Java Code Examples for gnu.trove.map.hash.TIntObjectHashMap#iterator()

/**
 * Filters the entity candidates against the given list of types
 *
 * @param entities Entities to filter'
 * @param filteringTypes Set of types to filter the entities against
 * @return filtered entities
 */
private Entities filterEntitiesByType(Entities entities, Set<Type> filteringTypes) throws EntityLinkingDataAccessException {
  if (filteringTypes == null) {
    return entities;
  }
  Entities filteredEntities = new Entities();
  TIntObjectHashMap<Set<Type>> entitiesTypes = DataAccess.getTypes(entities);
  for (TIntObjectIterator<Set<Type>> itr = entitiesTypes.iterator(); itr.hasNext(); ) {
    itr.advance();
    int id = itr.key();
    Set<Type> entityTypes = itr.value();
    for (Type t : entityTypes) {
      if (filteringTypes.contains(t)) {
        filteredEntities.add(entities.getEntityById(id));
        break;
      }
    }
  }
  return filteredEntities;
}
 

private void setupEntities(Entities entities) throws Exception {
  if (entities.size() == 0) {
    logger.debug("Skipping initialization of InlinkEntityEntitySimilarity for " + entities.size() + " entities");
    return;
  }

  logger.debug("Initializing InlinkEntityEntitySimilarity for " + entities.size() + " entities");

  entity2vector = new TIntObjectHashMap<>();

  TIntObjectHashMap<int[]> entityInlinks = DataAccess.getInlinkNeighbors(entities);

  for (TIntObjectIterator<int[]> itr = entityInlinks.iterator(); itr.hasNext(); ) {
    itr.advance();
    int entity = itr.key();
    int[] inLinks = itr.value();

    RoaringBitmap bs = new RoaringBitmap();
    for (int l : inLinks) {
      bs.add(l);
    }
    entity2vector.put(entity, bs);
  }

  logger.debug("Done initializing InlinkEntityEntitySimilarity");
}
 

private void setupEntities(Entities entities) throws Exception {
  logger.debug("Initializing MilneWittenEntityEntitySimilarity for " + entities.size() + " entities");

  collectionSize = DataAccess.getCollectionSize();

  TIntObjectHashMap<int[]> entityInlinks = DataAccess.getInlinkNeighbors(entities);

  // inlinks are assumed to be pre-sorted.
  entity2vector = new TIntObjectHashMap<>();

  for (TIntObjectIterator<int[]> itr = entityInlinks.iterator(); itr.hasNext(); ) {
    itr.advance();
    int entity = itr.key();
    int[] inLinks = itr.value();

    RoaringBitmap bs = new RoaringBitmap();
    for (int l : inLinks) {
      bs.add(l);
    }
    entity2vector.put(entity, bs);
  }

  logger.debug("Done initializing MilneWittenEntityEntitySimilarity for " + entities.size() + " entities");
}
 

public YagoEntityKeyphraseCooccurrenceDataProviderIterator(TIntObjectHashMap<TIntIntHashMap> superdocKeyphraseCounts) {
  entitiesIterator = superdocKeyphraseCounts.iterator();
  if (entitiesIterator.hasNext()) {
    entitiesIterator.advance();
    currentEntityKeyphrasesIterator = entitiesIterator.value().iterator();
  } else {
    currentEntityKeyphrasesIterator = null;
  }

}
 

public static Entities getAidaEntitiesForInternalIds(int[] internalIds) throws EntityLinkingDataAccessException {
  TIntObjectHashMap<KBIdentifiedEntity> kbEntities = DataAccess.getKnowlegebaseEntitiesForInternalIds(internalIds);
  Entities entities = new Entities();
  for (TIntObjectIterator<KBIdentifiedEntity> itr = kbEntities.iterator(); itr.hasNext(); ) {
    itr.advance();
    entities.add(new Entity(itr.value(), itr.key()));
  }
  return entities;
}
 

/**
 * Computes all entity occurrence probabilities based on their incoming links.
 *
 * @return Map of Entity->Probability.
 * @throws EntityLinkingDataAccessException
 */
public static TIntDoubleHashMap getAllEntityProbabilities() throws EntityLinkingDataAccessException {
  TIntObjectHashMap<int[]> entityInlinks = getAllInlinks();

  TIntDoubleHashMap entityProbabilities = new TIntDoubleHashMap(entityInlinks.size(), 0.5f);

  // Get the total number of links.
  long totalLinkCount = 0;

  TIntObjectIterator<int[]> itr = entityInlinks.iterator();

  while (itr.hasNext()) {
    itr.advance();
    totalLinkCount += itr.value().length;
  }

