Search in sources :

Example 1 with TextProcessingConfig

use of org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig in project stanbol by apache.

the class FstLinkingEngineTest method testFstLinkingWithProperNouns.

@Test
public void testFstLinkingWithProperNouns() throws Exception {
    Dictionary<String, Object> dict = new Hashtable<String, Object>();
    dict.put(PROCESSED_LANGUAGES, Arrays.asList("en;lmmtip;uc=LINK;prob=0.75;pprob=0.75"));
    dict.put(PROCESS_ONLY_PROPER_NOUNS_STATE, true);
    TextProcessingConfig tpc = TextProcessingConfig.createInstance(dict);
    EntityLinkerConfig elc = new EntityLinkerConfig();
    // this is assumed by this test
    elc.setMinFoundTokens(2);
    elc.setRedirectProcessingMode(RedirectProcessingMode.FOLLOW);
    FstLinkingEngine engine = new FstLinkingEngine("proper-noun-linking", LinkingModeEnum.LINKABLE_TOKEN, fstConfig, tpc, elc, null);
    processConentItem(engine);
    validateEnhancements(Arrays.asList("Chancellor", "Angela Merkel", "Greece", "Greeks", "Germany", "SPD"), Arrays.asList(DBPEDIA + "Christian_Democratic_Union_(Germany)", DBPEDIA + "Angela_Merkel", DBPEDIA + "Greece", DBPEDIA + "Germany", DBPEDIA + "Social_Democratic_Party_of_Germany"));
}
Also used : EntityLinkerConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.EntityLinkerConfig) TextProcessingConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig) Hashtable(java.util.Hashtable) FstLinkingEngine(org.apache.stanbol.enhancer.engines.lucenefstlinking.FstLinkingEngine) Test(org.junit.Test)

Example 2 with TextProcessingConfig

use of org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig 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);
}
Also used : IRI(org.apache.clerezza.commons.rdf.IRI) EntityLinkerConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.EntityLinkerConfig) TextProcessingConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig) PlainLiteralImpl(org.apache.clerezza.commons.rdf.impl.utils.PlainLiteralImpl) HashMap(java.util.HashMap) RDFTerm(org.apache.clerezza.commons.rdf.RDFTerm) StringSource(org.apache.stanbol.enhancer.servicesapi.impl.StringSource) TripleImpl(org.apache.clerezza.commons.rdf.impl.utils.TripleImpl) ContentItem(org.apache.stanbol.enhancer.servicesapi.ContentItem) Test(org.junit.Test)

Example 3 with TextProcessingConfig

use of org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig in project stanbol by apache.

the class EntityhubLinkingEngine method activate.

@Activate
@SuppressWarnings("unchecked")
protected void activate(ComponentContext ctx) throws ConfigurationException {
    Dictionary<String, Object> properties = ctx.getProperties();
    bundleContext = ctx.getBundleContext();
    EntityLinkerConfig linkerConfig = EntityLinkerConfig.createInstance(properties, prefixService);
    TextProcessingConfig textProcessingConfig = TextProcessingConfig.createInstance(properties);
    Object value = properties.get(SITE_ID);
    // init the EntitySource
    if (value == null) {
        throw new ConfigurationException(SITE_ID, "The ID of the Referenced Site is a required Parameter and MUST NOT be NULL!");
    }
    siteName = value.toString();
    if (siteName.isEmpty()) {
        throw new ConfigurationException(SITE_ID, "The ID of the Referenced Site is a required Parameter and MUST NOT be an empty String!");
    }
    // get the metadata later set to the enhancement engine
    String engineName;
    engineMetadata = new Hashtable<String, Object>();
    value = properties.get(PROPERTY_NAME);
    if (value == null || value.toString().isEmpty()) {
        throw new ConfigurationException(PROPERTY_NAME, "The EnhancementEngine name MUST BE configured!");
    } else {
        engineName = value.toString();
    }
    engineMetadata.put(PROPERTY_NAME, value);
    value = properties.get(Constants.SERVICE_RANKING);
    engineMetadata.put(Constants.SERVICE_RANKING, value == null ? Integer.valueOf(0) : value);
    // init the tracking entity searcher
    trackedServiceCount = 0;
    if (Entityhub.ENTITYHUB_IDS.contains(siteName.toLowerCase())) {
        entitySearcher = new EntityhubSearcher(bundleContext, 10, this);
    } else {
        entitySearcher = new ReferencedSiteSearcher(bundleContext, siteName, 10, this);
    }
    labelTokenizerTracker = new ServiceTracker(bundleContext, LabelTokenizer.class.getName(), new ServiceTrackerCustomizer() {

