public Communication convertToConcrete() {
if (comm != null) return comm;
AnalyticUUIDGeneratorFactory f = new AnalyticUUIDGeneratorFactory();
AnalyticUUIDGenerator g = f.create();
comm = new Communication();
comm.setId(id);
comm.setUuid(g.next());
comm.setType(communicationType);
comm.setMetadata(Conll2011.META_GENERAL);
// Tokenization for the words
// TokenTagging for the POS tags
// Parse for the constituency parse
// TokenTagging for NER labels
String sectionNum = null;
Section section = null;
for (Conll2011Sentence sent : sentences) {
if (sectionNum == null || !sent.getPart().equals(sectionNum)) {
if (section != null) comm.addToSectionList(section);
section = new Section();
section.setUuid(g.next());
section.setKind(Conll2011.SECTION_TYPE);
sectionNum = sent.getPart();
}
section.addToSentenceList(sent.convertToConcrete(g));
}
assert section != null;
comm.addToSectionList(section);
// SituationMentionSet for the SRL labels
propBankSrlSituationMentions = new SituationMentionSet();
propBankSrlSituationMentions.setUuid(g.next());
propBankSrlSituationMentions.setMetadata(Conll2011.META_SRL);
propBankSrlSituationMentions.setMentionList(new ArrayList<>());
for (Conll2011Sentence s : sentences) {
for (int pai = 0; pai < s.getNumPredicates(); pai++) {
SituationMention sm = s.getPredArg(pai, g);
assert sm.getTokens() != null || sm.getConstituent() != null;
propBankSrlSituationMentions.addToMentionList(sm);
}
}
comm.addToSituationMentionSetList(propBankSrlSituationMentions);
// EntitySet and EntityMentionSet for the coref labels
addCoref(comm);
// EntityMentionSet for the NER labels
if (this.conll2011.addNerAsEntityMentionSet) {
nerEms = new EntityMentionSet();
nerEms.setUuid(g.next());
nerEms.setMetadata(Conll2011.META_NER);
nerEms.setMentionList(new ArrayList<>());
for (Conll2011Sentence s : sentences)
for (EntityMention em : s.getNerEntityMentions()) nerEms.addToMentionList(em);
comm.addToEntityMentionSetList(nerEms);
}
return comm;
}