  // Derive probabilities from counts.
  itr = entityInlinks.iterator();

  while (itr.hasNext()) {
    itr.advance();
    double probability = (double) itr.value().length / (double) totalLinkCount;
    entityProbabilities.put(itr.key(), probability);
  }

  return entityProbabilities;
}
 

private synchronized void addToEntityKeyphrasesCache(String querySignature, TIntObjectHashMap<List<EntityKeyphraseData>> entityKeyphrases) {
  CachingHashMap<Integer, List<EntityKeyphraseData>> queryCache = entityKeyphrasesCaches.get(querySignature);
  if (queryCache == null) {
    int maxEntities = EntityLinkingConfig.getAsInt(EntityLinkingConfig.ENTITIES_CACHE_SIZE);
    queryCache = new CachingHashMap<Integer, List<EntityKeyphraseData>>(maxEntities);
    entityKeyphrasesCaches.put(querySignature, queryCache);
  }

  for (TIntObjectIterator<List<EntityKeyphraseData>> itr = entityKeyphrases.iterator(); itr.hasNext(); ) {
    itr.advance();
    int entityId = itr.key();
    List<EntityKeyphraseData> keyphrases = itr.value();
    queryCache.put(entityId, keyphrases);
  }
}
 

private synchronized void addToEntityKeywordsCache(String querySignature, TIntObjectHashMap<List<EntityKeywordsData>> entityKeywords) {
  CachingHashMap<Integer, List<EntityKeywordsData>> queryCache = entityKeywordsCaches.get(querySignature);
  if (queryCache == null) {
    int maxEntities = EntityLinkingConfig.getAsInt(EntityLinkingConfig.ENTITIES_CACHE_SIZE);
    queryCache = new CachingHashMap<Integer, List<EntityKeywordsData>>(maxEntities);
    entityKeywordsCaches.put(querySignature, queryCache);
  }

  for (TIntObjectIterator<List<EntityKeywordsData>> itr = entityKeywords.iterator(); itr.hasNext(); ) {
    itr.advance();
    int entityId = itr.key();
    List<EntityKeywordsData> keyphrases = itr.value();
    queryCache.put(entityId, keyphrases);
  }
}
 

@Override //TODO: language not used
public Map<String, int[]> getCategoryIdsForMentions(Set<String> mentions, Language language, boolean isNamedEntity) throws EntityLinkingDataAccessException {
  Map<String, Entities> entities = getEntitiesForMentions(mentions, 1.0, 0, isNamedEntity);

  Map<String, Set<Integer>> mentionCategories = new HashMap<>(entities.size());

  for (Entry<String, Entities> entry:entities.entrySet()) {
    Set<Integer> types = mentionCategories.get(entry.getKey());
    if (types == null) {
      types = new HashSet<>();
    }
    List<Integer> entityIds = new ArrayList<>();
    for (Entity entity:entry.getValue()) {
      entityIds.add(entity.getId());
    }
    TIntObjectHashMap<int[]> categories = getCategoryIdsForEntitiesIds(ArrayUtils.toPrimitive(entityIds.toArray(new Integer[0])));
    TIntObjectIterator<int[]> it = categories.iterator();
    while (it.hasNext()) {
      types.addAll(Arrays.asList(ArrayUtils.toObject(it.value())));
    }
  }

  Map<String, int[]> ret = new HashMap<>();
  for (String key:mentionCategories.keySet()) {
    Integer[] temp = mentionCategories.get(key).toArray(new Integer[0]);
    ret.put(key, ArrayUtils.toPrimitive(temp));
  }

  return ret;
}
 

private static Map<Integer, Set<Integer>> computeEntityOutlinks() throws EntityLinkingDataAccessException, IOException, SQLException {
	Map<Integer, Set<Integer>> inLinkOutLinkMap = new HashMap<>();
	String file_name = EntityLinkingManager.getAidaDbIdentifierLight() + "_" + ENTITY_OUTLINK_CACHE_JSON;
	if (Files.exists(Paths.get(file_name))) {
		logger.info("Loading " + file_name + " from cache");
		Map<Integer, Set<Integer>> cache = new GsonBuilder().enableComplexMapKeySerialization().create()
				.fromJson(new JsonReader(new FileReader(file_name)),
						new TypeToken<Map<Integer, Set<Integer>>>() {
						}.getType());

		if (cache != null) {
			inLinkOutLinkMap.putAll(cache);
			return inLinkOutLinkMap;
		}
	}
	logger.info("Computing " + ENTITY_OUTLINK_CACHE_JSON);
	TIntObjectHashMap<int[]> inlinkNeighbors = DataAccess.getInlinkNeighbors(DataAccess.getAllEntities());
	TIntObjectIterator<int[]> iterator = inlinkNeighbors.iterator();

	while (iterator.hasNext()) {
		iterator.advance();
		int outEntity = iterator.key();
		int[] inEntities = iterator.value();
		for (int inEntity : inEntities) {
			if (!inLinkOutLinkMap.containsKey(inEntity)) {
				inLinkOutLinkMap.put(inEntity, new HashSet<>());
			}
			Set<Integer> outLinks = inLinkOutLinkMap.get(inEntity);
			outLinks.add(outEntity);