        @Override
        public Object addingService(ServiceReference reference) {
            Object service = bundleContext.getService(reference);
            if (service == null) {
                return service;
            }
            synchronized (labelTokenizersRefs) {
                labelTokenizersRefs.add(reference);
                ServiceReference higest;
                try {
                    higest = labelTokenizersRefs.last();
                } catch (NoSuchElementException e) {
                    higest = null;
                }
                EntityLinkingEngine engine = entityLinkingEngine;
                ServiceTracker tracker = labelTokenizerTracker;
                if (engine != null && tracker != null) {
                    LabelTokenizer lt = (LabelTokenizer) (reference.equals(higest) || higest == null ? service : tracker.getService(higest));
                    if (!lt.equals(engine.getLabelTokenizer())) {
                        log.info(" ... setting LabelTokenizer of Engine '{}' to {}", engine.getName(), lt);
                        engine.setLabelTokenizer(lt);
                    }
                }
            // if engine or tracker is null deactivate was already called
            }
            return service;
        }

        @Override
        public void removedService(ServiceReference reference, Object service) {
            synchronized (labelTokenizersRefs) {
                // override
                labelTokenizersRefs.remove(reference);
                EntityLinkingEngine engine = entityLinkingEngine;
                ServiceTracker tracker = labelTokenizerTracker;
                if (engine != null && tracker != null) {
                    if (labelTokenizersRefs.isEmpty()) {
                        log.info(" ... setting LabelTokenizer of Engine '{}' to null", engine.getName());
                        engine.setLabelTokenizer(null);
                    } else {
                        LabelTokenizer lt = (LabelTokenizer) tracker.getService(labelTokenizersRefs.last());
                        if (!lt.equals(engine.getLabelTokenizer())) {
                            log.info(" ... setting LabelTokenizer of Engine '{}' to {}", engine.getName(), lt);
                            engine.setLabelTokenizer(lt);
                        }
                    }
                }
            // if engine or tracker is null deactivate was already called
            }
            bundleContext.ungetService(reference);
        }

        @Override
        public void modifiedService(ServiceReference reference, Object service) {
            synchronized (labelTokenizersRefs) {
                // override
                labelTokenizersRefs.remove(reference);
                labelTokenizersRefs.add(reference);
                ServiceReference higest;
                try {
                    higest = labelTokenizersRefs.last();
                } catch (NoSuchElementException e) {
                    higest = null;
                }
                EntityLinkingEngine engine = entityLinkingEngine;
                ServiceTracker tracker = labelTokenizerTracker;
                if (engine != null && tracker != null) {
                    LabelTokenizer lt = (LabelTokenizer) (reference.equals(higest) || higest == null ? service : tracker.getService(higest));
                    if (!lt.equals(engine.getLabelTokenizer())) {
                        log.info(" ... setting LabelTokenizer of Engine '{}' to {}", engine.getName(), lt);
                        engine.setLabelTokenizer(lt);
                    }
                }
            // if engine or tracker is null deactivate was already called
            }
        }
    });
    // create the engine
    entityLinkingEngine = new EntityLinkingEngine(engineName, // the searcher might not be available
    entitySearcher, textProcessingConfig, linkerConfig, null);
    // start tracking
    labelTokenizerTracker.open();
    entitySearcher.open();
}
Also used : EntityLinkerConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.EntityLinkerConfig) ServiceTracker(org.osgi.util.tracker.ServiceTracker) ServiceTrackerCustomizer(org.osgi.util.tracker.ServiceTrackerCustomizer) ServiceReference(org.osgi.framework.ServiceReference) EntityLinkingEngine(org.apache.stanbol.enhancer.engines.entitylinking.engine.EntityLinkingEngine) TextProcessingConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig) ConfigurationException(org.osgi.service.cm.ConfigurationException) LabelTokenizer(org.apache.stanbol.enhancer.engines.entitylinking.LabelTokenizer) NoSuchElementException(java.util.NoSuchElementException) Activate(org.apache.felix.scr.annotations.Activate)

Example 4 with TextProcessingConfig

use of org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig in project stanbol by apache.

the class FstLinkingEngineTest method testFstLinkingWithNouns.

