
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // string
    new Tolerance(1.0, ToleranceType.DIFFERENCE), // uri
    new Tolerance(1.0, ToleranceType.DIFFERENCE), new HashMap<String, List<String>>(), true);
    final DuplicateDataDetector duplicateDataDetector = new DuplicateDataDetector(duplicateDataConfig);
    final EntityStorage entityStorage = new MongoEntityStorage(super.getMongoClient(), RYA_INSTANCE_NAME, duplicateDataDetector);
    final Entity bobEntity = createBobEntity();
    entityStorage.create(bobEntity);
    assertTrue(entityStorage.get(bobEntity.getSubject()).isPresent());
    final Builder duplicateBobBuilder = Entity.builder(createBobEntity());
    duplicateBobBuilder.setSubject(createRyaUri("Robert"));
    // Modify a property for each type that is within tolerance
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 41)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(250.76f)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(72.499)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(50001)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 2)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(123456789013L)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 1).minusYears(40))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 1))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(true)));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bob.smitch01@gmail.com"))));
    duplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("124 Fake St. Washington, DC 20024")));
    duplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 556)));
    final Entity duplicateBobEntity = duplicateBobBuilder.build();
    // Try to create another entity that's considered a duplicate.
    // It will NOT be be created.
    boolean hasDuplicate = false;
    try {
        entityStorage.create(duplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        hasDuplicate = true;
    }
    assertTrue(hasDuplicate);
    assertFalse(entityStorage.get(duplicateBobEntity.getSubject()).isPresent());
    final Builder notDuplicateBobBuilder = Entity.builder(createBobEntity());
    notDuplicateBobBuilder.setSubject(createRyaUri("Not Bob"));
    // Modify a property for each type that is within tolerance
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_AGE, shortRyaType((short) 50)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_WEIGHT, floatRyaType(300.0f)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_HEIGHT, doubleRyaType(100.0)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_INCOME, intRyaType(60000)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_NUMBER_OF_CHILDREN, byteRyaType((byte) 5)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_LICENSE_NUMBER, longRyaType(9L)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_DATE_OF_BIRTH, dateRyaType(new DateTime(NOW.getTime() - 10000000L).minusYears(40))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EXPIRATION_DATE, dateRyaType(new Date(NOW.getTime() - 10000000L))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_GLASSES, booleanRyaType(false)));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_EMAIL_ADDRESS, uriRyaType(new URIImpl("mailto:bad.email.address@gmail.com"))));
    notDuplicateBobBuilder.setProperty(PERSON_TYPE_URI, new Property(HAS_ADDRESS, stringRyaType("123456789 Fake St. Washington, DC 20024")));
    notDuplicateBobBuilder.setProperty(EMPLOYEE_TYPE_URI, new Property(HAS_EXTENSION, shortRyaType((short) 1000)));
    final Entity notDuplicateBobEntity = notDuplicateBobBuilder.build();
    // It will be created.
    try {
        entityStorage.create(notDuplicateBobEntity);
    } catch (final EntityNearDuplicateException e) {
        fail();
    }
    assertTrue(entityStorage.get(notDuplicateBobEntity.getSubject()).isPresent());
}

Example 3 with DuplicateDataConfig

use of org.apache.rya.indexing.smarturi.duplication.conf.DuplicateDataConfig in project incubator-rya by apache.

the class DuplicateDataDetectorIT method testReadConfigFile.

@Test
public void testReadConfigFile() throws SmartUriException, ConfigurationException {
    final DuplicateDataConfig duplicateDataConfig = new DuplicateDataConfig();
    assertNotNull(duplicateDataConfig.getBooleanTolerance());
    assertNotNull(duplicateDataConfig.getByteTolerance());
    assertNotNull(duplicateDataConfig.getDateTolerance());
    assertNotNull(duplicateDataConfig.getDoubleTolerance());
    assertNotNull(duplicateDataConfig.getFloatTolerance());
    assertNotNull(duplicateDataConfig.getIntegerTolerance());
    assertNotNull(duplicateDataConfig.getLongTolerance());
    assertNotNull(duplicateDataConfig.getShortTolerance());
    assertNotNull(duplicateDataConfig.getStringTolerance());
    assertNotNull(duplicateDataConfig.getUriTolerance());
    assertNotNull(duplicateDataConfig.getEquivalentTermsMap());
    assertNotNull(duplicateDataConfig.isDetectionEnabled());
}