		}
	}

	if (!inLinkOutLinkMap.isEmpty()) {
		Gson gson = new Gson();
		Files.write(Paths.get(file_name), gson.toJson(inLinkOutLinkMap).getBytes());
	}

	return inLinkOutLinkMap;
}
 

@Test public void testGetTypes() throws EntityLinkingDataAccessException {
Entities entities = DataAccess.getEntitiesForMention("Merkel", 1.0, 0, true);
  TIntObjectHashMap<Set<Type>> types = DataAccess.getTypes(entities);
  Type politician = new Type("YAGO3", "<wordnet_politician_110450303>");

  Set<Type> allTypes = new HashSet<>();
  for (TIntObjectIterator<Set<Type>> itr = types.iterator(); itr.hasNext(); ) {
    itr.advance();
    allTypes.addAll(itr.value());
  }

  assertTrue(allTypes.contains(politician));
}
 

public String toString(){
	StringBuilder sb = new StringBuilder();

	TIntObjectHashMap<TLinkedHashSet<String>> sets = new TIntObjectHashMap<TLinkedHashSet<String>>();

	for(int i=0;i<alpahbet.size();i++){
		int head = getHead(i);
		TLinkedHashSet<String> s = sets.get(head);
		if(s==null){
			s = new TLinkedHashSet();
			sets.put(head, s);
		}
		s.add(alpahbet.lookupString(i));
	}

	TIntObjectIterator<TLinkedHashSet<String>> it = sets.iterator();
	while(it.hasNext()){
		it.advance();
		if(it.value().size()<2)
			continue;
		sb.append(wordProb.get(it.key()));
		sb.append(" ");
		TObjectHashIterator<String> itt = it.value().iterator();
		while(itt.hasNext()){
			String ss = itt.next();
			sb.append(ss);
			sb.append(" ");
		}
		sb.append("\n");
	}

	return sb.toString();

}
 

private void computeNonprocessedEntityMentionLabels(Set<Entity> processingEntities,
														HashMap<Integer, Map<String, MentionObject>> entityMentionLabelsMap)
			throws EntityLinkingDataAccessException, InterruptedException {
		TIntObjectHashMap<int[]> typesIdsForEntitiesIds =
				DataAccess.getTypesIdsForEntitiesIds(processingEntities
						.stream()
						.mapToInt(Entity::getId)
						.toArray());

		TIntObjectHashMap<List<MentionObject>> mentionsForEntities = DataAccess.getMentionsForEntities(new Entities(processingEntities));

		TIntObjectIterator<List<MentionObject>> entityMentionsIterator = mentionsForEntities.iterator();
		while (entityMentionsIterator.hasNext()) {
			Map<String, MentionObject> entityResult = new HashMap<>();
			entityMentionsIterator.advance();

			int eid = entityMentionsIterator.key();

			int[] typeIDs = typesIdsForEntitiesIds.get(eid);
			NerType.Label nerTypeForTypeIds = NerType.getNerTypeForTypeIds(typeIDs);

			for (MentionObject mentionObject : entityMentionsIterator.value()) {
				if (Thread.interrupted()) {
					throw new InterruptedException();
				}
				String entityMention = mentionObject.getMention();
				NerType.Label nerTypeForTypeIds_ = nerTypeForTypeIds;

//              getting rid of the junk from the aida database
				if (stopwords.contains(entityMention.toLowerCase()) ||
						entityMention.contains("<SPAN") ||
						entityMention.contains("=") ||
						entityMention.contains("<!--") ||
						entityMention.contains("(") && entityMention.contains(")") ||
						isDate(entityMention.trim()) ||
						entityMention.matches("[.,\\/#!$%\\^&\\*;:{}=\\-_`~()]") ||
						entityMention.endsWith("'S")
						) {
					continue;
				}

//                i.e. United States is a location, not an organization
				if (knownCountries.contains(entityMention.toLowerCase())) {
					nerTypeForTypeIds_ = NerType.Label.LOCATION;
				}
				if (languagesList.contains(entityMention.toLowerCase())) {
					nerTypeForTypeIds_ = NerType.Label.MISC;
				}
				MentionObject copy = mentionObject.copy();
				copy.setLabel(nerTypeForTypeIds_);
				entityResult.put(entityMention, copy);
			}
			entityMentionLabelsMapCache.put(eid, entityResult);
			entityMentionLabelsMap.put(eid, entityResult);
		}
	}