Examples with TripleImpl org.apache.clerezza.commons.rdf.impl.utils.TripleImpl used on opensource projects

Example 86 with TripleImpl

use of org.apache.clerezza.commons.rdf.impl.utils.TripleImpl in project stanbol by apache.

the class EntityLinker method processRedirects.

/**
     * Processes {@link EntitySearcher#getRedirectField() redirect field} values for
     * the parsed suggestions based on the {@link RedirectProcessingMode}
     * as configured in the {@link #config}.<p>
     * The results of this method are stored within the parsed {@link Suggestion}s
     * @param suggestion The suggestion to process.
     * @throws EntitySearcherException 
     */
private void processRedirects(Suggestion suggestion) throws EntitySearcherException {
    //if mode is IGNORE -> nothing to do
    if (linkerConfig.getRedirectProcessingMode() == RedirectProcessingMode.IGNORE) {
        return;
    }
    //therefore there is a small internal state that stores this information
    if (suggestion.isRedirectedProcessed()) {
        //Redirects for ResultMatch are already processed ... ignore
        return;
    }
    Entity result = suggestion.getResult();
    Iterator<IRI> redirects = result.getReferences(linkerConfig.getRedirectField());
    switch(linkerConfig.getRedirectProcessingMode()) {
        case ADD_VALUES:
            Graph entityData = result.getData();
            IRI entityUri = result.getUri();
            while (redirects.hasNext()) {
                IRI redirect = redirects.next();
                if (redirect != null) {
                    Entity redirectedEntity = entitySearcher.get(redirect, linkerConfig.getSelectedFields());
                    if (redirectedEntity != null) {
                        for (Iterator<Triple> data = redirectedEntity.getData().filter(redirectedEntity.getUri(), null, null); data.hasNext(); ) {
                            Triple t = data.next();
                            entityData.add(new TripleImpl(entityUri, t.getPredicate(), t.getObject()));
                        }
                    }
                    //set that the redirects where searched for this result
                    suggestion.setRedirectProcessed(true);
                }
            }
        case FOLLOW:
            while (redirects.hasNext()) {
                IRI redirect = redirects.next();
                if (redirect != null) {
                    Entity redirectedEntity = entitySearcher.get(redirect, linkerConfig.getSelectedFields());
                    if (redirectedEntity != null) {
                        suggestion.setRedirect(redirectedEntity);
                    }
                }
            }
        //nothing to do
        default:
    }
}

Example 87 with TripleImpl

use of org.apache.clerezza.commons.rdf.impl.utils.TripleImpl in project stanbol by apache.

the class EntityLinkingEngineTest method testEngine.

/**
     * 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);
    //this is assumed by this test
    linkerConfig.setMinFoundTokens(2);
    EntityLinkingEngine engine = new EntityLinkingEngine("dummy", searcher, new TextProcessingConfig(), linkerConfig, labelTokenizer);
    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")));
    //and add the AnalysedText instance used for this test
    ci.addPart(AnalysedText.ANALYSED_TEXT_URI, TEST_ANALYSED_TEXT);
    //compute the enhancements
    engine.computeEnhancements(ci);
    //validate the enhancement results
    Map<IRI, RDFTerm> expectedValues = new HashMap<IRI, RDFTerm>();
    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);
}

Example 88 with TripleImpl

use of org.apache.clerezza.commons.rdf.impl.utils.TripleImpl in project stanbol by apache.

the class NIFHelper method writePos.

/**
     * Writes the {@link NlpAnnotations#POS_ANNOTATION} as NIF 1.0 to the parsed
     * RDF graph by using the parsed segmentUri as subject
     * @param graph the graph
     * @param annotated the annotated element (e.g. a {@link Token})
     * @param segmentUri the URI of the resource representing the parsed 
     * annotated element in the graph
     */
public static void writePos(Graph graph, Annotated annotated, IRI segmentUri) {
    Value<PosTag> posTag = annotated.getAnnotation(NlpAnnotations.POS_ANNOTATION);
    if (posTag != null) {
        if (posTag.value().isMapped()) {
            for (Pos pos : posTag.value().getPos()) {
                graph.add(new TripleImpl(segmentUri, SsoOntology.oliaLink.getUri(), pos.getUri()));
            }
            for (LexicalCategory cat : posTag.value().getCategories()) {
                graph.add(new TripleImpl(segmentUri, SsoOntology.oliaLink.getUri(), cat.getUri()));
            }
        }
        graph.add(new TripleImpl(segmentUri, SsoOntology.posTag.getUri(), lf.createTypedLiteral(posTag.value().getTag())));
        graph.add(new TripleImpl(segmentUri, ENHANCER_CONFIDENCE, lf.createTypedLiteral(posTag.probability())));
    }
}

Example 89 with TripleImpl

use of org.apache.clerezza.commons.rdf.impl.utils.TripleImpl in project stanbol by apache.

the class EnhancementEngineHelper method setOccurrence.