@Test
public void testFstLinkingWithNouns() throws Exception {
    Dictionary<String, Object> dict = new Hashtable<String, Object>();
    dict.put(PROCESSED_LANGUAGES, Arrays.asList("en;lmmtip;uc=LINK;prob=0.75;pprob=0.75"));
    dict.put(PROCESS_ONLY_PROPER_NOUNS_STATE, false);
    TextProcessingConfig tpc = TextProcessingConfig.createInstance(dict);
    EntityLinkerConfig elc = new EntityLinkerConfig();
    // this is assumed by this test
    elc.setMinFoundTokens(2);
    elc.setRedirectProcessingMode(RedirectProcessingMode.FOLLOW);
    FstLinkingEngine engine = new FstLinkingEngine("proper-noun-linking", LinkingModeEnum.LINKABLE_TOKEN, fstConfig, tpc, elc, null);
    processConentItem(engine);
    validateEnhancements(Arrays.asList("Chancellor", "Angela Merkel", "Greece", "Greeks", "Germany", "SPD", "change", "election", "party", "policy"), Arrays.asList(DBPEDIA + "Christian_Democratic_Union_(Germany)", DBPEDIA + "Angela_Merkel", DBPEDIA + "Greece", DBPEDIA + "Germany", DBPEDIA + "Social_Democratic_Party_of_Germany", DBPEDIA + "Chancellor", DBPEDIA + "Election", DBPEDIA + "Party", DBPEDIA + "Policy"));
}
Also used : EntityLinkerConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.EntityLinkerConfig) TextProcessingConfig(org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig) Hashtable(java.util.Hashtable) FstLinkingEngine(org.apache.stanbol.enhancer.engines.lucenefstlinking.FstLinkingEngine) Test(org.junit.Test)

Aggregations

EntityLinkerConfig (org.apache.stanbol.enhancer.engines.entitylinking.config.EntityLinkerConfig)4 TextProcessingConfig (org.apache.stanbol.enhancer.engines.entitylinking.config.TextProcessingConfig)4 Test (org.junit.Test)3 Hashtable (java.util.Hashtable)2 FstLinkingEngine (org.apache.stanbol.enhancer.engines.lucenefstlinking.FstLinkingEngine)2 HashMap (java.util.HashMap)1 NoSuchElementException (java.util.NoSuchElementException)1 IRI (org.apache.clerezza.commons.rdf.IRI)1 RDFTerm (org.apache.clerezza.commons.rdf.RDFTerm)1 PlainLiteralImpl (org.apache.clerezza.commons.rdf.impl.utils.PlainLiteralImpl)1 TripleImpl (org.apache.clerezza.commons.rdf.impl.utils.TripleImpl)1 Activate (org.apache.felix.scr.annotations.Activate)1 LabelTokenizer (org.apache.stanbol.enhancer.engines.entitylinking.LabelTokenizer)1 EntityLinkingEngine (org.apache.stanbol.enhancer.engines.entitylinking.engine.EntityLinkingEngine)1 ContentItem (org.apache.stanbol.enhancer.servicesapi.ContentItem)1 StringSource (org.apache.stanbol.enhancer.servicesapi.impl.StringSource)1 ServiceReference (org.osgi.framework.ServiceReference)1 ConfigurationException (org.osgi.service.cm.ConfigurationException)1 ServiceTracker (org.osgi.util.tracker.ServiceTracker)1 ServiceTrackerCustomizer (org.osgi.util.tracker.ServiceTrackerCustomizer)1