private void addCoref(Communication comm) {
AnalyticUUIDGeneratorFactory f = new AnalyticUUIDGeneratorFactory(comm);
AnalyticUUIDGenerator g = f.create();
corefMentions = new EntityMentionSet();
corefMentions.setUuid(g.next());
corefMentions.setMetadata(Conll2011.META_COREF);
corefMentions.setMentionList(new ArrayList<>());
Map<String, List<EntityMention>> clusters = new HashMap<>();
int addedMentions = 0;
for (Conll2011Sentence s : sentences) {
// Get the entity mentions in this sentence
Map<String, List<EntityMention>> c = s.getCoref(g);
// Add these mentions to the EntityMentionSet
for (List<EntityMention> ems : c.values()) {
for (EntityMention em : ems) {
corefMentions.addToMentionList(em);
addedMentions++;
}
}
// Merge into the document-level view of the entities
for (Map.Entry<String, List<EntityMention>> se : c.entrySet()) {
String clustId = se.getKey();
List<EntityMention> existingMentions = clusters.get(clustId);
if (existingMentions == null) {
existingMentions = new ArrayList<>();
clusters.put(clustId, existingMentions);
}
existingMentions.addAll(se.getValue());
}
}
int addedEntities = 0;
corefClusters = new EntitySet();
corefClusters.setUuid(g.next());
corefClusters.setMetadata(Conll2011.META_COREF);
corefClusters.setMentionSetId(corefMentions.getUuid());
corefClusters.setEntityList(new ArrayList<>());
for (Map.Entry<String, List<EntityMention>> cluster : clusters.entrySet()) {
addedEntities++;
Entity ent = new Entity();
ent.setUuid(g.next());
ent.setConfidence(1);
for (EntityMention em : cluster.getValue()) ent.addToMentionIdList(em.getUuid());
corefClusters.addToEntityList(ent);
}
comm.addToEntitySetList(corefClusters);
comm.addToEntityMentionSetList(corefMentions);
if (conll2011.warnOnEmptyCoref && (addedMentions == 0 || addedEntities == 0)) {
LOGGER.warn(
"addedMentions="
+ addedMentions
+ " addedEntities="
+ addedEntities
+ " communication="
+ comm.getId());
}
}