/**
     * This method sets the fise:start, fise:end, fise:selection-prefix, 
     * fise:selected-text and fise:selection-suffix properties for the 
     * parsed fise:TextAnnotation instance according to the parsed parameters.<p>
     * While it is intended to be used for TextAnnotations this method can also
     * be used to add the mentioned properties to {@link IRI}s with different
     * type.<p>
     * <b>NOTE</b> the <code>allowSelectionHeadTail</code>: This parameter allows
     * to deactivate the usage of fise:selection-head and fise:selection-tail.
     * Typically users should parse <code>false</code> in case of 'named entities'
     * and <code>true</code> in case sections of the text (e.g. phrases, sentences,
     * chapters ...) are selected.
     * @param metadata The RDF graph to add the information
     * @param textAnnotation the IRI of the fise:TextAnnotation
     * @param content the plain text content as String
     * @param start the start index of the occurrence 
     * @param end the end index of the occurrence
     * @param lang the lanugage of the content or <code>null</code> if not known
     * @param prefixSuffixSize the size of the prefix, suffix. If the parsed
     * value &lt; 3 than the default 10 is used.
     * @param allowSelectionHeadTail if <code>true</code> the fise:selection-head
     * and fise:selection-tail properties are used instead of fise:selected-text
     * if the selected text is longer as <code>Math.max(30, prefixSuffixSize*5);</code>.
     * If <code>false</code> the fise:selected-text is added regardless of the
     * size of the selected area.
     * @since 0.11.0
     */
public static void setOccurrence(Graph metadata, IRI textAnnotation, String content, Integer start, Integer end, Language lang, int prefixSuffixSize, boolean allowSelectionHeadTail) {
    //set start, end
    metadata.add(new TripleImpl(textAnnotation, ENHANCER_START, lf.createTypedLiteral(start)));
    metadata.add(new TripleImpl(textAnnotation, ENHANCER_END, lf.createTypedLiteral(end)));
    //set selection prefix and suffix (TextAnnotation new model)
    prefixSuffixSize = prefixSuffixSize < MIN_PREFIX_SUFFIX_SIZE ? DEFAULT_PREFIX_SUFFIX_LENGTH : prefixSuffixSize;
    metadata.add(new TripleImpl(textAnnotation, ENHANCER_SELECTION_PREFIX, new PlainLiteralImpl(content.substring(Math.max(0, start - prefixSuffixSize), start), lang)));
    metadata.add(new TripleImpl(textAnnotation, ENHANCER_SELECTION_SUFFIX, new PlainLiteralImpl(content.substring(end, Math.min(content.length(), end + prefixSuffixSize)), lang)));
    //set the selected text (or alternatively head and tail)
    int maxSelectedTextSize = Math.max(MIN_SELECTEN_HEAD_TAIL_USAGE_LENGTH, prefixSuffixSize * 5);
    if (!allowSelectionHeadTail || end - start <= maxSelectedTextSize) {
        metadata.add(new TripleImpl(textAnnotation, ENHANCER_SELECTED_TEXT, new PlainLiteralImpl(content.substring(start, end), lang)));
    } else {
        //selected area to long for fise:selected-text
        //use fise:selection-head and fise:selection-tail instead
        metadata.add(new TripleImpl(textAnnotation, ENHANCER_SELECTION_HEAD, new PlainLiteralImpl(content.substring(start, start + prefixSuffixSize), lang)));
        metadata.add(new TripleImpl(textAnnotation, ENHANCER_SELECTION_TAIL, new PlainLiteralImpl(content.substring(end - prefixSuffixSize, end), lang)));
    }
}

Example 90 with TripleImpl

use of org.apache.clerezza.commons.rdf.impl.utils.TripleImpl in project stanbol by apache.

the class EnhancementEngineHelper method set.

/**
     * Replaces all current values of the property for the resource
     * with the parsed values
     * @param graph the graph
     * @param resource the resource
     * @param property the property
     * @param value the value. In case it is an instance of {@link RDFTerm} it
     * is directly added to the graph. Otherwise the parsed {@link LiteralFactory}
     * is used to create a {@link TypedLiteral} for the parsed value.
     * @param literalFactory the {@link LiteralFactory} used in case the parsed
     * value is not an {@link RDFTerm}
     */
public static void set(Graph graph, BlankNodeOrIRI resource, IRI property, Collection<?> values, LiteralFactory literalFactory) {
    Iterator<Triple> currentValues = graph.filter(resource, property, null);
    while (currentValues.hasNext()) {
        currentValues.next();
        currentValues.remove();
    }
    if (values != null) {
        for (Object value : values) {
            if (value instanceof RDFTerm) {
                graph.add(new TripleImpl(resource, property, (RDFTerm) value));
            } else if (value != null) {
                graph.add(new TripleImpl(resource, property, literalFactory.createTypedLiteral(value)));
            }
        }
    }
}