/** generate mention annotations (with entity numbers) based on the ACE entities and mentions. */
static void addMentionTags(Document doc, AceDocument aceDoc) {
ArrayList<AceEntity> entities = aceDoc.entities;
for (int i = 0; i < entities.size(); i++) {
AceEntity entity = entities.get(i);
ArrayList<AceEntityMention> mentions = entity.mentions;
for (int j = 0; j < mentions.size(); j++) {
AceEntityMention mention = (AceEntityMention) mentions.get(j);
// we compute a jetSpan not including trailing whitespace
Span aceSpan = mention.head;
// skip mentions in ChEnglish APF not aligned to any English text
if (aceSpan.start() < 0) continue;
Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
FeatureSet features = new FeatureSet("entity", new Integer(i));
if (flags.contains("types")) {
features.put("type", entity.type.substring(0, 3));
if (entity.subtype != null) features.put("subtype", entity.subtype);
}
if (flags.contains("extents")) {
String cleanExtent = mention.text.replaceAll("\n", " ");
features.put("extent", AceEntityMention.addXmlEscapes(cleanExtent));
}
doc.annotate("mention", jetSpan, features);
}
}
}
public static void addAnnotations(Document doc, AceDocument aceDoc) {
boolean monocase = Ace.allLowerCase(doc);
if (year.equals("2004")) gazetteer.setMonocase(monocase);
if (flags.contains("sentences")) addSentences(doc);
if (flags.contains("timex")) addTimexTags(doc, aceDoc);
if (flags.contains("mentions")) addMentionTags(doc, aceDoc);
if (flags.contains("names")) addENAMEXtags(doc, aceDoc);
}
public static void setFlag(String flag) {
if (flag.equals("sentences")
|| flag.equals("timex")
|| flag.equals("mentions")
|| flag.equals("extents")
|| flag.equals("types")
|| flag.equals("names")) {
flags.add(flag);
} else {
System.err.println("APFtoXML: invalid flag");
System.err.println("possible flags: sentences timex mentions extents types names");
System.exit(1);
}
}
public static void clearFlags() {
flags.clear();
}