private EnsembleMentionEntitySimilarity prepapreMES() {
Entities entities = new Entities();
for (Mention mention : input.getMentions().getMentions()) {
MentionTracer mt = new MentionTracer(mention);
tracer.addMentionForDocId(docId, mention, mt);
for (Entity entity : mention.getCandidateEntities()) {
EntityTracer et = new EntityTracer(entity.getName());
tracer.addEntityForMention(mention, entity.getName(), et);
}
entities.addAll(mention.getCandidateEntities());
}
logger.info(
"Disambiguating '"
+ docId
+ "' ("
+ input.getMentions().getMentions().size()
+ " mentions, "
+ entities.size()
+ " entities)");
if (includeNullAsEntityCandidate) {
entities.setIncludesNmeEntities(true);
}
EnsembleMentionEntitySimilarity mes = null;
try {
mes = new EnsembleMentionEntitySimilarity(input.getMentions(), entities, ss, docId, tracer);
return mes;
} catch (Exception e) {
e.printStackTrace();
return null;
}
}
protected void disambiguate(EnsembleMentionEntitySimilarity mes) throws Exception {
for (Mention mention : input.getMentions().getMentions()) {
List<ResultEntity> entities = new LinkedList<ResultEntity>();
for (Entity entity : mention.getCandidateEntities()) {
double sim = mes.calcSimilarity(mention, input.getContext(), entity);
entities.add(new ResultEntity(entity.getName(), sim));
}
if (!entities.isEmpty()) {
Collections.sort(entities);
} else {
entities.add(ResultEntity.getNoMatchingEntity());
}
Map<ResultMention, List<ResultEntity>> docSolutions = solutions.get(docId);
if (docSolutions == null) {
docSolutions = new HashMap<ResultMention, List<ResultEntity>>();
solutions.put(docId, docSolutions);
}
ResultMention rm =
new ResultMention(
docId, mention.getMention(), mention.getCharOffset(), mention.getCharLength());
docSolutions.put(rm, entities);
}
}
@Test
public void sentencesTokensTest() {
Preparator p = new Preparator();
PreparedInput input =
p.prepare(
"temp",
"This is a sentence. And another one. Just for the kicks.",
new StanfordManualPreparationSettings());
Tokens ts = input.getTokens();
List<List<Token>> sentences = ts.getSentenceTokens();
assertEquals(3, sentences.size());
assertEquals("This", sentences.get(0).get(0).getOriginal());
assertEquals("sentence", sentences.get(0).get(3).getOriginal());
assertEquals(".", sentences.get(0).get(4).getOriginal());
assertEquals("another", sentences.get(1).get(1).getOriginal());
assertEquals("kicks", sentences.get(2).get(3).getOriginal());
}
@Override
public void run() {
long beginTime = System.currentTimeMillis();
try {
AidaManager.fillInCandidateEntities(
docId, input.getMentions(), includeNullAsEntityCandidate, includeContextMentions);
} catch (SQLException e) {
logger.error("SQLException when getting candidates: " + e.getLocalizedMessage());
}
EnsembleMentionEntitySimilarity mes = prepapreMES();
try {
disambiguate(mes);
} catch (Exception e) {
logger.error("Error: " + e.getLocalizedMessage());
}
double runTime = (System.currentTimeMillis() - beginTime) / (double) 1000;
logger.info("Document '" + docId + "' done in " + nf.format(runTime) + "s");
}
public LocalDisambiguation(
PreparedInput input,
SimilaritySettings settings,
boolean includeNullAsEntityCandidate,
boolean includeContextMentions,
String docId,
Map<String, Map<ResultMention, List<ResultEntity>>> solutions,
Tracer tracer)
throws SQLException {
nf = NumberFormat.getNumberInstance(Locale.ENGLISH);
nf.setMaximumFractionDigits(2);
logger.debug(
"Preparing '" + docId + "' (" + input.getMentions().getMentions().size() + " mentions)");
this.ss = settings;
this.docId = docId;
this.solutions = solutions;
this.input = input;
this.includeNullAsEntityCandidate = includeNullAsEntityCandidate;
this.includeContextMentions = includeContextMentions;
this.tracer = tracer;
logger.debug("Finished preparing '" + docId + "'");
}