Examples with NounPhraseFilterer org.apache.stanbol.enhancer.engines.entitycoreference.impl.NounPhraseFilterer used on opensource projects

Example 1 with NounPhraseFilterer

use of org.apache.stanbol.enhancer.engines.entitycoreference.impl.NounPhraseFilterer in project stanbol by apache.

the class EntityCoReferenceEngine method activate.

@SuppressWarnings("unchecked")
@Activate
protected void activate(ComponentContext ctx) throws ConfigurationException {
    super.activate(ctx);
    Dictionary<String, Object> config = ctx.getProperties();
    /* Step 1 - initialize the {@link NounPhraseFilterer} with the language config */
    String languages = (String) config.get(CONFIG_LANGUAGES);
    if (languages == null || languages.isEmpty()) {
        throw new ConfigurationException(CONFIG_LANGUAGES, "The Languages Config is a required Parameter and MUST NOT be NULL or an empty String!");
    }
    nounPhraseFilterer = new NounPhraseFilterer(languages.split(","));
    /* Step 2 - initialize the {@link CoreferenceFinder} */
    String referencedSiteID = null;
    Object referencedSiteIDfromConfig = config.get(REFERENCED_SITE_ID);
    if (referencedSiteIDfromConfig == null) {
        throw new ConfigurationException(REFERENCED_SITE_ID, "The ID of the Referenced Site is a required Parameter and MUST NOT be NULL!");
    }
    referencedSiteID = referencedSiteIDfromConfig.toString();
    if (referencedSiteID.isEmpty()) {
        throw new ConfigurationException(REFERENCED_SITE_ID, "The ID of the Referenced Site is a required Parameter and MUST NOT be an empty String!");
    }
    if (Entityhub.ENTITYHUB_IDS.contains(referencedSiteID.toLowerCase())) {
        log.debug("Init NamedEntityTaggingEngine instance for the Entityhub");
        referencedSiteID = null;
    }
    int maxDistance;
    Object maxDistanceFromConfig = config.get(MAX_DISTANCE);
    if (maxDistanceFromConfig == null) {
        maxDistance = Constants.MAX_DISTANCE_DEFAULT_VALUE;
    } else if (maxDistanceFromConfig instanceof Number) {
        maxDistance = ((Number) maxDistanceFromConfig).intValue();
    } else {
        try {
            maxDistance = Integer.parseInt(maxDistanceFromConfig.toString());
        } catch (NumberFormatException nfe) {
            throw new ConfigurationException(MAX_DISTANCE, "The Max Distance parameter must be a number");
        }
    }
    if (maxDistance < -1) {
        throw new ConfigurationException(MAX_DISTANCE, "The Max Distance parameter must not be smaller than -1");
    }
    String entityUriBase = (String) config.get(ENTITY_URI_BASE);
    if (entityUriBase == null || entityUriBase.isEmpty()) {
        throw new ConfigurationException(ENTITY_URI_BASE, "The Entity Uri Base parameter cannot be empty");
    }
    String spatialAttrForPerson = (String) config.get(SPATIAL_ATTR_FOR_PERSON);
    String spatialAttrForOrg = (String) config.get(SPATIAL_ATTR_FOR_ORGANIZATION);
    String spatialAttrForPlace = (String) config.get(SPATIAL_ATTR_FOR_PLACE);
    String orgAttrForPerson = (String) config.get(ORG_ATTR_FOR_PERSON);
    String entityClassesToExclude = (String) config.get(ENTITY_CLASSES_TO_EXCLUDE);
    corefFinder = new CoreferenceFinder(languages.split(","), siteManager, entityhub, referencedSiteID, maxDistance, entityUriBase, spatialAttrForPerson, spatialAttrForOrg, spatialAttrForPlace, orgAttrForPerson, entityClassesToExclude);
    log.info("activate {}[name:{}]", getClass().getSimpleName(), getName());
}