/**
* This tests if the Enhancements created by the Engine confirm to the rules defined for the
* Stanbol Enhancement Structure.
*
* @throws IOException
* @throws EngineException
*/
@Test
public void testEngine() throws IOException, EngineException {
EntityLinkerConfig linkerConfig = new EntityLinkerConfig();
linkerConfig.setRedirectProcessingMode(RedirectProcessingMode.FOLLOW);
KeywordLinkingEngine engine =
KeywordLinkingEngine.createInstance(
openNLP, searcher, new TextAnalyzerConfig(), linkerConfig);
engine.referencedSiteName = TEST_REFERENCED_SITE_NAME;
ContentItem ci = ciFactory.createContentItem(new StringSource(TEST_TEXT));
// tells the engine that this is an English text
ci.getMetadata().add(new TripleImpl(ci.getUri(), DC_LANGUAGE, new PlainLiteralImpl("en")));
// compute the enhancements
engine.computeEnhancements(ci);
// validate the enhancement results
Map<UriRef, Resource> expectedValues = new HashMap<UriRef, Resource>();
expectedValues.put(ENHANCER_EXTRACTED_FROM, ci.getUri());
expectedValues.put(
DC_CREATOR, LiteralFactory.getInstance().createTypedLiteral(engine.getClass().getName()));
// adding null as expected for confidence makes it a required property
expectedValues.put(Properties.ENHANCER_CONFIDENCE, null);
// validate create fise:TextAnnotations
int numTextAnnotations =
validateAllTextAnnotations(ci.getMetadata(), TEST_TEXT, expectedValues);
assertEquals("Four fise:TextAnnotations are expected by this Test", 4, numTextAnnotations);
// validate create fise:EntityAnnotations
int numEntityAnnotations = validateAllEntityAnnotations(ci, expectedValues);
assertEquals("Five fise:EntityAnnotations are expected by this Test", 5, numEntityAnnotations);
}
/**
* Similar to {@link
* EnhancementStructureHelper#validateAllEntityAnnotations(org.apache.clerezza.rdf.core.TripleCollection,
* Map)} but in addition checks fise:confidence [0..1] and entityhub:site properties
*
* @param ci
* @param expectedValues
* @return
*/
private static int validateAllEntityAnnotations(
ContentItem ci, Map<UriRef, Resource> expectedValues) {
Iterator<Triple> entityAnnotationIterator =
ci.getMetadata().filter(null, RDF_TYPE, ENHANCER_ENTITYANNOTATION);
int entityAnnotationCount = 0;
while (entityAnnotationIterator.hasNext()) {
UriRef entityAnnotation = (UriRef) entityAnnotationIterator.next().getSubject();
// test if selected Text is added
validateEntityAnnotation(ci.getMetadata(), entityAnnotation, expectedValues);
// validate also that the confidence is between [0..1]
Iterator<Triple> confidenceIterator =
ci.getMetadata().filter(entityAnnotation, ENHANCER_CONFIDENCE, null);
// Confidence is now checked by the EnhancementStructureHelper (STANBOL-630)
// assertTrue("Expected fise:confidence value is missing (entityAnnotation "
// +entityAnnotation+")",confidenceIterator.hasNext());
// Double confidence = LiteralFactory.getInstance().createObject(Double.class,
// (TypedLiteral)confidenceIterator.next().getObject());
// assertTrue("fise:confidence MUST BE <= 1 (value= '"+confidence
// + "',entityAnnotation " +entityAnnotation+")",
// 1.0 >= confidence.doubleValue());
// assertTrue("fise:confidence MUST BE >= 0 (value= '"+confidence
// +"',entityAnnotation "+entityAnnotation+")",
// 0.0 <= confidence.doubleValue());
// Test the entityhub:site property (STANBOL-625)
UriRef ENTITYHUB_SITE = new UriRef(RdfResourceEnum.site.getUri());
Iterator<Triple> entitySiteIterator =
ci.getMetadata().filter(entityAnnotation, ENTITYHUB_SITE, null);
assertTrue(
"Expected entityhub:site value is missing (entityAnnotation " + entityAnnotation + ")",
entitySiteIterator.hasNext());
Resource siteResource = entitySiteIterator.next().getObject();
assertTrue("entityhub:site values MUST BE Literals", siteResource instanceof Literal);
assertEquals(
"'" + TEST_REFERENCED_SITE_NAME + "' is expected as " + "entityhub:site value",
TEST_REFERENCED_SITE_NAME,
((Literal) siteResource).getLexicalForm());
assertFalse("entityhub:site MUST HAVE only a single value", entitySiteIterator.hasNext());
entityAnnotationCount++;
}
return entityAnnotationCount;
}
/**
* Creates an {@link AnalysedText} instance for the parsed {@link Blob} and registers itself as
* {@link ContentItem#addPart(org.apache.clerezza.rdf.core.UriRef, Object) ContentPart} with the
* {@link UriRef} {@link AnalysedText#ANALYSED_TEXT_URI} to the parsed {@link ContentItem}.
*
* <p>If already a ContentPart with the given UriRef is registered this Method will throw an
* {@link IllegalStateException}.
*
* @param ci the ContentItem to register the created {@link AnalysedText} instance
* @param blob the analysed {@link Blob}
* @return the created {@link AnalysedText}
* @throws IllegalArgumentException of <code>null</code> is parsed as ContentItem or Blob
* @throws IllegalStateException if there is already an ContentPart is registered for {@link
* AnalysedText#ANALYSED_TEXT_URI} with the parsed ContentItem.
* @throws IOException on any error while reading data from the parsed blob
*/
public final AnalysedText createAnalysedText(ContentItem ci, Blob blob) throws IOException {
ci.getLock().readLock().lock();
try {
AnalysedText existing = ci.getPart(AnalysedText.ANALYSED_TEXT_URI, AnalysedText.class);
throw new IllegalStateException(
"The AnalysedText ContentPart already exists (impl: "
+ existing.getClass().getSimpleName()
+ "| blob: "
+ existing.getBlob().getMimeType()
+ ")");
} catch (NoSuchPartException e) {
// this is the expected case
} catch (ClassCastException e) {
throw new IllegalStateException(
"A ContentPart with the URI '"
+ AnalysedText.ANALYSED_TEXT_URI
+ "' already exists but the parts "
+ "type is not compatible with "
+ AnalysedText.class.getSimpleName()
+ "!",
e);
} finally {
ci.getLock().readLock().unlock();
}
// create the Analysed text
AnalysedText at = createAnalysedText(blob);
ci.getLock().writeLock().lock();
try {
// NOTE: there is a possibility that an other thread has added
// the contentpart
ci.addPart(AnalysedText.ANALYSED_TEXT_URI, at);
} finally {
ci.getLock().writeLock().unlock();
}
return at;
}
