use of org.apache.stanbol.entityhub.servicesapi.model.Text in project stanbol by apache.
the class KeywordLinkingEngine method writeEnhancements.
/**
* Writes the Enhancements for the {@link LinkedEntity LinkedEntities}
* extracted from the parsed ContentItem
* @param ci
* @param linkedEntities
* @param language
*/
private void writeEnhancements(ContentItem ci, Collection<LinkedEntity> linkedEntities, String language) {
Language languageObject = null;
if (language != null && !language.isEmpty()) {
languageObject = new Language(language);
}
Graph metadata = ci.getMetadata();
for (LinkedEntity linkedEntity : linkedEntities) {
Collection<IRI> textAnnotations = new ArrayList<IRI>(linkedEntity.getOccurrences().size());
//first create the TextAnnotations for the Occurrences
for (Occurrence occurrence : linkedEntity.getOccurrences()) {
IRI textAnnotation = EnhancementEngineHelper.createTextEnhancement(ci, this);
textAnnotations.add(textAnnotation);
metadata.add(new TripleImpl(textAnnotation, Properties.ENHANCER_START, literalFactory.createTypedLiteral(occurrence.getStart())));
metadata.add(new TripleImpl(textAnnotation, Properties.ENHANCER_END, literalFactory.createTypedLiteral(occurrence.getEnd())));
metadata.add(new TripleImpl(textAnnotation, Properties.ENHANCER_SELECTION_CONTEXT, new PlainLiteralImpl(occurrence.getContext(), languageObject)));
metadata.add(new TripleImpl(textAnnotation, Properties.ENHANCER_SELECTED_TEXT, new PlainLiteralImpl(occurrence.getSelectedText(), languageObject)));
metadata.add(new TripleImpl(textAnnotation, Properties.ENHANCER_CONFIDENCE, literalFactory.createTypedLiteral(linkedEntity.getScore())));
for (IRI dcType : linkedEntity.getTypes()) {
metadata.add(new TripleImpl(textAnnotation, Properties.DC_TYPE, dcType));
}
}
//now the EntityAnnotations for the Suggestions
for (Suggestion suggestion : linkedEntity.getSuggestions()) {
IRI entityAnnotation = EnhancementEngineHelper.createEntityEnhancement(ci, this);
//should we use the label used for the match, or search the
//representation for the best label ... currently its the matched one
Text label = suggestion.getBestLabel(linkerConfig.getNameField(), language);
metadata.add(new TripleImpl(entityAnnotation, Properties.ENHANCER_ENTITY_LABEL, label.getLanguage() == null ? new PlainLiteralImpl(label.getText()) : new PlainLiteralImpl(label.getText(), new Language(label.getLanguage()))));
metadata.add(new TripleImpl(entityAnnotation, Properties.ENHANCER_ENTITY_REFERENCE, new IRI(suggestion.getRepresentation().getId())));
Iterator<Reference> suggestionTypes = suggestion.getRepresentation().getReferences(linkerConfig.getTypeField());
while (suggestionTypes.hasNext()) {
metadata.add(new TripleImpl(entityAnnotation, Properties.ENHANCER_ENTITY_TYPE, new IRI(suggestionTypes.next().getReference())));
}
metadata.add(new TripleImpl(entityAnnotation, Properties.ENHANCER_CONFIDENCE, literalFactory.createTypedLiteral(suggestion.getScore())));
for (IRI textAnnotation : textAnnotations) {
metadata.add(new TripleImpl(entityAnnotation, Properties.DC_RELATION, textAnnotation));
}
//add the name of the ReferencedSite providing this suggestion
metadata.add(new TripleImpl(entityAnnotation, new IRI(RdfResourceEnum.site.getUri()), new PlainLiteralImpl(referencedSiteName)));
//add the RDF data for entities
if (dereferenceEntitiesState) {
metadata.addAll(RdfValueFactory.getInstance().toRdfRepresentation(suggestion.getRepresentation()).getRdfGraph());
}
}
}
}
use of org.apache.stanbol.entityhub.servicesapi.model.Text in project stanbol by apache.
the class Suggestion method getBestLabel.
/**
* Getter for the best label in the given language
* @param suggestion the suggestion
* @param nameField the field used to search for labels
* @param language the language
* @return the best match or {@link Suggestion#getMatchedLabel()} if non is found
*/
public Text getBestLabel(String nameField, String language) {
Representation rep = getRepresentation();
//start with the matched label -> so if we do not find a better one
//we will use the matched!
Text label = this.label;
// 1. check if the returned Entity does has a label -> if not return null
// add labels (set only a single label. Use "en" if available!
Iterator<Text> labels = rep.getText(nameField);
boolean matchFound = false;
while (labels.hasNext() && !matchFound) {
Text actLabel = labels.next();
if (label == null) {
//take any label at first
label = actLabel;
}
//now we have already a label check the language
String actLang = actLabel.getLanguage();
//use startWith to match also en-GB and en-US ...
if (actLang != null && actLang.startsWith(language)) {
//prefer labels with the correct language
label = actLabel;
if (this.label.getText().equalsIgnoreCase(label.getText())) {
//found label in that language that exactly matches the
//label used to match the text
matchFound = true;
}
}
}
if (label == null) {
//if no label was found ... return the one used for the match
label = getMatchedLabel();
}
return label;
}
use of org.apache.stanbol.entityhub.servicesapi.model.Text in project stanbol by apache.
the class MockEntityCorefDbpediaSite method findEntities.
@SuppressWarnings("deprecation")
@Override
public QueryResultList<Entity> findEntities(FieldQuery query) throws SiteException {
TextConstraint labelConstraint = (TextConstraint) query.getConstraint(RDFS_LABEL.getUnicodeString());
for (Entity entity : entities.values()) {
Iterator<Object> entityAttributes = entity.getRepresentation().get(RDFS_LABEL.getUnicodeString());
while (entityAttributes.hasNext()) {
Text entityAttribute = (Text) entityAttributes.next();
if (entityAttribute.getText().equals(labelConstraint.getText())) {
Collection<Entity> retEntities = new ArrayList<>(1);
retEntities.add(entity);
return new QueryResultListImpl<Entity>(null, retEntities, Entity.class);
}
}
}
return null;
}
use of org.apache.stanbol.entityhub.servicesapi.model.Text in project stanbol by apache.
the class FreebaseAbbrevationProcessor method process.
@Override
public Representation process(Representation rep) {
Iterator<Text> aliases = rep.getText(FB_ALIAS);
while (aliases.hasNext()) {
Text alias = aliases.next();
if (StringUtils.isAllUpperCase(alias.getText())) {
rep.add(FB_NAME, alias);
rep.add(RDFS_LABEL, alias);
}
}
return rep;
}
use of org.apache.stanbol.entityhub.servicesapi.model.Text in project stanbol by apache.
the class RdfIndexingSourceTest method testText.
private void testText(Representation rep) {
for (Iterator<String> fields = rep.getFieldNames(); fields.hasNext(); ) {
String field = fields.next();
Iterator<Text> values = rep.getText(field);
// assertTrue(values.hasNext());
while (values.hasNext()) {
Text text = values.next();
assertNotNull(text);
String lang = text.getLanguage();
//log.info(text.getText()+" | "+text.getLanguage()+" | "+text.getText().endsWith("@"+lang));
//this texts that the text does not contain the @{lang} as added by
//the toString method of the RDF Literal java class
assertFalse("Labels MUST NOT end with the Language! value=" + text.getText(), text.getText().endsWith("@" + lang));
}
}
}
Aggregations